Using clinical practice variations as a method for commissioners and clinicians to identify and prioritise opportunities for disinvestment in health care: a cross-sectional study, systematic reviews and qualitative study

Identifying and categorising procedures

We used the HES26 admitted patient care data set to identify inpatient procedures. HES is a routinely collected data set that records all episodes of care provided to all patients (NHS funded and privately insured) admitted, as a day case or inpatient, to NHS hospitals and NHS-funded patients treated in independent sector hospitals. We extracted pseudoanonymised individual episode records on all admissions contained in the HES data set between the financial years 2007/8 and 2011/12. Up to 24 clinical procedures per episode may be recorded using Office of Population, Censuses and Surveys (OPCS) (fourth revision) codes. OPCS codes are hierarchical and include more than 9000 four-character codes defining procedures at the finest level of detail. However, most of these codes are used infrequently, so we elected to define procedures using the three-character OPCS codes (n ≈ 1500).

We focused on the most clinically and economically consequential procedures by including only the 264 most widely used procedure codes, which accounted for over 90% of all inpatient procedures, in our analysis. OPCS codes include some very minor procedures (e.g. blood withdrawal) and diagnostic procedures (e.g. diagnostic echocardiography), which were predominantly not the primary reason for hospital admission. We decided to focus on more major therapeutic procedures, which were thought more likely to be recorded accurately and consistently between hospitals. We excluded diagnostic (n = 53), minor (n =38) and obstetric (n = 5) procedures from our analysis to focus on major therapeutic procedures. We removed a further three procedures because postcode of residence was missing in more than 10% of episodes. We excluded 11 procedures from all analyses because of changes in OPCS procedure codes between years (version 4.4 used in 2007/8 to version 4.6 introduced in 2011/12) which may have led to inconsistent clinical coding. This left 154 therapeutic procedures for analysis. These include procedures that are predominantly elective and those that are more frequently performed as an emergency procedure. A single stay in hospital may comprise more than one episode as patient care is transferred from one consultant to another. Therefore, a hospital stay may contain many procedures, all of which were eligible for inclusion in our analysis. When multiple records of the same procedure were recorded on the same patient with the same admission date, we included only the first record in our analysis in order to avoid the risk of double-counting procedures which were recorded more than once as a result of coding errors. 47

Estimating variation in procedure rates

Geographical variation was measured using PCT boundaries. Until April 2013, PCTs were responsible for commissioning most NHS services for their resident population; they represent geographically contiguous areas of England. In 2007/8 there were 152 (which decreased to 151 in April 2010) PCTs in England; on average each PCT was responsible for approximately 340,000 residents. Since April 2013, PCTs have been replaced by 212 CCGs.

We denote the observed number of utilisations of procedure j on residents of PCT i by ‘Observed_ij’ (i = 1, . . . , 152 PCTs; j = 1, . . . , 154 procedures). We used a two-stage approach to quantifying variation. In stage 1 we calculated expected numbers of procedure utilisations for procedure j on residents of PCT i based on demographic factors and factors that might affect clinical need for NHS care, denoted by ‘Expected_ij’. These expected numbers were calculated using indirect standardisation48 for age and sex (using quinary age groups and gender for England as the standard population49), to account for differences in the size and the age/sex composition of PCT populations. Then Poisson regression was used to further adjust rates for the ethnic50 and socioeconomic composition51 of PCTs, the prevalence of chronic diseases (asthma, atrial fibrillation, coronary heart disease, chronic kidney disease, dementia, diabetes, hypertension, stroke, all-cause cancer),52 and markers of unhealthy lifestyle (binge drinking, smoking and obesity). 52 We also adjusted for the prevalence of private medical care, which might substitute for NHS treatment, based on the number of private hospital beds within a 30-mile radius of the PCT headquarters,53 using mean imputation for the 29% of private hospitals where bed numbers were not recorded.

In stage 2 of our statistical analysis we quantified the residual inter-PCT variability in utilisation rates by using the expected counts as a covariate in random effects Poisson regression models:

The crucial parameter is σ²_j, which quantifies the remaining variability in utilisation of procedure j across PCTs, having adjusted for all factors reflected in the expected counts. Importantly, focusing attention on the inter-PCT standard deviation (SD) from a random effects model (or equivalently on a function of it, as we describe in the next paragraph) also appropriately adjusts for chance variability. In what follows, we refer to this parameter as the inter-PCT SD.

Models were fitted within a Bayesian framework using the WinBUGS software (version 1.4.3, MRC Biostatistics Unit, Cambridge, UK),54 which allowed us to estimate the probability of this unexplained variation in utilisation of each procedure exceeding a given threshold. We estimated the probability that a procedure was ‘very highly’ or ‘highly’ variable, which we arbitrarily defined as an inter-PCT SD greater than three or two times the median variability across all therapeutic procedures respectively. To improve interpretability, we transformed each model-based inter-PCT SD estimate into a ‘utilisation ratio’ (UR), which we defined as the rate in a high-utilisation PCT (at the 90th centile of the random effects distribution) divided by the rate in a low-utilisation PCT (at the 10th centile).

Estimating temporal changes in geographical variation and exploring factors associated with high variation

If variation is a proxy for clinical uncertainty about the value of a procedure, then temporal growth in variation could be an indicator of procedures where evidence is evolving, creating either local enthusiasm or disillusionment about the value of a procedure. In order to explore this possibility, for each procedure we performed a linear regression of the log of the estimated inter-PCT SDs on year, to quantify the change in inter-PCT variation between 2007/8 and 2011/12. The log transformation was applied to the SDs because of the large positive skew in the distribution and unequal variation in the error terms. We also conducted exploratory analyses to examine five factors that might be associated with high geographical variation in hospital procedure rates: (1) coding uncertainty; (2) variation in the quality of community care; (3) uncertainty about the appropriate setting for the procedure; (4) urgency and invasiveness of the procedure; and (5) evolving or uncertain evidence. The eight variables selected as potential proxies for these five factors, their definition and our rationale for their potential association with high inter-PCT variation are documented in Tables 1 and 2. We conducted a multivariable linear regression of the log of the inter-PCT SD for each of the 154 procedures on these eight variables.

Main results

There is a high degree of geographical variation in procedure rates for many commonly performed procedures that cannot be explained by proxies of clinical need. For six procedures, rates in PCTs at the upper and lower deciles still differed by more than 10-fold, after adjustments for chance variation and proxies of clinical need. Variation was most pronounced in procedures where utilisation was increasing or decreasing most rapidly and those where a substitute procedure was available. Policy-makers could use geographical variation as a starting point to identify procedures where HTR or new RCTs might be needed to inform investment and disinvestment decisions.

Strength and weaknesses

The main strengths of this study lie in the large sample, novelty of the research question and breadth of procedures considered. Our model-based approach accounts appropriately for statistical chance, and the UR provides a simple summary measure of disparity between PCTs. In order to make valid comparisons of procedure rates between PCTs we have adjusted for a number of indicators of clinical need. However, bias may still occur because of unmeasured aspects of clinical need or if the accuracy of demographic and morbidity measures varies by PCT. In particular, local variations in the quality of community and primary health services could plausibly cause variation in procedure rates for ambulatory care sensitive conditions. 28 Future work could use proxies for ambulatory care quality such as community health service expenditure or the primary care Quality and Outcomes Framework to assess their association with procedure rates. We used the number of private hospital beds as a proxy measure for the number of private hospital procedures which may be inappropriate, particularly for some conditions (e.g. excision of appendix), where private treatment is less likely. It is also possible that some of the residual variation in procedure rates observed in our study actually reflects differences in local PCT criteria for allowing access to procedures paid for by the NHS. We do not have the data needed to test this hypothesis, but, as PCTs typically did not have access criteria for many of the procedures included in our study, we do not think it would be a predominant factor. Arguably, such PCT criteria ‘lotteries’ still reflect uncertainty about the value of care that might be reduced by better evidence informing more uniform criteria. PCTs have now been superseded by CCGs. We cannot use the CCG as the unit of analysis because demographic and morbidity statistics are not currently available for CCGs. Although it is unlikely that boundary changes will have any immediate impact on geographical variation of procedure rates, new CCG approaches to commissioning will have an impact in the medium term. 9

Our analyses are reliant on the OPCS coding framework, which was not developed for this purpose. The OPCS codes are reviewed annually, but contain anomalies whereby some substitute procedures can be distinguished by the primary OPCS three-character code, for example cemented versus uncemented joint replacement, whereas others cannot, for example open versus endoscopic carpal tunnel decompression. Furthermore, some OPCS procedure codes span more than one diverse clinical group, for example adenoidectomy in children with chronic tonsillitis or sleep apnoea. Methodological research is needed to create an OPCS/International Classification of Diseases (ICD) diagnosis code algorithm that would define a clinically nuanced matrix of distinct procedure/diagnosis group pairings. As both coding frameworks contain thousands of codes, such research would be time-consuming and was beyond the scope of this study. It is important to recognise that different levels of procedure code aggregation will be useful in different circumstances. By using three-character OPCS codes we were able to identify high inter-PCT variation in how varicose vein procedures are conducted (e.g. transluminal or combined), but could not explore variation between PCTs in whether or not to do perform varicose vein procedures.

We have restricted our analysis to major therapeutic procedures to reduce the likelihood of inconsistent coding of minor medical and diagnostic procedures affecting our results. However, inaccurate coding of procedures is still a concern which may contribute to high variation. As hospitals have moved to ‘payment by results’ there is a stronger incentive for complete coding of procedures, but also a possibility of up-coding to increase revenues. 56 We have focused on inpatient and day case procedures because outpatient procedures and those performed in other settings tend to be poorly recorded in routine data. An ideal analysis would include procedures across all settings to provide a more comprehensive picture.

Comparison with other studies

Geographical variation in hospital procedure rates has long been documented in the UK21,22,57 and internationally. 58,59 Indeed, its persistence between countries and through time may give the impression that it is an intractable issue. However, practice variation can be reduced;60 in Scotland, a 50% decline in appendectomy rates between 1973 and 1993 was accompanied by increasing uniformity of procedure rates between health boards. 24 Our findings demonstrate that there is now relative uniformity in tonsillectomy rates between PCTs.

There has been relatively little work exploring the potential causes of geographical variation in health care. Westert and Groenewegen23 and Weinstein et al. 61 describe how an individual clinician’s conservative or liberal use of a procedure might be transferred to colleagues, creating local signatures of procedure use that persist over time. They also stress that the development of these practice patterns may be intensified by fee for service clinician and hospital payments. Birkmeyer et al. 58 and others have demonstrated that geographical variation was higher in discretionary procedures, such as back surgery, where indications are fuzzy and the appropriateness is debated than for procedures, such as hip fracture repair, where the diagnosis and the need for surgery is clear cut. Our analysis also found that procedures that were predominantly performed electively had higher variation.

Unlike most other studies on variation, we did not begin with a list of procedures where we suspected clinical uncertainty existed. Instead, we selected all commonly performed NHS procedures and allowed the data to identify extreme variation. Many of the usual suspects for clinical uncertainty did not appear at the top of our list (e.g. excision of tonsil had a low estimated UR of 1.5; excision of lumbar intervertebral disc had a modest estimated UR of 2.5), underlining the potential of this approach to identify emerging areas of clinical uncertainty.

Implications

Our finding that variation is high in rapidly diffusing or declining procedures, and procedures where a substitute is available, supports the theory that variation is a marker for clinical uncertainty. This suggests that regular monitoring of geographical variation might help NHS regulators and commissioners identify procedures ripe for HTR. However, the appropriate HTR question may take many forms. For some of the ‘high variance’ procedures identified, the question is likely to be ‘which of two procedures (e.g. traditional versus various endovenous varicose vein procedures)62 is more cost-effective and for which patient subgroups)?’63 In other cases, the question may be ‘is this procedure (e.g. radiofrequency denervation of spinal facet joints) more cost-effective than conservative care and if so in which patients?’64 Answering these questions and disseminating the findings is vital if patients across the NHS are to receive best care. However, it is important to realise that, while this should lead to disinvestment from procedures that are not being used cost-effectively, it will not necessarily save the NHS money if they are replaced by more effective but more expensive procedures.

We can only speculate on whether or not our data contain a modern-day equivalent of tonsillectomy, where the clinical indications will narrow drastically in the future. The rise of evidence-based medicine should reduce this risk, but large high-quality RCTs of surgical procedures are still not commonplace and many longstanding procedures have never been fully evaluated in rigorous RCTs. 65 Lack of evidence about which clinical subgroups benefit from a procedure and which do not is the major challenge for commissioners and clinicians. 15 A natural response from commissioners with high local procedure rates is to try to enforce convention through criteria-based access (CBA) and referral management, and many new CCGs have adopted this approach. 9 However, the convention might be wrong66,67 and a more far-sighted response would be for research funders to use geographical variation to prioritise procedures where primary research assessing value in specific clinical subgroups is most needed.

It is also likely that the approach outlined in this chapter, being reliant on coding accuracy, will identify some red herrings. In early iterations of our analysis, in vitro fertilisation (IVF) procedures had extremely high apparent inter-PCT variation with a substantial number of PCTs documenting zero procedures. Media reports have highlighted variations in access to IVF across England;68 however, on closer inspection of the data, we felt that the zero rates in some PCTs were in fact more probably due to variable practice in recording the postcode of patient residence for potentially sensitive fertility treatments. 69

Unanswered questions and future research

Similar studies of unplanned admissions,70 diagnostic tests,71 referral, prescription rates72 and other process measures45 will be useful in identifying other aspects of health care where uncertainty manifests through variation. We chose hospital procedures because there are a nationwide routine data set and a financial incentive for hospitals to document procedures. The integration of routine primary and secondary care data sets has the potential to provide much richer analyses of the intersectorial causes of variation. However, such analyses are dependent on complete and accurate coding and may be jeopardised by any future fragmentation of data sets between NHS and independent sector providers. Additional qualitative research in areas with extreme procedure rates would help us gain a better understanding of the causes of observed variations.

Conclusions

The widespread geographical variations in hospital procedure rates in England are not solely due to variance in clinical need and are likely to reflect clinical discretion regarding appropriate procedure use and setting (e.g. day case or outpatient). NICE and NHS England might, with appropriate caution, use geographical variation to identify candidates for HTR and potential disinvestment. These variations should also be used to set a research agenda for investment in RCTs in surgery.

Benchmarking inter-PCT variation in procedure rates

The methods for identifying inter-PCT variation in procedure rates and adjusting for proxies of clinical need are explained in detail in the previous chapter. However, as the number of procedures, study dates and method for adjusting for private health-care provision are different for the benchmarking work, we briefly recap those methods in this section.

For the benchmarking work, we extracted anonymised individual episode records from HES between 1 April 2007 and 31 March 2010. All interventions and procedures performed during each episode are recorded using OPCS fourth revision codes. OPCS codes include more than 9000 four-character codes defining procedures at the finest level of detail. However, most of these codes are used infrequently, so we categorised procedures using the three-character OPCS codes (n ≈ 1500). We focused on the most clinically and economically consequential procedures by including only the 269 most widely used procedures, which accounted for over 90% of all inpatient procedures. We excluded minor (n = 45) and diagnostic (n = 34) procedures and procedures where codes changed between OPCS version 4.4 (2007/8) and version 4.5 (used since 2009/10), (n = 9). This left 181 major therapeutic procedures for analysis. The number of included procedures differs from Chapter 3, as prior to the analysis in Chapter 3 we tightened our procedure exclusion criteria to focus on major therapeutic procedures and because of coding changes in 2011/12. Up to 24 procedures can be recorded during each episode and a single stay in hospital may comprise more than one episode as patient care is transferred from one consultant to another. All procedures from both admission and subsequent episodes were eligible for inclusion in our analysis.

Crude annual procedure rates (per 100,000) were calculated by dividing the number of procedures undertaken on PCT residents by the total PCT population. 49 We counted the observed number of each procedure within every PCT, ‘Observed_ij’, and estimated the expected number, ‘Expected_ij’, indirectly standardising48 for age and sex and accounting for differences in PCT ethnic and socioeconomic composition and the prevalence of chronic disease and unhealthy lifestyle as described in the previous chapter. However, in the local benchmarking work we used data on the regional prevalence of private medical insurance, which might substitute for NHS treatment, from the British Household Panel Survey to adjust for private health-care provision. 78 The observed number of procedures within each PCT, ‘Observed_ij’, was divided by the expected, ‘Expected_ij’, and multiplied by the national rate of procedure use (i.e. rate in the standard population) to yield adjusted PCT rates.

Benchmarking

For the work in this chapter, we studied procedure utilisation rates in two PCT commissioning groups in more detail. The first PCT commissioning group (henceforth PCT1) served a region that was relatively rural, with a predominantly white and comparatively wealthy population. The second PCT commissioning group (henceforth PCT2) served a predominantly urban region, consisting of areas with high proportions of ethnic minority populations and pockets of high deprivation. Although we label them as a PCT throughout, one of the commissioning groups supported more than one neighbouring PCT. For each procedure, we quantified the difference between the individual PCT’s rate and the national rate using both the absolute and the relative difference. For each of the two PCTs we presented the 20 procedures with the largest estimated absolute difference as candidates for further study. We calculated a summary measure of the national geographical variability of each of these 20 procedures (described in more detail in Chapter 3). We defined procedures as ‘very high’ variance if there was a greater than 0.95 probability that inter-PCT variation in procedure utilisation was more than three times as high as the median inter-PCT variation for all 181 procedures. Similarly we identified ‘high’ (> 2 times as high as the median), ‘average’ (> 0.5 times as high as the median) and ‘low’ (> 0.33 times as high as the median) variance procedures. The remaining procedures were classified as ‘very low’ variance procedures. In addition, for each procedure, we present the estimated national rank of the specific PCT, where a value of 1 would indicate that the PCT had the largest (of all 152 PCTs) estimated absolute difference in rate from the national rate.

We further explored intra-PCT variation by tabulating procedure numbers by hospital, geographical variation at the sub-PCT (i.e. local borough) level and temporal variation in local procedure rates since 2001/2. A list of 20 procedures with the highest estimated absolute differences between the local PCT rate and national rate was discussed at a project team meeting in October 2011 with public health and commissioning representatives from the two PCT commissioning groups. As a result of the project team meeting we selected one procedure for rapid technology assessment at each PCT. These were ‘release of entrapment of peripheral nerve at wrist’ (OPCS code A65) at PCT1 and ‘incision of capsule of lens’ (OPCS code C73) at PCT2. The choice of procedures was a joint one between the research team and representatives of the two PCTs. These procedures were selected because procedure rates were substantially higher than the national average after adjustment for clinical need and were thought worth further investigation by the PCT representatives. Both procedures were thought likely to have had an important financial impact on PCT spending. Additionally, each procedure code was thought to represent a relatively well-defined procedure and patient population which would allow a subsequent technology appraisal to be undertaken.

Benchmarking: PCT1

The adjusted rate of procedure use is much greater in PCT1 than the national average for a large number of procedures (Table 7). In 2009/10 the procedure with the largest estimated absolute difference from the national rate was ‘operations on spinal nerve root’, where adjusted local use was 62 (95% CI 55 to 70) procedures higher per 100,000 residents than the national mean. If PCT1 reduced its rate of this procedure to the national rate, a reduction of around 370 procedures per annum would be made, leading to substantial cost savings. For three procedures PCT1’s utilisation is in the top 10% of all PCTs. Furthermore, three of the 20 procedures with the largest absolute differences between PCT1 rate and the national rate also exhibit high or very high national variation, suggesting that there may be uncertainty about the appropriate procedure rate across England. For 17 of the procedures (the exceptions being ‘excision of gall bladder’, ‘excision of tonsil’ and ‘endoscopic resection of outlet of male bladder’) utilisation in PCT1 remained above the national median for each year between 2007/8 and 2009/10.

The procedure ‘release of entrapment of peripheral nerve at wrist’ (i.e. CTR surgery) was selected for further evaluation in PCT1. The adjusted local rate was 28 (95% CI 20 to 37) procedures higher than the national rate per 100,000 residents. PCT1 had the 19th-largest estimated absolute difference from the national rate. Nationally there was ‘average’ variation between PCTs in the use of this procedure, which suggests reasonable agreement in appropriate procedure rates across England.

PCT1 had a higher rate of CTR procedures than each of three neighbouring PCTs, although with overlapping 95% CIs (Table 8). About 60% of CTR procedures in PCT1 were performed at hospital A and 30% were performed at hospital B. Analysis of the rates of CTR surgery by local authority area of residence demonstrated substantial differences of CTR surgery rates across the PCT (Table 9). The residents of local authority areas predominantly served by hospital A had the highest rates, for example local authority areas U, V and W all had crude CTR surgery rates 62% higher than the national average, whereas districts X, Y and Z, which were more commonly served by hospital B, had crude CTR surgery rates between 22% and 37% higher than the national average.

Crude CTR surgery rates in PCT1 have been higher than the national average since 2004/5 (Figure 3), indicating that high utilisation is not a temporary phenomenon, which might occur if capacity for CTR surgery had temporarily been boosted to meet a waiting list target. Although a reduction in the crude CTR surgery rate was observed between 2004/5 and 2006/7, substantial increases have taken place since then, despite only moderate increases in the national rate.

FIGURE 3.

Temporal trends in CTR surgery in PCT1 and in England.

The gender and age distribution of patients admitted for CTR surgery in PCT1 was similar to the national picture (Table 10), as was the high proportion of procedures (≈ 95%) that were done as day cases. Revision CTR surgery, which is one potential explanation of the high procedure rate in PCT1, is not recorded more frequently. Patients treated in PCT1 were slightly more likely to have bilateral CTR surgery coded. As this finding is based on a supplemental laterality procedure code being recorded, it is possible that this difference represents a variation in coding rather than in clinical practice.

Benchmarking: PCT2

The rate of procedure use is much greater in PCT2 than the national average for a large number of procedures (Table 11). The greatest absolute difference is seen in the ‘destruction of lesion of retina’ procedure, where PCT2 carries out an additional 57 (95% CI 50 to 64) procedures per 100,000 residents than the national average. If PCT2 reduced its rate of this procedure to the national rate, a reduction of around 516 procedures per annum would be made, leading to substantial cost savings. For five procedures, PCT2’s utilisation is in the top 10% of all PCTs. Furthermore, 5 of the 20 procedures with the largest absolute differences between PCT2 rate and the national rate also exhibit high or very high national variation, suggesting that there may be clinical uncertainty about the appropriate procedure rate across England. For the 20 high-utilisation procedures, there was a positive correlation between the procedure rates of neighbouring PCTs (Figure 4). For 18 of the procedures (the exceptions being ‘other vein related operations’ and ‘other therapeutic transluminal operations on vein’), utilisation in PCT2 remained above the national median between 2007/8 and 2009/10.

FIGURE 4.

Comparison of utilisation in three neighbouring PCTs. (a) Graph displaying PCT procedures rates compared to the national procedure rate; (b) key of procedure label.

The ‘incision of capsule of lens’ procedure was selected for further evaluation and technology assessment. The adjusted local rate was 43 (95% CI 39 to 47) procedures higher than the national rate per 100,000 residents. PCT2 had the 26th largest estimated absolute difference from the national rate. Nationally there was ‘very high’ variation between PCTs in the use of this procedure, which suggests that there might be a high degree of clinical uncertainty about the appropriate use of this procedure. The procedure, which is more commonly known as capsulotomy, is typically performed in people who have developed PCO following extracapsular cataract extraction and implantation of an intraocular lens for age-related cataract. Neighbouring PCTs, all of which commissioned ophthalmology services from the same hospital trust, had capsulotomy procedure rates well above the national average (Table 12). Capsulotomy procedure rates in PCT2 have been higher than the national average since at least 2001/2 (Figure 5), indicating that high utilisation is not a temporary phenomenon.

FIGURE 5.

Temporal trends in capsulotomy in PCT2 and in England.

The gender and age distribution of patients admitted for capsulotomy in PCT2 was similar to the national picture (Table 13), as was the high proportion of procedures that were recorded as day cases (≈ 95%). Patients in PCT2 were more likely to have the specific diagnosis code of ‘complications of other internal prosthetic device’ whereas nationally ‘other cataract’ or other diagnosis codes were more commonly used. However, this may simply reflect local differences in diagnosis coding.

Main findings

We used benchmarking of geographical variations in procedure rates to help health-care commissioners identify procedures with high local utilisation which might be candidates for disinvestment. In both PCT commissioning groups, the rate of procedure use was much greater than the national average for a large number of procedures. For some procedures (e.g. operations on spinal nerve root in PCT1 and destruction of lesion of retina in PCT2), rates were more than 50 procedures per 100,000 residents in excess of the national average. In many cases the high local use procedures were not ones that are commonly cited on ‘low value’ procedure lists.

Strengths and weaknesses

The benchmarking approach that we describe does not start from any preconceptions about ‘low value’ procedures and it does not rely on clinician or commissioner introspection to identify areas where local health funds might be used more efficiently. Instead, the data are used to identify procedures where the PCT has a high utilisation rate and then HTR can be employed to explore whether or not the high utilisation is justified. While several procedures identified in PCT1, including varicose vein and carpal tunnel surgery, have appeared on lists of ‘low value procedures’, others have not. The high rate of cholecystectomy is an example where PCT1 is an outlier despite very low inter-PCT variation in procedure rates nationally, suggesting general agreement about the appropriate use of this procedure.

Benchmarking procedure codes does not necessarily lead to straightforward technology reassessment questions. In PCT2, adenoidectomy was initially flagged as having locally high utilisation rates. However, as this procedure may be used in combination with other procedures to treat otitis media, sleep apnoea or chronic tonsillitis, any resulting technology appraisal will be complex. Benchmarking also needs to be conducted carefully to avoid false conclusions. During the course of our work with the PCTs we found apparently high procedure rates due to double counting of procedures where the patient was admitted via an NHS hospital but transferred to an independent sector treatment centre for surgery. This highlights the importance of commissioners working with clinicians from the outset, with willingness from both parties to identify procedures with genuinely high local utilisation and explore the potential causes of this and responses to it.

We used the number of private hospital beds as a proxy measure for the number of private hospital procedures which may be inappropriate, particularly for some conditions (e.g. excision of appendix) where private treatment is unlikely. We have ranked PCTs according to how frequently they used each procedure, relative to national utilisation rates. However, ranks should be interpreted with caution since they are themselves subject to sampling variation. 79 More statistically rigorous methods are available for quantifying whether or not a PCT is truly ‘unusual’ in terms of its rate of utilisation of a procedure. 80,81

Comparisons with other studies

Benchmarking health care in an attempt to improve performance is not a new phenomenon,82 although current financial constraints have brought renewed interest. Benchmarking tools for the NHS in England are now more publicly available than ever, most prominently through the NHS Atlas of Variation in Healthcare45 series, NHS comparators,83 payment by results benchmarker84 and similar tools. However, while the majority of PCTs are aware of the NHS Atlas data, only 34% report using the data to explore causes of variation and potential responses to it. 85 Schang et al. 85 pinpoint one key barrier to the real-world applicability of variations data as the ‘essential ambiguity over the meaning of observed variations’. PCTs can compare their performance with the national norm, but not with any ‘gold standard’ procedure rate.

There is also international interest in the potential of benchmarking to improve health-care efficiency. The seminal work of the Dartmouth group in the USA demonstrated that elderly Medicare patients in higher spending regions did not have better health outcomes or satisfaction with care despite receiving approximately 60% more care,86,87 raising the possibility that substantial savings could be made without detriment to care quality. In Tuscany, a benchmarking exercise including GP and pharmaceutical expenses, inappropriate hospitalisation rates, hospital length of stay and readmission rates estimated that savings equivalent to 2–7% of the health-care budget could be achieved if all health authorities achieved the regional average. 39 However, this finding is based on the questionable assumption that the regional average is desirable and has best outcomes for patients.

Benchmarking should inform and complement other local disinvestment initiatives, such as programme budgeting and marginal analysis (PBMA) and sociotechnical allocation of resources (STAR) approaches. 38,88 In both of these approaches, stakeholder groups discuss and evaluate the evidence supporting numerous investment and disinvestment options within a programme, or programmes, of health care. The aim is to reconfigure care pathways and packages to offer better value care for patients. The NHS spend and outcomes tool89 is designed to help local commissioners identify which health-care programmes are in most need of reconfiguration.

Implications

If there is compelling evidence that a high local use procedure is no longer cost-effective in some or all of the clinical subgroups in which it is currently used, then there is a clear case for local commissioners to intervene to reduce procedure rates. However, when evidence is not compelling, local variation may simply represent legitimate differences in patient preferences for a procedure or unimportant eclectic differences in practice by clinicians. 67 The limited evidence available suggests that patient preferences are unlikely to be a major driver of the large variations in procedure rates observed. One study found that urologists focus more on clinical parameters than patient views when recommending prostatectomy. 90

It has been argued that variation due to eclectic clinical practice may be preferable to an enforced consensus around the prevailing, non-evidence-based, norm,67 provided that the variation is used as a natural experiment to generate evidence comparing patient costs and outcomes between regions of the country. In practice, however, variation is seldom used as an opportunity for research. Furthermore, inpatient procedures are known to be initially costly and in many cases carry known risks to patients; it is the long-term health benefits and potential NHS savings which are more often in doubt. Therefore, it seems reasonable, from the perspective of a local commissioner, that the onus should be on clinicians with high procedure rates to demonstrate that their approach results in better patient outcomes at an acceptable cost, rather than on the commissioners to demonstrate otherwise. Using benchmarking to identify and reduce inappropriately high procedure rates will be only a small part of the response to the ‘Nicholson challenge’ to find £20B in productivity improvements by 2015. 91 For example, if PCT1 did reduce the carpal tunnel surgery rate to the national average it would commission approximately 200 fewer procedures each year and could potentially save approximately £200,000 per annum in inpatient costs. This potential saving will only actually be achieved if, over time, the number of theatre slots and hospital beds are reduced to reflect changing patterns of care. Furthermore, any savings in inpatient costs may be offset to some extent if patients not treated surgically require additional non-surgical care. The relatively small size of the potential savings and the difficulties of intervening to reduce procedure rates might well discourage individual PCTs from investing more time and resources in benchmarking and intervening to regulate procedure rates. The NHS may need to harness economies of scale to address the Nicholson challenge by encouraging nationwide initiatives, such as shared decision-making tools or national guidance on appropriate criteria for surgery, rather than rely on initiatives by individual local commissioners.

There is a real danger that by benchmarking in just one setting (i.e. inpatient care) and on what is recorded routinely (i.e. therapeutic procedures) commissioners will miss the bigger picture and possibly larger opportunities for disinvestment. For example, high rates of carpal tunnel surgery might reflect ‘upstream’ factors such as referral pathways from primary to secondary care or the availability of hand therapy or steroid injections. Therefore it is important that benchmarking lead to a broad conversation about the pathway of care and service configuration, not just a narrow focus on surgeons and the eligibility criteria for surgery.

In 1999, NICE was established to help reduce inequality in access to new treatments across England and Wales. 92 It has attempted to do so through establishing a rigorous, transparent and mandatory technology appraisal process for selected new technologies. However, our benchmarking work reveals the extent to which the postcode lottery of access to treatments in the NHS remains for more established interventions. Indeed the NICE technology appraisal process for new technologies may indirectly exacerbate this by forcing PCTs to cut back in other areas of health care without a transparent or consistent process for making these disinvestment decisions.

Conclusions

Benchmarking can help local decision-makers identify procedures where local utilisation appears to be substantially higher than the national average, even after adjusting for proxies of clinical need. This might represent local overutilisation and an opportunity to provide better value care for the local population. However, benchmarking is vulnerable to inaccurate coding of clinical activity and does not necessarily lead commissioners to an understanding of why local rates are high or, more importantly, what the optimal procedure rate should be based on the available evidence. It is important to recognise that the national average represents a norm and may not be an evidence-based gold standard or provide good clinical outcomes for the patient. We explore these questions further in the next two chapters, which review the evidence on surgical interventions for treating CTS (PCT1) and interventions for treating PCO (PCT2).

Overview of carpal tunnel syndrome

Carpal tunnel syndrome results from pressure on the median nerve as it passes through the carpal tunnel of the wrist. Pressure is much higher in patients with CTS than in patients with asymptomatic wrists and is raised by wrist or finger flexion. 93 Chronic CTS has insidious onset and is related to several prognostic factors including pregnancy, diabetes mellitus, hypothyroidism, osteoarthritis, gout, overweight and occupation. 94 The annual incidence of CTS in UK primary care is approximately 88 per 100,000 males and 193 per 100,000 females. 95 Incidence peaks in the late fifties for women and the late seventies for both men and women. 96 Although most people present with CTS in one hand, more than 50% subsequently develop bilateral CTS. 96

Compression of the median nerve produces pain, paraesthesia (pins and needles) and numbness in the hand. Mild CTS presents with intermittent symptoms often at night. As symptoms progress, paraesthesia may become constant and interfere with daily activities such as holding objects. Severe CTS may cause permanent atrophy of the thenar muscles and permanent loss of sensation in the hand. There is no gold standard for diagnosing CTS. In practice, diagnosis can be by clinical examination and careful history-taking, alone or in combination with electrophysiological testing (i.e. nerve conduction and/or electromyography). 97 Specific diagnostic signs and symptoms such as Phalen’s manoeuvre, Tinel’s sign, and the Durkan test and the patient-completed Katz hand diagram may contribute to the diagnosis. 94 Electrophysiological tests stimulate the median nerve and measure the muscle or sensory response at a point further along the nerve pathway94 with the aim of quantifying the severity of the damage and define the pathophysiology. Prognosis varies with the duration and severity of symptoms. 98,99 A cohort study of 196 patients (274 hands) with CTS observed symptomatic improvement in 34% after 10–15 months of non-surgical care. However, a further 21% reported worsening of symptoms. 98 People with mild to moderate symptoms may respond well to non-surgical interventions. 100 In more severe cases CTR surgery is often performed.

Overview of evidence on non-surgical interventions

There are many putative non-surgical treatments for CTS, of which relatively few are supported by high-quality evidence. 100 The most commonly described treatments for CTS are hand splinting, injection of steroids and oral steroids. 93 A Cochrane systematic review100 of non-surgical treatment for CTS found limited evidence, from one small trial at high risk of bias, that nocturnal hand splinting was better than control (no treatment) at improving symptoms and function in the short term. The Cochrane review also found moderate evidence (consistent findings in more than one RCT) to support the short-term benefit of oral steroids101–103 and therapeutic ultrasound,104,105 and limited evidence (findings from one RCT) for the effectiveness of yoga106 and carpal bone mobilisation. 107

Evidence from a second Cochrane systematic review108 indicated that local corticosteroid injections provided greater improvement in clinical symptoms after 1 month than placebo injections. Two RCTs109,110 (141 participants) at low risk of bias demonstrated a significant increase [risk ratio (RR) = 2.58; 95% CI 1.72 to 3.87] in the incidence of short-term (2–4 weeks) clinical improvements after steroid injection. These two studies recruited patients whose symptoms were refractory to prolonged periods of conservative care (> 3 months110 and 6 weeks of hand splinting109). In both trials symptomatic improvement was achieved in ≥ 70% of patients who had steroid injection, but in only 20–34% of patients in the placebo arm. The Cochrane review authors concluded that, while local corticosteroid injection provided greater clinical improvement in the short term, relief beyond 1 month had not been demonstrated.

Overview of surgical interventions and guidelines for surgery

The number of CTR procedures performed by the NHS has increased since the early 1990s. 96 HES data document a 41% increase in NHS CTR surgery in England in the decade between 2001/2 (n = 39,724 procedures) and 2010/11 (n = 55,957). CTR surgery is usually preformed as a day case procedure under local anaesthesia. 111 Open decompression of the transverse carpal ligament involves an incision in the wrist and then the roof of the carpal tunnel is cut. Surgery can also be done endoscopically; the endoscope is passed through the incision and the transverse carpal ligament is cut from within. Comparisons of endoscopic and open surgical techniques indicate that success and complication rates are broadly similar112,113 and selection of technique is largely based on surgeon preference. 93,113 Potential complications of surgery include infection, bleeding, nerve injury, scarring, persistence or return of CTS symptoms and, rarely, complex regional pain syndrome. 111

The British Society for Surgeons of the Hand (BSSH) recommends surgery for patients with severe symptoms (constant numbness or pain, wasting and/or weakness of thumb muscles) or in mild and moderate cases where conservative treatment has failed (unchanged or increasing severity of symptoms after 3 months). 114 The BSSH suggest that nocturnal wrist splints, activity/workplace modification, hand therapy and/or steroid injections should be considered first in mild and moderate cases of CTS. The BSSH also recommends that electrophysiological testing be used only in equivocal cases (e.g. atypical symptoms, suspected neuropathy or medico-legal issues).

Specific aims of this chapter

The aims of this chapter are:

1. to conduct a rapid systematic review of the evidence on the clinical effectiveness and cost-effectiveness of CTR surgery compared with conservative care

2. to discuss the potential causes of and responses to the high rate of CTR surgery in PCT1.

Defining the research question

We first mapped the A65 procedure code to a clinical question by defining the patient group, intervention, comparators and outcomes of interest. The vast majority of patients (≈ 95% nationally and in PCT1) had ICD-10 diagnosis codes indicating ‘mononeuropathy of upper limb’ while the remainder had diagnoses of forearm fracture or fibroblastic disorders. Almost all procedures (> 99% nationally and in PCT1) were coded as ‘carpal tunnel release’. The procedure code includes both open and endoscopic procedures and was not strongly negatively correlated (i.e. Pearson’s rho < –0.15) with any other OPCS procedure, which indicates that the high rate in PCT1 was not due to the substitution of one surgical procedure for another. Based on these factors the inclusion criteria for our rapid systematic review and technology appraisal were (1) types of studies – RCTs or systematic reviews of RCTs; (2) participants – patients with a clinical diagnosis of CTS with or without electrophysiological confirmation; (3) intervention(s) – any surgical intervention; (4) comparator(s) – any non-surgical intervention, including no treatment; and (5) outcome measures – any reported outcome at 1 month or more after randomisation.

Search strategy

We searched the Cochrane Library and Database of Abstracts of Reviews of Effects (DARE) databases to identify relevant systematic reviews using the term ‘carpal tunnel syndrome’. To find RCTs we searched MEDLINE and EMBASE on Ovid and Cochrane Central Register of Controlled Trials (CENTRAL) on The Cochrane Library. We combined terms for CTS and CTR surgery with a sensitive RCT filter, recommended by the Cochrane Handbook115 (see Appendix 2). Databases were searched from inception to January 2012 and all searches were performed in January 2012. Studies published in languages other than English, letters to journals and abstracts of conference papers were excluded.

Titles and abstracts identified by the searches were screened independently by two reviewers (TM and JB). The full text of potentially eligible studies was assessed for inclusion and data were extracted by one reviewer (TM) and checked by a second (JB). At both stages, disagreements were resolved by a third reviewer (WH). Two reviewers (TM and WH) independently assessed the studies using The Cochrane Collaboration’s tool for assessing risk of bias. 116

Effect sizes were calculated based on the groups to which patients were randomised including all available cases reported by the authors. For dichotomous outcomes, RRs were computed. The published reports did not always contain sufficient information on the outcomes of interest. We contacted the authors of two trials117,118 and subsequently received unpublished data from one117 to supplement our analyses. We also used unpublished data from an additional RCT119 that had been previously obtained by a Cochrane review. 120

We used a forest plot to graphically display the results of primary studies. Visual inspection of the graphs and the I² statistic121 were used to investigate statistical heterogeneity between primary studies. Because of substantial differences in the non-surgical interventions (e.g. splinting, injection or other) used as comparators and high variance in study findings, we elected to provide a narrative summary of results rather than pool results in a meta-analysis. All analyses were completed using Review Manager software (RevMan version 5.1, The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark).

Main findings

Based on a case study of CTR surgery, where local use was 20% higher than the national average, we conducted a rapid technology assessment. Six RCTs which randomised approximately 600 patients have provided evidence on the clinical effectiveness and cost-effectiveness of CTR surgery compared with a variety of non-surgical therapies. All RCTs were vulnerable to detection bias and/or performance bias due to lack of blinding. We elected not to pool study findings because of the differences in comparison groups and outcome measures. Three of four RCTs that reported findings at 3 months concluded that clinical improvement in function or symptoms was better in patients randomised to surgery. In two of these three RCTs the benefit persisted, albeit attenuated, at 6 and 12 months. One RCT concluded that CTR surgery was more cost-effective than wrist splinting. However, the RCTs provide very little information to help surgeons or commissioners identify the patient subgroups where the costs and risks of surgery counterbalance the health benefits.

Strengths and weaknesses

We elected to conduct a rapid systematic review in order to complete it in a short time frame, close to that which would be needed in a real-world commissioning setting. We chose a narrow research question and limited our search to a small number of key electronic bibliographic databases, excluding papers not published in English. Rapid reviews risk exacerbating publication bias if studies in the grey literature or foreign language journals are missed. 126 However, our search identified all RCTs included in a previous Cochrane review on the topic. 120 We also retained the Cochrane risk of bias tool in order to describe the quality of the RCTs included in our review.

Our review was based on aggregate data provided in the published report of RCTs rather than individual patient data (IPD). IPD have several potential advantages including the ability to assess how covariates (e.g. patient age, and symptom duration and severity) modify the aggregate treatment effect. 127 These nuances, which remain hidden in aggregate data, would be very valuable for commissioners and clinicians wishing to tailor threshold policies to target the clinical subgroups likely to benefit most from surgery. However, IPD analyses tend to be more time-consuming and potentially introduce bias if IPD can be obtained from only a proportion of all RCTs. By reviewing evidence from randomised trials we have focused on higher quality evidence. Prospective cohort studies of patient outcomes after surgical and non-surgical care128,129 can provide useful information to help predict the types of patients who will have poor outcomes after treatment. However, cohort studies are vulnerable to selection bias.

Implications

The benchmarking exercise and rapid technology assessment leave PCT1 with a dilemma. It is evident that over a number of years the CTR procedure rate, adjusted for proxies of clinical need, has been significantly higher than the national average. However, evidence from the best available RCTs suggests that surgery is more effective and cost-effective than many forms of non-surgical care in the average patient with mild to moderate CTS. Apart from one post-hoc subgroup analysis, there is no trial evidence available to help refine the indications for surgery.

The benchmarking indicates that PCT1 is an outlier, but commissioners cannot be certain whether the local rate is too high, appropriate or even too low. Although the evidence suggests that CTR is likely to be cost-effective in the average patient with mild to moderate CTS, it might still be overutilised locally in patients where the benefit does not justify the cost and risks. If it wants to act to reduce procedure rates, the PCT has several options available to it. It could tighten up the existing CBA thresholds required before surgery will be funded (the thresholds for PCT1 and neighbours are provided in Table 15); for example by demanding longer trials of splinting or corticosteroid injections or requiring prior authorisation from the PCT before surgery. Emerging evidence indicates that the newly established CCGs frequently use this approach in an attempt to trim costs. 9 However, PCT1 already stipulated a relatively lengthy 6-month trial of conservative care before surgery was allowed in mild to moderate cases. Furthermore, for procedures such as CTR where some criteria are subjective (e.g. symptoms significantly interfere with daily activities) this approach can easily be undermined by patients or clinicians who wish to game the system. Alternatively, the PCT could introduce tougher referral management systems ranging from light touch peer review of GP referrals through to new referral management centres and prior approval to triage patients. However, neither approach is cheap and there is a dearth of evidence that referral management centres are cost-effective. 75

A further option would be for the PCT to implement a CTR surgery patient decision aid to ensure that treatment choices are based on a shared decision between patient and clinicians. By providing standardised and balanced information on the disease and the potential benefits and harms of treatment options, these aids should, in theory, increase the patient’s role in the decision and reduce the potential for procedure rates to be driven predominantly by GP or surgeon judgements about the value of the procedure. Trial evidence demonstrates that decision aids typically do decrease the proportion of patients selecting more invasive surgery. 76 The introduction of decision aids is not cheap; it requires staff training and access to the aids plus good support and potentially longer primary care sessions to support decisions. Finally, the PCT might use contractual levers to bring about a change in practice patterns. This might involve cost and volume contracts to discourage providers from exceeding a given volume of procedures or opening up the service to tender from other providers.

PCT1 decided to put CTR surgery out to tender at a locally agreed price and set up a prior approval process. The tender was awarded to providers operating at two local independent sector hospitals. Patients still had the option to have surgery at the local NHS hospitals, which continued to provide surgery to patients although a reduced number. The likely effect of this on the local procedure rate is unclear. If providers consider the price per procedure too low or the prior approval system too cumbersome, this will act as a disincentive for them to offer surgery, particularly in complex cases. It is clear, however, that it will change the relative balance between costs to the NHS and the benefits and risks to patients. Despite having a relatively high procedure rate, PCT1 may achieve more cost-effective care for its patients with CTS than other PCTs, by keeping the cost of surgery low. However, this does assume that the quality of the service, which was monitored by the PCT clinical quality team, was not affected by transferring more care to the independent sector.

Unanswered questions

Existing RCT publications on the average (cost-)effectiveness of a procedure have a limited value to commissioners and clinicians who are interested in identifying marginal subgroups of patients where the costs and risks of surgery may outweigh the benefits. It has been argued that open sharing of IPD from completed RCTs should be mandated to facilitate subsequent subgroup analyses, potentially pooling data across similar trials to increase statistical power. 130 For example, if the preliminary finding that carpal tunnel surgery is ineffective in patients with least impairment in median motor latency times is replicated in other existing trials, then surgical criteria could be adjusted accordingly. An alternative is to commission new trials with eligibility criteria targeting patient subgroups where the benefit of surgery is believed to be most questionable. However, selecting eligibility criteria which maintain sufficient clinical equipoise for surgeons to participate in the trial will be difficult in established procedures such as CTR. A further approach is to use geographical variation as a natural experiment67 and conduct observational research to compare processes and outcomes of care between high and low procedure rate regions and explore whether or not higher intensity of care is associated with better outcomes. 87 However, these approaches do not provide a short-term solution to the dilemma faced by commissioners.

Conclusions

Benchmarking can help local decision-makers identify procedures where local utilisation is substantially higher than the national average, even after adjusting for proxies of clinical need. This might represent local overutilisation and an opportunity to provide better value care for the local population through reinvestment of funds in other, more cost-effective, interventions. However, evidence is often not sufficient to identify the appropriate indications for the procedure. This leaves local commissioners with a dilemma: they can either allow clinicians to continue to operate at a higher rate than their peers elsewhere in the country or try to enforce standardisation around a norm that is not fully evidence based but might help them meet their duty to balance the budget.

Overview of posterior capsule opacification

Capsulotomy is typically performed in people who have developed PCO following implantation of a posterior chamber intraocular lens (IOL) for age-related cataract. The natural lens is held in place in the eye by a structure called the capsule. During cataract surgery the anterior part of the capsule is removed (or partially removed) to facilitate removal of the natural lens but the posterior capsule of the lens is left intact and in place. The new synthetic IOL is placed on the posterior capsule, which acts as a scaffold or structure to support the new lens. 131 Lens epithelial cells (LECs) are the cause of PCO and can affect vision in two ways. Pearl-type PCO is caused by LECs from the anterior chamber migrating to the lens over the axis of vision, proliferating and then becoming cloudy as collagen is laid down and lens fibre regeneration occurs. Fibrosis-type PCO is caused by proliferation and migration of LECs which transform to myofibroblasts, causing wrinkling and contraction of the posterior capsule, deformation of the capsule bag, and decentralisation of the newly inserted lens, thus reducing visualisation of the peripheral retina. Both processes cause vision to degenerate, light becomes scattered and patients suffer from glare from lights at night and reduction in visual acuity (VA). 131–133 PCO or ‘secondary cataract’ is the most common complication of cataract surgery;134 incidence is reported to be 12%, 21% and 28% at 1, 3 and 5 years respectively. 135

Overview of capsulotomy

Posterior capsule opacification is most often treated using a neodymium:yttrium–aluminium–garnet (Nd:YAG) laser. 132,136 Quick pulses of the laser make precise ablations in the posterior capsule and create a small circular opening in the visual axis. The treatment is usually done using topical anaesthesia with non-sedated patients. The procedure takes approximately 15 minutes, without the need for surgical cuts or stitches and patients can return to normal activities straight away. 137 Early large case series (n = 2110) supported the effectiveness of the Nd:YAG laser procedure, with 98% of procedures successfully opening the posterior capsule. 138 VA was dramatically improved in 84% of cases, with 81% improved to at least 20/40 vision. 138 In a second case series (n = 595), VA improved by at least two Snellen lines in 75% of cases at 6–12 months after the procedure. 139 Improvement in VA was swift: 30% within 24 hours; 75% within 1 week. 139 Contrast sensitivity and glare sensitivity in patients with PCO also improved for all visual angles tested after capsulotomy. 140

Neodymium:yttrium–aluminium–garnet laser capsulotomy may cause complications. Data reported in initial safety studies in the 1980s suggest that, at that time, operative complications included (1) damage (pitting or marking) to the IOL in 20% of cases; (2) rupture of the hyaloid face in 19% of cases; and (3) other, rarer, operative complications such as corneal oedema, bleeding and iris damage in less than 1% of cases. Postoperative complications included (1) raised intraocular pressure – 39% of cases experienced a > 5-mmHg increase in pressure 2–6 hours after the procedure and in 28% of cases the pressure was greater than 30 mmHg; 1% of cases had persistent elevation of pressure at 3–6 months – (2) other, rarer, postoperative complications such as cystoid macular oedema, retinal detachment, pupillary block glaucoma, retinal haemorrhage, iritis and vitritis (which occurred in up to 1.2% of cases). 138 Improved cataract surgery and capsulotomy techniques have reduced the incidence of many of these complications. 141 To further minimise these risks the Royal College of Ophthalmologists (RCOphth) guidelines recommend the use of minimal laser energy to avoid pitting the IOL and note that some clinicians routinely give a hypotensive agent immediately after treatment. 142 Nevertheless a recent survey of UK practice identified considerable variation in yttrium–aluminium–garnet capsulotomy technique and postcapsulotomy management, and highlighted the need for evidence-based guidelines. 143

Other preventative measures focus on improving cataract surgery to prevent the development of PCO and the need for capsulotomy. A meta-analysis of 13 studies of 1456 eyes found strong evidence that fewer eyes fitted with sharp-edge IOLs went on to have laser capsulotomy than those with round-edge IOLs. The review authors also concluded that there was no clear evidence of a difference in outcomes between optic materials, although some studies found that silicone IOLs resulted in less PCO. Few studies evaluated the choice of post-operative anti-inflammatory treatment, but, in those that did, there was little evidence of an effect on PCO development. 144

Specific aims of this chapter

The aims of this chapter are to:

1. conduct a rapid systematic review of the evidence on the clinical effectiveness and cost-effectiveness of capsulotomy and current UK guidelines on the use of this procedure

2. discuss the potential causes of the high rate of day case or inpatient capsulotomy among PCT2 patients.

Defining the research question

Before embarking on the rapid systematic review, we translated the procedure (capsulotomy) into a critical appraisal question by defining the patient group, intervention, appropriate comparators and outcomes of interest. From work described in Chapter 4 (see Table 13), the predominant reasons for performing capsulotomy were coded as ‘other cataract’ (89% nationally versus 83% locally) and ‘complications of . . . implants’ (3% nationally and 14% locally). The gender and age distribution of local patients was similar to the national picture, as was the high proportion of procedures (≈ 95%) that were done as a day case rather than inpatient. Despite the discrepancy between diagnosis codes, which may well be due to local coding practices, Table 13 suggests that patients treated locally are broadly similar to those treated nationally. The OPCS-4 three-character code for capsulotomy (C73) is a broad grouping of six subcodes, giving the opportunity for coders to describe the procedure in more detail. We examined the HES data for this code (C73) and found that, in PCT2 during 2009/10, more than 99% of these procedures were coded as ‘C73.3 Capsulotomy of the posterior lens capsule’ (Table 16). This is quite different from the national situation, where 30% of procedures are coded as ‘C73.4 Capsulotomy of lens not elsewhere classified’. It is unclear whether this represents actual differences in the types of procedure being performed locally or merely variation in coding preferences across the country. Nevertheless, it is clear that it is posterior capsulotomy that dominates in PCT2 and elsewhere.

High use of day case or inpatient posterior capsulotomy in PCT2 may result from a number of reasons. One possibility is that an alternative surgical intervention is used in other PCTs to treat PCO. However, our scoping of the literature did not suggest that this was likely. Given this, in our rapid systematic review, we considered any other non-surgical intervention or watchful waiting as suitable comparator interventions for our technology assessment. Outcomes included VA and measures of vision-related quality of life. As our scoping searches indicated very little trial evidence on the effectiveness of capsulotomy for PCO, we decided not to restrict our review to those studies that measured specific outcomes. Based on the preliminary steps described above, we systematically reviewed the literature to assess the effects of capsulotomy for treating PCO. The inclusion criteria are described in Table 17.

Search strategy

Existing systematic reviews were sought through searches of DARE and The Cochrane Library. Clinical guidelines produced in the UK were also sought to identify the current consensus on the use of capsulotomy among ophthalmologists. To find primary studies (RCTs) we searched MEDLINE, EMBASE and Cochrane CENTRAL using methods and filters described in the Cochrane Handbook in November 2011. 115 We combined terms for PCO, capsulotomy and Nd:YAG laser with a sensitive RCT filter, recommended by the Cochrane Handbook for use in MEDLINE, on Ovid and in EMBASE (see Appendix 6). We searched The Cochrane Library and DARE database to identify relevant systematic reviews in November 2011 (see Appendix 6). Studies published in languages other than English, letters to journals and abstracts of conference papers for which no full study publication was available were excluded. We searched for UK guidelines on the websites of NHS Evidence, NICE, the Scottish Intercollegiate Guidelines Network and the RCOphth. We used simple terms for ‘posterior capsule opacification’; if no relevant records were identified we used the broader term ‘cataract’.

Titles and abstracts identified by the searches were screened for relevance independently by two reviewers (TM and JB). Disagreements were resolved through referral to a third reviewer (WH). Studies that appeared potentially relevant were ordered and full-text papers were assessed for inclusion by one reviewer (TM) and checked by a second (JB). For lack of trial evidence we did not pool RCT data and instead provide a narrative summary of clinical guideline recommendations.

Results of search for systematic reviews and randomised controlled trials

We found no reviews of interventions for treating PCO. The search of DARE found 10 references: one was on the incidence of PCO and the rest were on cataract surgery techniques or interventions aimed at preventing PCO. The searches for RCTs identified 1249 references (Figure 10) after we removed duplicates. However, there were no RCTs comparing Nd:YAG laser surgery with no treatment or delayed capsulotomy. Details of excluded articles are provided in Appendix 7. The literature on this topic tended to focus on interventions employed during cataract surgery to prevent PCO; use of corticosteroids to prevent LECs from proliferating; exploring in vitro methods to prevent PCO; case series of patient outcomes following Nd:YAG laser surgery; preventing or treating elevated intraocular pressure following Nd:YAG laser surgery; comparison of the effects of different Nd:YAG lasers; and the effect of different sizes of hole cut during capsulotomy.

FIGURE 10.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram for selection of PCO RCTs.

Results of the search for guidelines

The RCOphth guidance on the pathogenesis of PCO draws attention to the effects of different IOL shape, edge profile, haptics (side struts of the IOL that hold it in place) and the material from which they are made on development of PCO. The evidence on these topics was graded 1a (the highest – systematic reviews and RCTs). 142 Specifically, square-shaped IOLs with a sharp optic edge profile are known to inhibit migration of LEC.

The RCOphth recommends that, in treating PCO, a Nd:YAG laser be used and that a more invasive surgical approach is required only rarely. The guidelines do not provide references or a level of evidence for these recommendations. The RCOphth notes that in some units laser treatment is performed by appropriately trained paramedical staff. The RCOphth does not discuss VA thresholds or referral criteria for capsulotomy but does state that PCO should be confirmed by presence of characteristic signs visible on slit lamp examination and that symptoms are more important than tests of visual function: severity of PCO correlates poorly with high-contrast VA, and blurred vision, glare, dysphotopsia and reduced contrast in the presence of PCO on slit lamp examination are the common symptoms. We found no national guidance on referral for intervention for PCO,142 although local referral criteria have been developed. 145

Further exploration of the causes of high variation

Capsulotomy rates will, to some extent, be dependent on rates of the initial cataract surgery. However, the evidence suggests that the rate of cataract surgery between 2007/8 and 2009/10 was only slightly higher in PCT2 than the rest of England (Table 18). The only clear difference evident in this table is in the proportion of patients readmitted for inpatient or day case capsulotomy within 1 or 2 years of cataract surgery. The PCT2 rate is approximately 5% per year compared with approximately 1% per year elsewhere. This raises the possibility that some aspect of the initial cataract procedure in PCT2 (e.g. the IOL design or pharmacological agents used to prevent PCO) or the thoroughness of follow-up (e.g. frequency of screening for PCO) may be a cause of the high capsulotomy rate. However, the PCT2 readmission rate for capsulotomy is still below the literature estimate of the annual incidence of PCO following cataract surgery, which is estimated to be 12% at 1 year. 135

In our view the most likely cause of the apparently high rate of inpatient or day case capsulotomy in PCT2 is the uncertainty about whether the procedure should be performed (and coded) as a day case or outpatient procedure. Outpatient procedures are not included in the HES admitted patient data set and are therefore excluded from our analysis. Department of Health reference cost data in 2009/10146 indicate that the majority of capsulotomy procedures (57%) are classified as outpatient rather than day case procedures. It therefore seems quite probable that it is the clinical coding of the setting of care that causes the high rate of inpatient or day case capsulotomy in PCT2. This theory is supported by evidence that many PCTs (see Figure 11) have close to zero capsulotomy procedures recorded in HES admitted patient care data sets. The number of day case or inpatient capsulotomies per cataract procedure performed in PCT2 (Figure 11) was among the top 10 of all PCTs in 2009/10. It seems very likely, therefore, that in many PCTs the vast majority of capsulotomies are recorded as outpatient procedures. This theory was borne out by subsequent investigations in PCT2 that revealed in 2009/10 no (0%) capsulotomy procedures were recorded as outpatient procedures.

FIGURE 11.

Distribution of the proportion of day case or inpatient capsulotomy to cataract procedures.

Main findings

In 2009/10 PCT2 had 43 (95% CI 39 to 47) more inpatient or day case capsulotomy procedures per 100,000 residents than the national average. We found no systematic reviews or RCTs of surgical versus non-surgical interventions for treating PCO. Despite the lack of RCT evidence, Nd:YAG laser capsulotomy is a well-established method for treating PCO, with more than 14,000 inpatient or day case procedures recorded annually. There is limited national guidance on appropriate thresholds for capsulotomy and most PCTs appear to have no policy in place to guide its use beyond clinical discretion.

Strengths and weaknesses

Working with routine data entails a number of challenges for analysis and interpretation. The analysis relies on accurate coding of clinical activity which is consistent between NHS trusts. This means that, rather than assuming that clinical differences drive the headline procedure rates, the first area for exploration should be the potential for coding anomalies to be the cause of variation. Capsulotomy provides a good example of this. Nationally there is very high variance in day case and inpatient capsulotomy rates between PCTs, and PCT2 is at the upper extreme of the distribution. However, as discussed, this is probably purely because of the setting of care (i.e. other NHS trusts performing capsulotomy as an outpatient procedure rather than a day case) rather than variation in the total number of capsulotomies performed. The HES outpatient data set, available since 2004/5, provides the potential for benchmarking analyses to span settings and address this problem. However, in practice, despite improvements over recent years, the sporadic recording of outpatient diagnosis and procedure codes has limited the value of such cross-setting analyses. 147,148

Our findings in context

Advances in medical technologies mean that over time many more procedures will move from inpatient to day case or day case to outpatients. 149 The Department of Health defines day surgery as ‘day case patients who require full operating theatre facilities and/or a general anaesthetic’. 149 By this definition, most capsulotomy procedures might be more appropriately classified as outpatient procedures. According to NHS reference costs,146 the national average unit cost of performing this procedure (Healthcare Resource Group code BZ04Z) in the outpatient setting is £144 (18,950 procedures in 2009/10) compared with £328 in a day case setting (14,174 procedures in 2009/10). Between 2009/10 and 2011/12, PCT2 achieved a complete transformation in the proportion of capsulotomy procedures recorded as being performed in the outpatient setting (0% to 100%), a potential annual saving of £233,000 [1268 × (£328 – £144)] in hospital reimbursements. Potentially then there are substantial savings for commissioners in moving this procedure from the day case to the outpatient setting, although such a move could have a detrimental impact on provider finances. Aggregate HES data indicate that the number of ‘inpatient/day case’ capsulotomy procedures did not decline substantially between 2007/8 (n = 14,218) and 2012/13 (n = 11,562) suggesting that many local commissioners may still be paying a premium for day case procedures. Even though capsulotomy procedures in PCT2 have been transferred to the outpatient setting, there is still very little evidence on the most appropriate indications for the procedure and therefore little to guide commissioners on the evidence-based referral criteria or appropriate procedure rates.

A recent report by the Audit Commission noted widespread inconsistencies in the way short stay procedures were recorded in the NHS. 150 The report concluded that, while NHS coding guidance could be clarified, the main cause of coding inconsistencies was the reluctance of NHS trusts to code procedures as outpatient procedures because of the negative financial implications of doing so. They also noted that valuable data describing a patient’s diagnoses and treatment are poorly recorded when a service moves from an inpatient to an outpatient setting and lost entirely when a service moves to the community. The Audit Commission emphasised the vital role that new NHS bodies (e.g. NHS England and Monitor) should play in improving coding accuracy and consistency.

Conclusions

PCT2 has a high rate of day case or inpatient capsulotomy, predominantly performed as a day case procedure. This rate cannot be explained by statistical chance or by greater clinical need leading to higher rates of cataract surgery and therefore a higher incidence of PCO. Observational studies suggest that a high proportion of patients have improvements of VA after Nd:YAG laser capsulotomy and the procedure is endorsed in guidelines. However, we found no RCT evidence that demonstrated the effectiveness of Nd:YAG laser capsulotomy compared with conservative treatment or delayed surgery and there was very little information on the appropriate clinical thresholds for performing capsulotomy.

The discrepancy between PCT2 and other PCTs is probably due to the coding of capsulotomy as a day case rather than outpatient procedure. There are substantial potential savings for commissioners in moving this procedure from the day case to the outpatient setting. Coding inconsistencies distort PCT differences in capsulotomy procedure rates and distract attention from a debate about appropriate procedure use and rates.

Design and methodology

The study adopted a qualitative design using an ethnographic approach. Two ‘commissioning advisory groups’ situated within separate PCTs were studied over a year. CCGs replaced PCTs following NHS reforms that took place during data collection. The remit of PCTs and CCGs were similar for the purposes of this investigation. Data collection methods included observations of routine commissioning advisory group meetings, combined with semistructured interviews with individuals affiliated with these groups, and individuals who had the potential to be affected by the groups’ decisions.

Observations helped to immerse the researcher (LR) into the field, promoting an in-depth understanding of the context within which those assigned responsibility to disinvest were operating. Interviews allowed further exploration of select topics, from the perspectives of individual members of stakeholder groups involved in or affected by disinvestment. Document analysis of policies and meeting minutes or agendas complemented observations and interviews. As the study progressed, a case study of a specific example of disinvestment was conducted for a more comprehensive view of disinvestment in practice. The case study looked at provision of a given procedure within each region. Data collection consisted of interviews with clinicians whose practice would be potentially influenced by disinvestment, combined with document analysis and informal discussion with commissioning leads from each study site.

Settings and participants

Each commissioning advisory group was responsible for developing commissioning and disinvestment recommendations to the PCTs (replaced by local CCGs as data collection progressed). The PCT/CCG had authority to implement these recommendations locally. Preliminary contact with commissioning group leads checked that the commissioning advisory group meetings were relevant forums for disinvestment decision-making. The groups were situated in sociodemographically contrasting regions of England. Group A served a region that was more rural, with a predominantly white and comparatively wealthy population (PCT1). Group B served a predominantly urban region, consisting of areas with high proportions of ethnic minority populations and pockets of high deprivation (PCT2). Each commissioning advisory group consisted of an array of professional commissioners, including public health consultants, pharmacists, medical directors and directors of finance. Group B also included lay members. Each group hosted regular meetings, attended by representatives from acute NHS trusts and the PCT/CCG in the area.

The selected case study for each region was disinvestment from surgical management of CTS (referred to from here on as ‘CTS surgery’). CTS surgery is a minor orthopaedic procedure, generally conducted under local anaesthetic (also known as CTR or carpal tunnel decompression). Case study clinicians were all hand surgeons who provided this procedure to NHS patients within the study regions. Labels have been used to categorise the various groups of informants participating in this study. Professionals employed by the PCT/CCG have been labelled as ‘commissioners’ (C). Representatives from acute trusts affiliated with commissioning groups are referred to as ‘providers’ (P). Case study clinicians have been labelled as ‘clinicians’ (Clin). Ethical approval for the observation of commissioning group meetings and interviews with individual group members was granted by the University of Bristol, Faculty of Medicine and Dentistry Committee for Ethics (application 111210, January 2012). Ethical approval for the patient and clinician interviews on access to CTS surgery was granted by the National Research Ethics Service Committee South Central – Southampton B (Research Ethics Committee reference 12/SC/0418; August 2012).

Sampling and recruitment

Data collection

Analysis

Audio recordings of interviews and meetings were transcribed in full using standard notation. Transcripts were analysed thematically using the constant comparison method derived from grounded theory methodology. 164 This involves line-by-line coding of transcripts, categorising codes into themes, and developing codes and themes as transcripts are reread in the light of newly collected data. Analysis was primarily conducted by LR, and supported through use of NVivo (version 9, QSR International, Warrington, UK). A sample (10%) of transcripts from interviews and observations was independently analysed by AO-S midway through data collection. Any differences in coding and thematic interpretations were discussed, and additional areas suggested for addition to the topic guide. Descriptive accounts of observation and interview findings were written once analysis was complete. Matrices for major themes from interviews were drawn up and populated with individuals’ quotes. Informants were grouped according to their role to visualise patterns of meaning within each group’s accounts (i.e. commissioners, providers, clinicians). These differences had become apparent throughout the study, but mapping data in this manner helped to identify ‘negative’ cases that conflicted with emerging theories. These exceptions were revisited by rereading transcripts, and described accordingly in reported findings.

Final sample

Eight meetings were observed in total: five from group A and three from group B (Table 19). Twenty-eight individuals took part in interviews. The breakdown of commissioners, providers, lay members and clinicians interviewed in each region can be seen in Table 20.

Presentation of data

Quotations from interviews and observations have been selected on the basis of how clearly and succinctly they illustrate the dominant themes to emerge from this research. Tensions and inconsistencies have been discussed, and quotations from divergent cases presented where relevant. Some quotations have been edited to ease comprehension and/or protect individuals’ anonymity.

Reducing/stopping activity

Most informants defined disinvestment as the reduction or cessation of activity, and/or the removal of funding from an area of health care. Reducing activity and reducing spend are connected, and this clearly came through in all interviews. Some informants mentioned these concepts together in their initial definitions. One commissioner and three providers mentioned finances alone:

What does disinvestment mean to you, or how would you define disinvestment?

Um . . . reducing the amount of money that’s spent on a service.

Interview, group A, PCT1

Other informants focused solely on cessation or restriction of activity (one lay member and a mix of providers and commissioners):

It means that there are certain procedures which will perhaps no longer be done.

Interview, group A, PCT1, lay member 1

To either decommission (treatments/procedures), or try to increase the threshold so that some people do get it, but not all of them. So that is what I think, in very broad terms, disinvestment means.

Interview, group A, PCT1, C1

Choosing not to invest

Two commissioners defined disinvestment in terms of choosing to avoid commissioning new activity:

Disinvestment means there is something there that you’ve identified as an activity that you really want to do [. . .]. But because of conflicting priorities disinvestment means that you’re not going to be able, or you potentially are not going to be able to find the money to do that.

Interview, group A, PCT1, C7

One of the two commissioners (C12) also demonstrated an understanding of disinvestment that was in line with the majority of informants, but clearly felt ‘not investing’ was a component of disinvestment. The statement below was made while discussing the events of a meeting that both LR and C12 had recently attended. C12 had enthusiastically supported an application to fund a new drug, although this was eventually rejected by the wider group. The commissioner’s comment below was in reference to this experience:

I suppose there’s many forms of disinvestment, isn’t there? It’s like making it so difficult to get funding for something that people don’t bother – they either get better or die, you know?

Interview, group B, PCT2, C12

Finding cheaper alternatives

Most informants who discussed funding in their definitions implied, or directly stated, that this would be coupled with reductions or cessations of activity. In contrast, the commissioner below talked about finding cheaper ways of providing a service, with no mention of reduced activity:

Ideally it means that this is a service that we can – or an issue that we can handle in a different way that is cheaper.

Interview, group A, PCT1, C3

Later in the interview, it became apparent that this informant’s idea of disinvestment was more aligned with cessation or restriction of activity. Nonetheless, it was interesting that their initial response, prior to probing, was distinct from that of others.

Summary: disinvestment in theory

This section has unravelled various ways in which individuals understood the term ‘disinvestment’. Whereas there was some confusion about what disinvestment meant, most informants saw it as a term used to denote a reduction of activity or funding within an area of health care. Within this broad group, there were some individuals who emphasised additional criteria as being necessary to classify an activity as disinvestment. Individuals’ interpretations of local disinvestment practices will be discussed in the next section. Once informants had offered their initial definitions, disinvestment tended to be framed in very particular and distinct ways throughout the remainder of interviews, with clear patterns emerging within provider and commissioner accounts. This suggests that, although there can be general agreement over what disinvestment is in theory (i.e. reducing or stopping activity), disinvestment in practice can be interpreted in different ways.

Observed examples of disinvestment

Observed meetings were dominated by new requests for funding and implementation of NICE mandates. Disinvestment decisions were rarely discussed, despite a constant undertone of concern about the affordability of newly commissioned activity:

The business case before us is very clear about how much money this is going to cost, and I’m sure the Director of Finance will say, ‘Well, what should we stop doing instead?’ Um, which is very challenging, and we don’t have any proposals before us to do that.

Observation, group A, PCT1, C2

Disinvestment initiatives, when mentioned in meetings, related to ‘threshold policies’ or ‘not routinely funded’ policies written for specific treatments or services. As mentioned earlier, ‘threshold policies’ set out eligibility criteria patients needed to fulfil prior to access the treatment/service in question. ‘Not routinely funded’ policies stated that the treatments/services would not normally be provided through the NHS. These will jointly be termed ‘restrictive policies’ throughout this chapter. When mentioned, restrictive policies tended to be included on meeting agendas if they were pre-existing and undergoing review. There was no discussion or reference to new disinvestment decisions being made in Group A (PCT1). Group B (PCT2) implemented eight new restrictive policies, although all but one of these new policies were written to establish a formal line of what is and what is not currently funded in the local health-care community. The single example of disinvestment came in the form of a threshold policy designed to ‘control the volume’ of referrals to orthopaedic knee surgeons. A near finalised version of the policy had been developed by a public health consultant outside of the meeting. This draft was brought to the group to obtain final approval of the criteria, and decide on its mode of implementation. A clear process of ‘policy development’ was not observed within the meetings attended by the researcher.

Overall, meetings did not provide much opportunity to observe how disinvestment decisions are made, on account of the rare occurrence of such activities. However, observing these meetings provided valuable insight that helped to place previous disinvestment initiatives (discussed in interviews) in context. Furthermore, meetings provided an opportunity to observe some of the difficulties with implementing disinvestment decisions, even though these decisions had been made prior to the period of observations. These difficulties will be covered in the next main section, Barriers to disinvestment. Finally, our observations provided an insight into the context within which commissioners are expected to engage in disinvestment activity. This allowed us to speculate on some of the reasons underlying the lack of disinvestment decision-making observed. These reasons are presented in the section, subsection Tools and capacity.

Interview informants’ experiences: examples of disinvestment

Identifying candidates for disinvestment

The meetings provided little insight into how disinvestment opportunities were identified. This was largely owing to the lack of disinvestment decision-making occurring within the observed meetings. Interview informants were probed to discuss the approaches used to identify candidate areas for disinvestment, based on previous local experiences. What was notable across all interviews was the absence of any description of a systematic process that could be routinely used. Informants’ responses were based on specific examples of disinvestment recalled, and the routes by which these had been identified. Some responses, particularly those involving benchmarking, tended to be triggered by questions from other areas of the topic guide (i.e. initiated by LR).

Working through a disinvestment process

Interviews provided an opportunity to investigate how previous disinvestment decisions had been made. Most accounts, based largely on restrictive policy formation, were vague and limited in detail. Interview accounts, supported by observations, led to the impression that there was no clear disinvestment agenda in place in either region. There was a very broad process in place (within both regions) for determining the criteria stated within restrictive policies. This process was similar for both regions. First, individuals working within commissioning groups would conduct ‘evidence syntheses’ of published literature to inform draft criteria. These drafts would then be sent to local expert clinicians for comment and feedback (a process commonly referred to as ‘consultation’), although there was no guarantee that this feedback would be taken into account in finalised policies. Implemented policies were placed on the local commissioners’ website, but rarely proactively disseminated to local clinicians. Although interview informants were not aware of any clear disinvestment processes, two broad topics were discussed by commissioners and providers/clinicians from both regions: the process of evidence synthesis and the process of consultation. Each of these topics was the source of disagreement within observed meetings, and a clear point of contention revealed through interview accounts.

Collaboration issues

The issue of providers and clinicians feeling excluded from the decision-making process threatened the sense of collaboration between commissioners and providers/clinicians. With poor collaboration came greater potential for turbulence in implementing disinvestment decisions. This was deep rooted, extending beyond the process of policy formation. An environment of mistrust was apparent through the accounts of commissioners and providers/clinicians from both regions.

Five commissioners from across both regions expressed doubt over providers’ and clinicians’ capacity to remain unbiased in the face of disinvestment decisions. The potential reduction of paid activity was often the backdrop to this:

Because . . . that is the hidden agenda. Because people won’t tell you openly that ‘I’m getting hit’, but they might use other things, some true, some untrue. That is my view, but they may say something different . . . they may use patient safety.

Interview, group A, PCT1, C1

Not all commissioners discussing these ideas were as direct as the informant above. Concerns were sometimes expressed more subtly:

There is a mechanism for making sure the consultants aren’t, um . . . overtreating, shall I say.

Interview, group A, PCT1, C4

Providers and clinicians raised the issue of commissioners’ mistrust as a felt phenomenon. Over half of the providers and most of the clinicians interviewed raised concerns that commissioners had very set views on their motives:

I mean I don’t think doctors want to operate on people for the sake of operating. They always want to make people better. And I don’t think that’s always thought of by the commissioners: it’s almost as though we’re dragging in people off the street for A&E [accident and emergency] and to operate on, and actually, we don’t want to do that.

Interview, group B, PCT2, P4

One commissioner from group B (PCT2) shared others’ views that clinicians have difficulties viewing disinvestment impartially, but was exceptional in that they attributed this to clinicians’ genuine belief that the treatments they conduct are of clinical value, even if this does not hold up to external scrutiny by commissioners.

There was evidence from both regions that providers and clinicians felt particularly uncomfortable with the idea that some restrictive policies had pre-set criteria that defined whether or not a patient was eligible for treatment. This was seen as a reductionist approach to what should be a deliberative decision-making process, based on each individual patient:

That [threshold policies] suggests that it’s black and white, that a person with such and such a scar and such and such a condition is fully deserving of an operation, and this person who has got a real problem, but it doesn’t fit that particular tick box, doesn’t deserve an operation.

Interview, group A, PCT1, P6

That’s why medicine is a long process – it’s an art essentially [. . .], because no books can tell you exactly what to do, for that very reason. And now you’ve been through all of that, and overnight a criteria tick box exercise has been introduced? It just doesn’t agree with medicine, full stop.

Interview, group B, PCT2, Clin5

Commissioners from both regions were aware that clinicians were uncomfortable with the loss of clinical freedom to judge need; however, the issues of mistrust generally prevailed:

We’re challenged on a regular basis, probably weekly, where people are saying, ‘You’re not trusting my clinical judgement,’ or, ‘You’re thinking I’m lying’. [Later] And it’s been said to me, ‘Well I’m a GP, I have a relationship with our patient, I know that they are truly suffering from tonsillitis and they need this procedure, and you should trust me as the clinical lead to refer them in for this treatment’. That’s fine, but for every one GP who is aware of the procedure, and is aware they can only refer in properly, there’s another five who aren’t . . .

Interview, group B, PCT2, C10

Other practices and procedures set by commissioners contributed to providers’/clinicians’ sense of exclusion from decision-making processes. For instance, the auditing or ‘policing’ of compliance with restrictive policies had created tension in PCT1. Providers felt commissioners had not been explicit about their intentions to withhold payment in the event of administrative tasks not being completed correctly:

But I think that the way they went about it wasn’t very clever, and they didn’t really take me into their confidence with what they were trying to achieve. I don’t like that style of saying, ‘Oh just do a clinical audit, just so we know everything is OK,’ and then turning round and saying, ‘And now we’re not going to pay you for 15% of the work’. So that was something that seriously angered me.

Interview, group A, PCT1, P6

The sense of being ‘penalised’ reinforced the sense of division between the two stakeholder groups in the face of disinvestments:

It is very much more the sensation of, as an acute secondary care doctor, is that it [disinvestment] is something that is done to us.

Interview, group A, PCT1, P7

Finally, group A’s approach to finalising decisions clearly marked out the divisions and seemingly unequal distribution of power among commissioners and providers. Group A tended to make decisions on the basis of a voting system, but providers around the table did not share this right to vote. One provider from group A expressed how this reinforced feelings of exclusion from the group:

it doesn’t feel as if the secondary care providers are equal members of that committee.

Interview, group A, PCT1, P1

Reluctance to be explicit about financial constraints

A number of clinicians, providers and commissioners commented on the culture of discomfort in being explicit about cost in the context of health care. Providers and clinicians from both regions were conscious of commissioners’ reluctance to associate disinvestment practices with a desire to save money. Commissioners’ tendencies to portray local disinvestments as ‘waste minimisation’ implied that activity had been conducted unnecessarily. This notion sat uncomfortably with providers and clinicians, promoting disengagement rather than collaboration:

[I think]: ‘don’t turn round and tell me that I am unnecessarily operating on people. Because that’s not ever going to engage me in um – in trying to work through a rationing or disinvesting process.’

Interview, group A, PCT1, P6

One commissioner from PCT2 was exceptional in referring to similar issues touched on by providers:

Yeah and you just think, well, I think, if we’re going to say no to something, then let’s be honest about it. Because these days you can say, ‘OK, you know, it’s got some evidence base, perhaps not a lot of safety, but basically can’t afford it’. At least you’ve been honest then.

Interview, group B, PCT2, C12

Providers’ and clinicians’ impressions that money was the key instigator to previous restrictive policy implementation were paramount. Some provided additional anecdotes supporting their views. For instance, the auditing of compliance with restrictive policies was seen as a covert money-saving exercise, littered with opportunities for commissioners to withhold payment. This was demonstrated through providers’ and clinicians’ accounts:

I mean it’s become a game [. . .]. I mean it’s, ‘How many different ways can we find to not pay you for doing the work, or fine you for doing the work too slowly, too quickly, in the wrong way or whatever?’

Interview, group A, PCT1, Clin4

The use of the term ‘game’ was apparent in two separate clinicians’ accounts of commissioners’ auditing exercises. This use of language suggested that the process was not only a waste of time, but one that could be learnt and manipulated:

It was a way where the [commissioners] could stop paying for stuff even though it was clinically necessary. And we’re now better at ticking the box that says ‘this patient is in pain’. It’s a financial game that’s being played.

Interview, group A, PCT1, P3

Evidence of commissioners’ reluctance to talk about ‘cutbacks’ or financial difficulties was apparent during observations. One of group B’s meetings involved discussion about implementation of a policy restricting access to a cosmetic surgical procedures; although they were not currently provided, commissioners wished to establish a formal line by developing a policy. The discussion that ensued was fraught with concerns that the wording of the policy implied that ‘cost’ was the underlying cause for restriction. This was not deemed to be defensible, with the group concluding that the policy would require substantial reworking:

Yeah because I’m inclined to agree with [commissioner] in that it does look like it’s possibly money, you know, a money thing, when obviously that’s against the spirit . . . but I appreciate it’s challenging if you have lots of people who want to access it.

Observation, group B, PCT2, C9

The culture of disassociating ‘affordability’ from health care in the public arena was seen to be counterproductive to making disinvestment decisions. One commissioner explained how the fear of public outcry or media response had prevented disinvestment initiatives in the past:

If you get somebody who is dissenting – which might be the Local Medical Committee, or it might be a local consultant making a lot of noise in the paper – that’s where your chief exec goes very wobbly, certainly in this organisation it’s happened in the past, and you go back from it and decide that actually you’re not going to be the big brave people to disinvest.

Interview, group A, PCT1, C4

Lack of central support

As indicated earlier, some informants referred to one-off central guidance on potential areas for disinvestment. However, there was general agreement across interviews that central bodies did little to publicly progress the disinvestment agenda, while some informants suggested that central bodies actually hindered disinvestment. For example, NICE mandates to fund new treatments represented a substantial pressure on the groups, and were rarely accompanied with disinvestment suggestions. The way in which these mandates were presented in group A’s meeting was suggestive that NICE was out of touch with local financial pressures:

This is a classic example of somebody making a decision at the centre without any regard to the resources or anything – and then we of course, in future it’ll be the CCGs, to pick up the bill – because the bill is going to be enormous.

Observation, group A, PCT1, C1

Perceptions of the government’s silence when it came to disinvestment made these decisions all the more challenging. Informants from both regions had a strong sense that the government attempted to distance itself from disinvestment, which was thought to be a ‘vote loser’ (interview, PCT1, P1). One provider expressed that devolving these decisions to local bodies had the potential to introduce ‘postcode lotteries’:

And they [government] run a mile from being held responsible for that [. . .]. Politicians want to devolve these decisions to a local level [. . .]. And so, almost by definition, they are introducing a postcode lottery.

Interview, group A, PCT1, P7

Concern over public response over postcode lotteries was raised by numerous commissioners, demonstrating a further hurdle to engaging in disinvestment locally. Nationally applied disinvestment decisions were thought to be more palatable to the public:

Say for instance [the commissioners] don’t pay for homeopathy, so we’ve made a decision, and that’s supported nationally, that’s easy. But where it’s a much more localised decision it’s much more difficult to support.

Interview, group B, PCT2, C10

Overall, the government’s reluctance to become embroiled in disinvestment was suggestive of a more general avoidance of being explicit about financial constraints within the NHS. This was commonly referred to as the main barrier to engaging in disinvestment:

I think it’s very difficult to disinvest when the whole front case for the government is that every time a patient and a doctor sit down and talk about something, then the PCT, the CCG shall somehow magic it up. Until there is some restraint put forward to the public – I mean it’s like, you’ve got to expect treatment whenever you want it, wherever you want it, whatever you want, whatever is in the Daily Mail you ought to be able to have it. Expectations are cranked right up.

Interview, group B, PCT2, C11

Tools and capacity

The lack of systematic tools to identify opportunities for disinvestment was a clear barrier to engaging in disinvestment. As discussed earlier, commissioners and providers were aware of this deficiency. One commissioner went on to point this out as a substantial barrier to disinvestment:

The bit that’s challenging is where you’ve got embedded practice which is not um – does not have a sufficient evidence base behind it. Then you’ve got to identify that practice and disinvest in it. And identifying it is the issue.

Interview, group A, PCT1, C2

Three commissioning group members (two commissioners, one lay member) identified a lack of time and resource capacity as a barrier to pursuing disinvestment:

We haven’t had the opportunity and the time, I think it’s probably as simple as that – having the resources.

Interview, group B, PCT2, C9

The issue of limited capacity clearly emerged through our observations of meetings. Both commissioning groups were inundated with investment proposals. Observations revealed the extensive time and attention invested in working through these funding requests, especially treatments that had been recommended by NICE (mainly group A). Implementing NICE guidance was a process of considering relevance to local practice and fine tuning the details of how national policies could be implemented into specific local systems. These processes were presented as complex and time-consuming in observed discussions:

This is really a piece of work that ties together quite a lot of NICE guidance on the use of antiplatelet drugs. And it was extraordinarily difficult to work out from all these different bits of guidance exactly how we should be using antiplatelet drugs.

Observation, group A, PCT1, C4

Observations also revealed occasional cases where NICE guidance was perceived to lack specificity, thereby creating additional work:

There’s other NICE guidance which gives a little bit – I mean some of the NICE guidance is really unhelpful: ‘It is an option’. You know, I mean and, ‘It is an option,’ is really quite an unhelpful thing.

Observation, group A, PCT1, P2

The NICE guidance is unbelievably woolly.

Observation, group B, PCT2, P5

Main findings in relation to existing literature

The barriers to disinvestment identified in this report were practical and ideological in nature. Practical barriers included limited knowledge, tools and capacity to engage in disinvestment. Ideological barriers related to reluctance to be seen to ration health care, and difficulties in collaboration between commissioners and health-care providers. Interestingly, the practical barriers identified are similar to previously reported obstacles to use of economic evaluation in local settings, as reported by Eddama and Coast. 74 We also noticed parallels between the ideological barriers to disinvestment we identified and Greenhalgh et al. ’s11 reports of professional ‘power struggles’ hindering diffusion of knowledge and innovations. Each of the specific barriers to disinvestment identified in this report has been summarised below, and considered in the light of existing literature.

Strengths and limitations

Our study was unique in that it focused on disinvestment as it is experienced in practice. A key strength of the study was its reliance on inductive approaches to formulate findings that are grounded in actual experiences and events. We sought to approach the research topic with a blank slate, asking questions that might be taken for granted or preconceived prior to this work. Empirical evidence that has been inductively generated is crucial for forming a foundation for subsequent research, and any interventions that seek to push forward the disinvestment agenda. The economic difficulties faced by health-care systems worldwide make this an important area of research, especially given that disinvestment is widely viewed as a challenging prospect. The methods we used not only enabled us to put forward novel barriers to effective disinvestment, but also confirmed and added dimensions to obstacles that have previously been suggested.

Adopting multiple data sources helped to lend credibility to reported findings. Using observations and interviews helped to view the complex issue of disinvestment from different angles, helping to build a more comprehensive view of the phenomenon. Where similar themes have emerged, this has lent weight to findings. Contradictions have also been reported accordingly. These added richness to the findings, and re-emphasised the complex social factors underpinning disinvestment. For instance, the fact that disinvestment was explained differently in group versus individual contexts relates to wider issues, such as the reluctance to be explicit about rationing. We were able to uncover some of the barriers to disinvestment by adopting source triangulation. Comparison of different stakeholder groups’ responses to the same questions illuminated issues of division and threats to collaboration. In some instances, commissioners’ and providers’ responses converged, lending credibility to reported themes. For instance, the lack of central guidance when it came to disinvestment appeared to be a substantial and widely felt issue, highlighting it as an easy starting point for intervention. Our findings will be relevant to health-care decision-makers operating at the local level of resource allocation, in that we have identified broad, fundamental issues that are likely to be transferable. We can confidently report findings with the knowledge that themes were not specific to any of the two sociodemographically contrasting research sites. Individual decision-making groups operating in different geographical and political contexts are likely to experience their own set of barriers, and some of the more specific barriers we cover may not be applicable.

There were a number of limitations to this study. First, as with any controversial subject discussed by public figures, socially desirable responses could have acted as a substantial source of bias. Questions must also be raised about the influence of the NHS reforms that were occurring throughout the period of data collection. The plans to restructure the NHS have been controversial, and could have impact on professional roles. These factors may have indirectly influenced informants’ accounts, particularly if they felt threatened or aggrieved by the reforms. Similarly, disinvestment may be approached very differently, or have a different level of priority, once the reforms have settled in. The provider/commissioner divisions discussed may be retained, but could be altered in light of GPs taking the helm in commissioning. GPs may also be better placed to identify areas for disinvestment, given their experience and knowledge of patient preference and care pathways.

We were able to continue sampling interview informants from each site until we were satisfied that saturation had been achieved. Making claims of saturation for the observed meetings is more problematic, as we had less control over when and how often data collection could occur. While we were able to make a choice to stop data collection at one site (on grounds of saturation), the second site’s meetings were less frequent and often cancelled. Data collection had to end because of timing constraints, although the three meetings observed replicated the same themes, and were broadly similar to the themes emerging from the first site. Assessment of this site’s meeting agendas that were received after the close of data collection reassured us that no new themes were likely to emerge had we continued observations.

As with all qualitative research, the extent to which our findings can extend to other settings is questionable. Further research in this area could adopt survey methodology to investigate the generalisability of some of the views or perspectives to emerge from our study.

Finally, discussions about current disinvestments that were in the process of implementation were absent from meetings. It is possible that disinvestment discussion took place away from the observed commissioning meetings, although the research team established at the outset that the commissioning advisory group meetings were key forums where disinvestment decisions would be discussed. However, the lack of discussion within meetings reflects one of the key issues our research highlights: disinvestment still does not appear to be adopted as a routine exercise. Our adoption of interview methods was an alternative means of investigating how disinvestment is tackled in practice, and, importantly, allowed in-depth exploration of the underlying reasons for the lack of disinvestment decision-making observed. The various practical challenges to disinvestment to emerge through interview, and our observations of the competing demands commissioners face within meetings, offer some explanations of why disinvestment does not occupy a more substantial proportion of commissioners’ agendas.

Conclusion/recommendations

Disinvestment at the local level of health-care decision-making is fraught with difficulties, owing to a lack of tools and capacity to engage in the complex decision-making process. Implementation of disinvestment decisions would benefit from greater inclusion of provider groups. This will require promotion of a shared dialogue, and greater transparency in the process of identifying and negotiating opportunities for disinvestment. Local decision-making groups will either need more guidance from central organisations when it comes to disinvestment, or promote democratic processes in disinvestment decision-making, whereby the public has greater involvement. This may help to tackle some of the social factors underlying the reluctance to take responsibility for disinvestment decisions. Finally, ‘disinvestment’ is a poorly demarcated term that sparks a range of different understandings. Any guidance to improving engagement with disinvestment will need to take account of this, and consider promotion of a less ambiguous term that reduces inconsistent interpretations.

Study design and methodology

The qualitative substudy of patient views was a component of the wider ethnographic study described in the previous chapter. Methodological details relating to the clinician interviews can also be found in the previous chapter. Semistructured interviews were conducted with patients over a 3-month period (January–March 2013). Interviews were conducted face to face wherever possible, but telephone interviews took place if it was not feasible to meet in person.

Sampling and recruitment

The sample of patients was drawn from three NHS acute hospital trusts (all three acute trusts were participating in the wider ethnographic study, described in the previous chapter) situated within the two study regions introduced in the previous chapter (i.e. PCT2 and PCT1). Eligible patients needed to have attended an outpatient appointment in relation to a working diagnosis of CTS within 12 months preceding the date of recruitment. The aim was to achieve a sample of maximum variation on the basis of age, gender and outcome of the most recent outpatient appointment (i.e. admitted for CTS surgery, not admitted for CTS surgery). Clinicians and research nurses identified eligible patients. Clinicians leading this process were participants in the wider ethnographic study. Hospital administrative staff posted recruitment packs to potential participants from December 2012 to March 2013. Each pack consisted of a letter introducing the study; a participant information sheet; a pre-paid postage envelope; and instructions on what to do if the recipient did not speak English. Instructions were translated into the top five most widely spoken non-English languages within each region (based on local council statistics). Patients interested in participating were asked to contact the researcher (LR). Once the researcher had confirmed eligibility in relation to the criteria above, a mutually convenient interview time and location were agreed on. Recruitment packs were sent within time periods to control the volume of responses received. The aim was to stop recruitment at the point of data saturation. This was achieved in PCT1, but recruitment in PCT2 ceased 1 month prior to the study’s official close (to allow time for analysis). Data saturation was considered separately for each region on account of the different contexts of CTS provision.

Data collection: patients

Analysis and presentation of data

The approach to analysis and presentation of data consistent with methods described in the previous chapter.

Threshold policies: PCT1 and PCT2

Threshold policies for CTS surgery were implemented within both regions (edited versions shown in Box 1). The eligibility criteria set out in these policies were broadly similar, although subtle differences in wording were apparent. Both regions’ policies included a statement conveying eligibility for surgery once conservative therapy has been attempted for a specified period (3 or 6 months). Both regions allowed patients to bypass conservative therapy if they had clinically specified symptoms indicating severity, or if symptoms were having a significant impact on patients’ day-to-day lives.

The issues of ‘clinical consultation’ and transparency of the disinvestment decision-making process were discussed in the previous chapter. In summary, clinicians from PCT2 had little awareness of how threshold policies were formed, or the identity of individuals making prior approval decisions. These clinicians expressed frustration that their views had not been acknowledged prior to, or after, implementing these policies. By contrast, clinicians from PCT1 knew the creators of the threshold criteria by name, and were actively involved the formation of these policies.

Prior approval process: PCT2 only

Although each region had broadly similar thresholds for funding CTS surgery, their implementation differed. Trusts in PCT1 were paid for activity, provided that clinicians had documented how patients satisfied the threshold criteria, retrospectively assessed by periodic audit. In contrast, trusts in PCT2 were paid only for activity that had been approved by the PCT in advance. Patients’ fulfilment of threshold criteria was assessed by the PCT’s ‘exceptional funding panel’. This panel was responsible for assessing eligibility for all clinical activity that carried a ‘prior approval’ status. Clinicians were required to submit a prior approval application to this panel for each patient they intended to refer for surgery. The PCT had the right to withhold payment for any CTS surgery that had not been approved in advance, regardless of the fulfilment of threshold criteria.

Introduction of intermediary services: PCT2 only

PCT2 had introduced a physiotherapy-like intermediary service between general practice and the orthopaedic secondary care. Physiotherapists and related clinical experts working in these centres provided conservative care and/or submitted prior approval applications to the PCT.

Standard patient pathways

The standard patient pathway to receiving CTS surgery in each region is summarised in Figure 12. The pathway for PCT1 was relatively simple: GPs were able to directly refer to secondary care, and surgery could take place without prior approval. There was no formal direct route between general practice and orthopaedic services in PCT2. GPs were expected to either refer patients to the intermediary services, or apply for prior approval themselves if they felt surgery was indicated. Clinician interviews revealed that patients were sometimes referred to secondary care orthopaedic clinics from other secondary care departments. In these instances, consultants were required to apply for prior approval themselves.

FIGURE 12.

Typical routes to surgery for PCT1 (left) and PCT2 (right).

Carpal tunnel syndrome surgery out to tender: PCT1 and PCT2

Both regions had put CTS surgery out to tender, allowing other health-care providers in the region (some non-NHS) to compete for contracts. CTS surgery provision was thus fragmented across different providers within each region.

Clinicians’ views on threshold criteria (PCT1 and PCT2)

Looking at threshold criteria alone, clinicians across both regions felt these had little impact on their clinical practice. For clinicians in PCT1, this was largely because of their involvement in writing the threshold criteria. Those consulted during policy formation explained how they had included a ‘catch all’ clause that preserved their ability to act in accordance with their own clinical opinion:

We haven’t rigidly defined ‘severe symptoms’. So there’s a clause that says you can operate on them if they’ve got severe symptoms . . . it’s the ones I want to operate on.

Interview, group A, PCT1, Clin2

Careful framing of patients’ symptoms also allowed clinicians to demonstrate eligibility if they felt the patient warranted surgery. Across the board, clinicians from PCT1 demonstrated that they were very much in control when it came to making decisions about surgery:

Well um you don’t want to um – you can’t invent symptoms, you can’t invent something that isn’t there, and one shouldn’t do it, because that’s a very slippery slope if you do that. But of course, you can guide the patient to give the answers that enable you to put them on a waiting list, giving them questions that will lead them to – um – perhaps make their symptoms more intrusive than they possibly are, if you feel strongly.

Interview, group A, PCT1, Clin3

Clinicians in PCT2 expressed similar views with regard to the threshold policies, but the intermediary services and prior approval system largely removed clinicians’ responsibility of assessing patients against these criteria. By the time a patient had progressed to seeing a consultant, their eligibility for surgery will have been determined by the PCT:

But now they [patients] don’t get to me, because someone else is making some sort of decision that they need steroid injections or they don’t need carpal tunnel surgery or they do.

Interview, group B, PCT2, Clin7

Despite this, patients still filtered through the system without funding approval – usually when referred from other secondary care specialists (e.g. rheumatologists, neurologists). In these situations, the consultant would apply for prior approval, using the threshold criteria to justify why a patient required surgery. One clinician described this process, maintaining that they had never allowed threshold criteria to prevent them from operating on a patient they felt truly required surgery:

It’s like everything else: you find ways around the system. [. . .] I have a patient in front of me who needs treatment. And if I think the best thing for them is carpal tunnel surgery then quite clearly, on paper, they will meet the criteria. So the whole thing is a nonsense.

Interview, group B, PCT2, Clin9

The above perspective was similar to that of a clinician from PCT1:

I mean, I feel I’m being interfered and manoeuvred, but actually when a patient is sitting in front of me, if I need to operate on that patient, a lack of funding or anything like that has not stopped me from doing it – I’ve always been able to do that.

Interview, group A, PCT1, Clin4

One clinician from PCT2 expressed concern that threshold criteria were inappropriate if adopted as the sole decision-making tool. Whereas clinical specialists could draw on tacit knowledge and experience, others (i.e. PCT commissioners) would be fully reliant on an overly reductionist tick-box system:

I fully appreciate we’re in a time of austerity, but you cannot pick a condition and put criteria for treatment. That’s why medicine is a long process – it’s an art essentially [. . .], because no books can tell you exactly what to do, for that very reason. And now you’ve been through all of that, and overnight a criteria tick-box exercise has been introduced? It just doesn’t agree with medicine, full stop.

Interview, group B, PCT2, Clin5

Clinicians’ views on the prior approval process/intermediary services (PCT2)

Most clinicians from PCT2 commented on how the intermediary services and prior approval system delayed patients’ progression to surgery. These delays were thought to increase the risk of poorer surgical outcomes. One clinician felt there had been a notable increase in the proportion of patients seen with late stage CTS since the introduction of the disinvestment initiatives:

So I now see people who have had carpal tunnel syndrome for years, have permanent nerve injury, and several years ago would have been referred a lot earlier by their GP, and would have had surgery, and would not have had the permanent nerve injury which they definitely now have as a result of the delay coming through for surgery.

Interview, group B, PCT2, Clin8

Clinicians from PCT2 were adamant that the intermediary service and prior approval system would not prove cost-effective in the long term, pointing out the possible financial repercussions if patients with poor surgical outcomes attributed this to delays brought about by the convoluted care pathway:

Whilst the population remains relatively uneducated we will get away with it. But some patients will get denied access and will have irreversible neurological damage and will then sue somebody – and then your cost–benefit analysis will go out the window.

Interview, group B, PCT2, Clin8

Another clinician commented on how the administrative effort demanded by the prior approval process was a threat to cost-effectiveness in its own right:

So I can tell you that my secretary’s job, and she’s very competent, now more than 50% of her time she spends doing administration with respect to prior approval.

Interview, group B, PCT2, Clin9

There was a common view among clinicians that most CTS cases would eventually require surgery. One clinician held the more extreme opinion that surgery was the only permanent treatment option for CTS (with the exception of cases related to pregnancy). Delaying surgery was thus viewed as delaying the inevitable, having clear implications for long-term cost-effectiveness:

Because eventually patients with carpal tunnel syndrome need surgery. [Later] They think they’re saving money, but they’re not, it’s costing them a lot more money in the long run.

Interview, group B, PCT2, Clin7

Explaining disinvestment processes to patients (PCT1 and PCT2)

Most clinicians from PCT1 had no experience of discussing disinvestment with CTS patients, on the basis that they had been able to maintain standard practices (by their accounts). There was one exception to this, where one clinician acknowledged that he had, on occasion, had to explain to patients that their surgery would need to wait until they had tried conservative therapy. This clinician reported being very open about the lack of control he had over the situation, diverting responsibility to the PCT:

Yes I have told them that um the reason why they are having this treatment, and not an operation which they possibly expected to get, is that they don’t fulfil the criteria set up by the PCT.

Interview, group A, PCT1, Clin3

Another clinician from PCT1 echoed the above ideas, although this was in reference to other ‘low priority’ procedures:

I mean we blame it squarely and fairly on the PCT; we take no responsibility for it. We say, ‘Look, this is the health authority’s rules,’ and if we’re feeling jaded we do tell them, ‘Yes, it’s just a form of rationing’.

Interview, group A, PCT1, Clin2

Most clinicians from PCT2 saw patients only once their prior approval for funding had been granted. Two clinicians reported seeing patients without secured approval. In these cases, they too reported having no qualms in being explicit about the PCT’s control over the provision of surgery. Although the prior approval forms required clinicians to explain the possibility that funding might not be granted, both clinicians felt they would have explained the process to patients regardless of this obligation:

Oh God yeah, you have to. On the form it says you need to explain to the patient that the funding might get turned down. So, I mean, it’s absolutely clear. I do anyway, because I feel obliged to. They need to understand why they’re waiting for their surgery.

Interview, group B, PCT2, Clin5

One PCT2 clinician who had never needed to explain the prior approval process still suggested a readiness to be open about the loss of clinicians’ control in decision-making. This was reflected in his comment that he would readily support a patient if they were to challenge the PCT on legal grounds:

I suspect at some point someone will successfully sue somebody preventing access. That has not occurred yet as far as I am aware, but I would support that if they ask me as an expert witness.

Interview, group B, PCT2, Clin8

Overview of patients’ pathways

Patients’ pathways from initial consultation to their final secondary care appointments varied considerably, in terms of both the treatments received and the settings in which these treatments were delivered. Patients were divided into two groups based on the types of treatments received. These groups are summarised in Box 2. There was an even mix of patients who had received their CTS diagnosis from primary and secondary care clinicians.

The summary in Box 2 shows how patient choice may have had a role to play in treatment provision. Reports of shared decision-making tended to come from patients from PCT1. Beyond patient choice, reasons underlying the structure or content of treatment pathways could not be reliably ascertained on the basis of patients’ accounts alone. Such information needed to be derived by analysing patient notes or interviewing clinicians about specific patient cases. Thus, our focus was not so much on the extent to which clinicians complied with threshold criteria, but more on patients’ rationalisations of their treatment pathways. We were particularly interested in elucidating patients’ awareness of rationing/disinvestment initiatives, and the extent to which different rationing approaches were acceptable.

Patients’ expectations for surgery

All but a few patients reported that they had been familiar with CTS and its management prior to receiving their diagnosis. Most patients had an expectation that surgery was a certainty or possibility. Lay experiences played an important role in building patients’ knowledge of the condition and its treatment options. When asked to explain their expectations, most patients commented that they were familiar with at least one other person who had received surgery for CTS:

I belong to a craft class and several ladies there have had it done.

Interview, PCT1, P1

Similarly, three patients’ personal experiences of having CTS in the past shaped their expectations for surgery. Two patients had come to expect surgery having conducted their own research:

I was expecting to be referred for surgery, which I was. I think that was the . . .

What created that expectation?

I think a bit of research and everything else as much as anything.

Interview, PCT2

Patients without expectations reported being unfamiliar with CTS prior to their diagnosis, having little knowledge of the condition or its treatment options.

Patients’ rationalisation of pathway stages

Patient involvement in decision-making

As demonstrated above, some patients felt they had influenced the treatment they received by expressing a preference when given a choice. All of these patients were from PCT1. Other patients conveyed their personal involvement or control over their care in more subtle terms. The following accounts suggested that patients had a clear idea of what they wanted, seeing the clinicians as providers of this service. This was suggestive of a very consumerist style of doctor–patient interaction:

Well that’s what I wanted [surgery], and that’s what I said, and that’s what I was expecting.

Interview, PCT1, P2

At the opposite end of the spectrum, a handful of patients indicated a willingness to go along with whatever their clinician recommended:

I thought, ‘I’d come to you and that’s what you recommended was what was best for me so, but I’m not a surgeon and I don’t know what’s what’.

Interview, PCT1, P9

The findings relating to patient involvement and doctor–patient communication should be interpreted with caution. Patients tended to recall the outcomes of appointments, rather than how these outcomes were determined. It may not be surprising that patients who actually chose their treatments were better able to explain how treatment decisions were reached. Preferences of other patients may also have been explored in consultation without the patient being aware of or recalling this.

Patient awareness and understanding of rationing

The focus for other stakeholder group interviews was on ‘disinvestment’, implying a change in CTS surgery provision from how it once was. The actual changes implemented (i.e. introducing thresholds, prior approval systems) will be referred to as ‘rationing’ processes for the purposes of this section. Patients’ awareness of rationing processes were explored through different approaches as the interviews progressed: initially through open-ended questions that encouraged narrative accounts, and later through more structured questioning. For ethical reasons, the interviewer made no suggestion that rationing had influenced patients’ care. This was discussed only if initiated by the participant. Table 22 summarises patients’ awareness of (1) the existence of threshold criteria (PCT1 and PCT2) and (2) the prior approval process (PCT2 only). Most patients were not aware of rationing in any capacity. Of the patients who were aware (n = 5), most were from PCT2 (n = 4). This section focuses on the five patients’ views on the rationing processes that they had become aware of in relation to CTS surgery. These patients perceived that their care pathway had been influenced by factors outside their GP or consultant’s control. In all cases, this knowledge emerged unprompted.

Patients’ perceived acceptability of rationing/disinvestment approaches

Patients’ perspectives on the threshold policies and prior approval processes were explored across all interviews. As most patients had been completely unaware of any rationing practices influencing their care, this concept often had to be introduced by the interviewer in hypothetical terms. As a lead-in, patients were probed to discuss their awareness of the financial state of the NHS, and any previous experiences that may have provoked them to consider costs in the context of health care. Patients were very much aware of constraints, but quick to add that these had never come through to their care. Generally, the topic of financial constraint provoked patients to express their personal satisfaction and sense of privilege in having a health service that was free at the point of care. Financial constraints were often associated with wider austerity measures occurring at the time:

I mean, obviously they are doing various cutbacks. I mean interestingly enough, I only went to the surgery to have the stitches removed yesterday, and I picked up a leaflet on the counter that they were giving to everybody, which said they were encouraging people to do telephone appointments now.

Interview, PCT1, P2

A few patients touched on the idea that the health system, by its own nature, would always be under financial pressure:

They need [to have money], don’t they? And people are living longer so obviously they’re needing more.

Interview, PCT2, P7

Once the topic of finance had been raised, patients were asked for their views on a threshold-based system in place for accessing elective procedures. The CTS case study was used to gauge patients’ views on such a system, by presenting conservative therapy as a prerequisite for accessing surgery. This was presented as a hypothetical scenario for patients who had no awareness of rationing processes in their region. Most patients had been informed by their own clinicians that conservative treatments generally had a lower success rate than surgery, but those not aware of this were provided with this information within the interview. Patients were also asked to assume that CTS surgery was an expensive procedure relative to other approaches to managing CTS.

Patients’ views on the prior approval process were also explored. Those unfamiliar with the prior approval system were asked to comment on who they felt was best placed to make decisions about patients’ eligibility to access elective procedures such as CTS. Some patients found it difficult to engage with this question and required additional probing; in these instances, they were given a list of examples (i.e. GP, NHS managers, consultants, the patients) and asked to provide reasons for their responses.

Summary of findings

Based on clinicians’ accounts, the introduction of threshold policies posed little intrusion or disruption to their standard practices. Clinicians were confident of their ability to ensure that patients’ symptoms fitted with the criteria if necessary. Patients in true need of surgery were thought to be able to receive this, as long as clinicians themselves assessed eligibility against threshold criteria. In contrast, clinicians operating under a prior approval system expressed a loss of control over decision-making. The introduction of an intermediary service and prior approval system was thought to be detrimental to patient outcomes. Clinicians suggested these approaches would eventually impose greater long-term financial penalties on the NHS. Clinicians across both regions expressed little hesitation in being open with patients about rationing/disinvestment processes. Their personal objections to disinvestment, and ability to dissociate themselves from decision-making processes, may have facilitated this.

Patients were very often unaware of rationing/disinvestment measures that had the potential to influence their care. They tended to explain their own care pathways in terms of clinical justification, clinicians’ recommendations and their own treatment preferences. Patients were more likely to be aware of rationing or disinvestment initiatives if this took the form of a prior approval process. Even within this subgroup, knowledge of how decisions were made was limited. Patients showed support for a threshold-based approach to controlling access to CTS, seeing this as a sensible and fair system. Patients emphasised the financial benefit of potentially avoiding surgery by exploring conservative options first, and showed little concern for the potential delay this would impose on receiving surgery. However, this acceptance was connected with the perceived low-priority status of CTS, and clear patient preferences for non-invasive treatment. Patients strongly disagreed with a prior approval process, largely because of the loss of clinician control this implied. Similarly to clinicians’ views, there was also some expressed concern that the prior approval system compromised patients’ potential to benefit from surgery. The only patients to express dissatisfaction with their care had been informed of the prior approval process midway through their care pathways, despite having already been informed that they were in need of surgery.

Our findings suggest that the timing of introducing disinvestment processes may influence patient satisfaction, but this theory will need further exploration. Our findings also raise ethical issues of when and how patients should be informed about disinvestment decisions. Previous commentary on these issues suggests that patients are likely to view clinicians as their advocates. Based on this, care informed by CBA policies – when under the control of clinicians – could still be perceived by patients as being in their best interest. Patients may be more likely to be aware of health resource concerns/rationing if access criteria are seen to be enforced by non-clinicians (e.g. managers, commissioners). 184 Our findings also highlight the importance of effective diffusion of commissioning policies to front-line clinicians. This is particularly important in the case of disinvestment decisions, as policies represent a change in what clinicians might be accustomed to. This distinguishes research in this area from wider rationing issues, where the discussion may be focused on treatments that never were routinely available.

The findings from this case study have implications for refining the disinvestment process proposed in the wider project. The process could benefit from the addition of a step that exclusively considers how disinvestment proposals, once identified and fine tuned, can be disseminated to patient groups and communicated to those receiving care. This step could involve open discussion between commissioners and clinicians involved in the patient pathways under consideration. Key points for discussion could include how, when and by whom disinvestment plans are disclosed to patients.

Strengths and limitations

Our findings provide a detailed insight into patients’ awareness of implemented disinvestment initiatives that had the potential to have substantial influence over their care. We present findings on patient’s views on the acceptability of rationing approaches in a particular clinical context, focusing on people who were able to draw on actual experiences. The literature on qualitative accounts of disinvestment is sparse, especially in relation to the patient/public perspective. A key strength of this research is its detailed exploration of different policies as experienced by clinicians and patients, and the potential it lends to formulating more focused future research questions.

There were a number of limitations to this study, most of which centre around the case-specific nature of each patient’s interview. The individual pathways to surgery challenged our ability to gauge when data saturation had been achieved. The research could have benefited from additional interviews, potentially from other sites that have implemented threshold or prior approval approaches to controlling the volume of CTS surgery. We have conducted exploratory work that provides a good foundation for further research. It is likely that the extent to which patients noticed any policy-derived impact on their care was partially dependent on the specific nature of their case. For example, some of the prior approval patients may not have noticed any significant impact on their care if they were cases that clearly exceeded eligibility criteria. This may have affected how quickly/easily their application was processed by the prior approval panel. We did not observe the prior approval decision-making process, and are thus unable to comment on whether certain patients are moved through the system quicker than others. A notable limitation of this work was the absence of patients who had been denied access to surgery. Sampling from secondary care clinics alone immediately excluded patients at earlier stages of the pathway. This group may have had greater experience of being denied access to surgery on the basis of disinvestment policies. This is especially true for PCT2, where the prior approval process and intermediary treatment centres could have significantly filtered the population of patients we had access to. This raises questions about the transferability of findings to the wider CTS population.

The findings from patient interviews may be limited by recall bias and patients’ interpretations of what had occurred during their care pathways. This was not so much of a problem in terms of exploring patient perspectives on rationing approaches, but had implications for our reports of shared decision-making and disclosure of rationing processes within consultations. Further research in this area may overcome these issues by observing/recording actual consultations within which issues of disinvestment may be discussed. Furthermore, the study focused on one procedure (CTS) in one specialty (orthopaedics). The patient informants perceived the procedure to be low priority in general, particularly relative to life-saving treatments. These views may have also been influenced by informants’ perceptions of symptoms having little to no impact on their quality of life. Further research is needed to explore how patient perspectives vary in relation to disinvestment from a range of health interventions, with a focus on how attitudes vary according to the nature of the intervention (e.g. life-extending treatments versus interventions that improve quality of life). Similar work has already been undertaken in relation to rationing interventions for chronic obesity versus breast cancer. 181 Further work could also investigate how patient attitudes vary within a single procedure according to symptom severity. The present study selected CTS in line with the wider project’s aims. This work was successful in highlighting avenues for potential further research, while future studies will be in a better position to purposefully select clinical specialities best suited for exploring these new lines of enquiry. It is important to acknowledge that we interviewed secondary care clinicians who potentially had a conflict of interest in maintaining provision of health services they provide. The trusts these clinicians were operating within may have experienced a significant loss of revenue. The clinicians’ accounts presented need to be interpreted with these factors in mind.

Findings in relation to the wider literature

Like other studies,181 we found that patients were generally aware of constrained resources in the context of health care, but often failed to relate this to their own care. Drawing conclusions about awareness of disinvestment from their own care is problematic: we cannot be certain if individual patients’ care was actually shaped by disinvestment policies (and, if so, to what extent). This limitation is particularly evident from PCT1 clinicians’ portrayal of thresholds, which were deemed to have little effect on their standard practices. We can be more confident that disinvestment had an impact on some prior approval patients’ pathways.

Prior research in this area suggests an overall desire for openness about rationing with regard to an individual’s care, from both public185,186 and patient181 points of view. However, reasons underlying this desire have been centred around the prospect of not being aware of the type of services that could be available, thus missing out on opportunities to access this care via alternative means (e.g. appeals, private care, etc.). Whether or not these issues are transferable to a disinvestment context will depend on the nature of the disinvestment in question, and patients’ pre-existing awareness of the services that would have been available prior to disinvestment. Previous research suggests clinicians have a desire to be open with patients about rationing, but find this challenging in practice. 181,187 Practising clinicians’ views on being explicit about financial issues have not been investigated within the unique context of disinvestment. We found that clinicians who sensed a loss of control in making treatment decisions reported having no qualms in being open with patients about the reasons for service restriction. Clinicians felt able to divert responsibility to the PCT whenever explaining rationing processes. Views on such matters were more difficult to elicit from clinicians in PCT1, as informants felt they had retained control over treatment decisions, and so any restriction of treatment was effectively their clinical decision.

Future research

This exploratory research suggests a number of further research questions. There is a clear need to investigate how disinvestment decisions are communicated to patients, and patients’ subsequent acceptance of such approaches. This research would benefit from observations of actual consultations, and paired analyses of clinical and patient perspective for specific cases. This emerging area of research would also benefit from application to a range of conditions with different treatment pathways, comparing how responses vary by the nature of the condition and the treatment options available. Although our sample was limited, we found evidence to suggest that clinicians are not always aware of disinvestment decisions made by commissioners. Our findings thus support further research into the best ways to communicate disinvestment decisions to practising clinicians, especially GPs, who often create or maintain patient expectations of what their care pathway is likely to entail. Further research also needs to consider the appropriateness of prior approval systems, and particularly whether or not a check-list operated by health-care managers can safely replace specialist clinical opinion. The implications of such systems on clinical outcomes and patients’ quality of life (as well as cost-effectiveness) need to be urgently investigated, as indicated by the patients and clinicians in this study.

Implications for local commissioners

Implications for national policy-makers and research funders

Developing tools for commissioners to work with stakeholders to prioritise disinvestment

Commissioners are faced with an array of tools which aim to help them identify, prioritise and implement investment and disinvestment initiatives. These include a variety of benchmarking tools and various multicriteria decision analysis methods such as PBMA, sociotechnical allocation of resources and the Portsmouth scorecard. 202 These methods vary in the extent of stakeholder input and the level of complexity and time required. Research comparing these methods as means of identifying and implementing disinvestment initiatives and exploring their optimal design in order to engage clinicians and the public in a collaborative and transparent decision-making process would help commissioners establish sustainable local investment and disinvestment procedures.

Evaluate the costs and consequences of commissioner interventions to regulate procedure rates

It is already clear that CCGs will continue, and possibly ramp up, the methods used by PCTs (e.g. CBA, prior approval, referral management centres) to regulate access to medical procedures on the NHS. However, there is very little evidence on which methods are effective and some concern that the costs and bureaucracy of more intensive methods may defeat their objective of providing more efficient care, and have the potential to lead to poorer patient outcomes. Alternatives to these criteria-based approaches include patient-centred approaches, such as shared-decision-making aids, and contractual approaches (e.g. cost and volume contracts and tendering with the independent sector), all aimed at efficiently providing appropriate levels of care. Research on the relative merits and drawbacks of these approaches is needed to guide commissioners. This includes the research that we initially planned to evaluate the effectiveness of local commissioners’ policies for CBA in reducing the volume of procedures of uncertain value.

Evaluate the costs and outcomes of care in regions with high and low procedure rates

Observational and qualitative research on procedures with high geographical variation in localities with ‘high’ and ‘low’ procedure rates will help better our understanding of the reasons for variation and appropriate responses to it. This research should focus on not just those patients who receive surgery (e.g. individuals who receive carpal tunnel surgery), but the wider cohort (e.g. individuals with CTS) seen in primary and secondary care. This would help tease apart high procedure rates caused by high primary care consultation and referral rates from those caused by lower thresholds for surgery. Measuring the costs and outcomes of care for this wider cohort could help determine if high-cost care is associated with better outcomes. 86,87

Methodological work to develop a procedure/diagnosis code matrix

We have highlighted the limitations of working with the OPCS coding framework, which was developed for clinical rather than research purposes. For some purposes, the level of aggregation provided by OPCS three-character codes was too crude, while for other purposes the level of aggregation was too detailed. In all cases, procedure rates had to be cross-matched with diagnosis codes to account for procedures that were used for multiple distinct clinical groups (e.g. tonsillectomy and adenoidectomy for chronic tonsillitis and sleep apnoea). In order to make these coding frameworks more amenable for research, work is needed to develop a clinically coherent matrix of commonly occurring procedure/diagnosis group pairings. Given that there are thousands of OPCS procedure codes and ICD diagnosis codes this would be a major undertaking. It would also need to occur at different levels of aggregation so that it could be used at a high level (e.g. to identify the number of total hip replacements for patients with osteoarthritis) and also at a more detailed level (e.g. to identify the number of cemented, cementless and hybrid hip replacements for patients with osteoarthritis).

Conduct benchmarking on patient-reported outcomes collected in routine data

Our geographical variations analysis and local benchmarking was conducted on the process measure of procedure rates. However, now that patient-reported outcome measures are routinely collected for four surgical procedures (hip and knee replacements, hernia and varicose vein repair), there are opportunities to examine interhospital variations and benchmarking on the costs and outcomes of surgery. 203 Estimating the counterfactual (e.g. the costs and outcomes that would have occurred if surgery had not been performed) is problematic, but preliminary research in this area has found quite large variation in estimated costs and QALYs following hip replacement. This work could be expanded particularly if patient-reported outcome measures are introduced more widely in NHS routine data.

Search strategy

1. randomized controlled trial.pt. (316,214)

2. (randomized or randomised).ab,ti. (282,509)

3. placebo.ab,ti. (130,897)

4. dt.fs. (1,490,078)

5. randomly.ab,ti. (160,982)

6. trial.ab,ti. (274,934)

7. groups.ab,ti. (1,082,267)

8. 1 or 2 or 3 or 4 or 5 or 6 or 7 (2,773,556)

9. exp animals/ (15,618,000)

10. exp humans/ (11,977,239)

11. 9 not (9 and 10) (3,640,761)

12. 8 not 11 (2,353,235)

13. Carpal Tunnel Syndrome.mp. or Carpal Tunnel Syndrome/ (7170)

14. (carp$ tunn$ or tunn$ syndrom$ or carp$ syndrom$).mp. (8672)

15. (nerve entrapment or nerve compression or entrapment neuropath$).mp. (10,579)

16. median nerve entrapment.mp. (92)

17. nerve compression syndromes/ or nerve compression syndrom$.mp. (9137)

18. or/13-17 (17,977)

19. epineurotomy.mp. (35)

20. reconstruct$.mp. (152,873)

21. release.mp. (372,093)

22. SURGERY/ or surgery.mp. (679,668)

23. neurosurgery/ (11,727)

24. SURGICAL PROCEDURES, OPERATIVE/ or surgical.mp. (789,562)

25. operat$.mp. (670,857)

26. 19 or 20 or 21 or 22 or 23 or 24 or 25 (2,014,211)

27. 18 and 26 (6826)

28. 12 and 27 (797)

Search strategy

1. Carpal Tunnel Syndrome.mp. or Carpal Tunnel Syndrome/ (9992)

2. (carp$ tunn$ or tunn$ syndrom$ or carp$ syndrom$).mp. (12,406)

3. (nerve entrapment or nerve compression or entrapment neuropath$).mp. (11,509)

4. 1 or 2 or 3 (21,802)

5. epineurotomy.mp. or carpal tunnel release/ (81)

6. nerve compression/ (9898)

7. surgical approach/ or surgical technique/ or nerve surgery/ (264,182)

8. (surgery or surgical or operation or reconstruct$).mp. (1,968,769)

9. 5 or 6 or 7 or 8 (1,974,934)

10. Randomized Controlled Trial/ (295,607)

11. Clinical Trial/ (823,118)

12. Multicenter Study/ (86,014)

13. Controlled Study/ (3,671,242)

14. Crossover Procedure/ (31,644)

15. Double Blind Procedure/ (102,550)

16. Single Blind Procedure/ (14,668)

17. exp RANDOMIZATION/ (55,456)

18. Major Clinical Study/ (1,563,506)

19. PLACEBO/ (191,051)

20. Meta Analysis/ (58,505)

21. phase 2 clinical trial/ or phase 3 clinical trial/ or phase 4 clinical trial/ (42,895)

22. (clin$ adj25 trial$).tw. (259,385)

23. ((singl$ or doubl$ or tripl$ or trebl$) adj25 (blind$ or mask$)).tw. (138,179)

24. placebo$.tw. (164,030)

25. random$.tw. (677,054)

26. control$.tw. (2,484,239)

27. (meta?analys$ or systematic review$).tw. (40,103)

28. (cross?over or factorial or sham? or dummy).tw. (119,732)

29. ABAB design$.tw. (104)

30. or/10-29 (6,557,827)

31. human/ (12,776,609)

32. nonhuman/ (3,772,499)

33. 31 or 32 (15,850,473)

34. 30 not 33 (864,912)

35. 30 and 31 (4,084,918)

36. 34 or 35 (4,949,830)

37. 4 and 9 and 36 (3065)

Search strategy

1. randomized controlled trial.pt. (322,599)

2. controlled clinical trial.pt. (84,057)

3. randomized.ab. (227,373)

4. placebo.ab. (130,354)

5. drug therapy.fs. (1,520,338)

6. randomly.ab. (163,568)

7. trial.ab. (235,465)

8. groups.ab. (1,082,545)

9. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 (2,816,687)

10. exp animals/ not humans.sh. (3,722,514)

11. 9 not 10 (2,390,384)

12. exp cataract extraction/ (25,710)

13. (phacoemulsificat$ or capsulorhexis).tw. (5506)

14. ((extract$ or surg$) adj3 cataract$).tw. (16,486)

15. exp lenses, intraocular/ (11,795)

16. exp lens implantation, intraocular/ (6718)

17. exp lens capsule crystalline/ (3514)

18. 12 or 13 or 14 or 15 or 16 or 17 (35,557)

19. (posterior adj3 capsul$ adj3 opaci$).tw. (1323)

20. PCO.tw. (3164)

21. aftercataract.tw. (20)

22. secondary cataract$.tw. (402)

23. capsulotom$.tw. (2078)

24. ((Nd or Neodymium) adj3 YAG).tw. (2188)

25. 19 or 20 or 21 or 22 or 23 or 24 (7817)

26. 11 and 18 and 25 (493)

Search strategy

1. exp randomized controlled trial/ (292,701)

2. exp randomization/ (54,986)

3. exp double blind procedure/ (101,701)

4. exp single blind procedure/ (14,442)

5. random$.tw. (665,175)

6. 1 or 2 or 3 or 4 or 5 (779,881)

7. (animal or animal experiment).sh. (3,122,387)

8. human.sh. (12,650,241)

9. 7 and 8 (504,084)

10. 7 not 9 (2,618,303)

11. 6 not 10 (707,167)

12. exp clinical trial/ (872,064)

13. (clin$ adj3 trial$).tw. (215,171)

14. ((singl$ or doubl$ or trebl$ or tripl$) adj3 (blind$ or mask$)).tw. (132,889)

15. exp placebo/ (187,497)

16. placebo$.tw. (161,437)

17. random$.tw. (665,175)

18. exp experimental design/ (6601)

19. exp crossover procedure/ (31,195)

20. exp control group/ (27,882)

21. exp latin square design/ (189)

22. 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 (1,452,360)

23. 22 not 10 (1,370,140)

24. 23 not 11 (671,160)

25. exp comparative study/ (881,702)

26. exp evaluation/ (164,837)

27. exp prospective study/ (176,527)

28. (control$ or prospectiv$ or volunteer$).tw. (2,872,425)

29. 25 or 26 or 27 or 28 (3,725,532)

30. 29 not 10 (3,106,375)

31. 30 not (11 or 23) (2,495,266)

32. 11 or 24 or 31 (3,873,593)

33. exp cataract extraction/ (30,724)

34. (phacoemulsificat$ or capsulorhexis).tw. (6653)

35. ((extract$ or surg$) adj3 cataract$).tw. (18,706)

36. lens implantation/ (4322)

37. lens implant/ (15,141)

38. exp lens capsule/ (2937)

39. 33 or 34 or 35 or 36 or 37 or 38 (41,455)

40. exp aftercataract/ (315)

41. (posterior adj3 capsul$ adj3 opaci$).tw. (1566)

42. PCO.tw. (2554)

43. aftercataract.tw. (49)

44. secondary cataract$.tw. (462)

45. capsulotom$.tw. (2301)

46. ((Nd or Neodymium) adj3 YAG).tw. (2824)

47. 40 or 41 or 42 or 43 or 44 or 45 or 46 (8317)

48. 32 and 39 and 47 (895)

Search strategy

#1 (MeSH descriptor Cataract Extraction) 91

#2 phacoemulsificat* or capsulorhexis 1610

#3 cataract* NEAR/3 (extract* or surg*) 3374

#4 MeSH descriptor Lenses, Intraocular explode all trees 777

#5 MeSH descriptor Lens Implantation, Intraocular explode all trees 779

#6 MeSH descriptor Lens Capsule, Crystalline explode all trees 223

#7 (#1 OR #2 OR #3 OR #4 OR #5 OR #6) 3924

#8 MeSH descriptor Posterior Capsule of the Lens explode all trees 3

#9 posterior NEAR/3 capsul* NEAR/3 opaci* 228

#10 PCO 363

#11 aftercataract 25

#12 secondary cataract* 386

#13 Nd near3 YAG 502

#14 Neodymium near3 YAG 101

#15 capsulotom* 247

#16 (#8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15) 1369

#17 (#7 AND #16) 564

Search strategy

1. MeSH DESCRIPTOR cataract extraction EXPLODE ALL TREES 21

2. MeSH DESCRIPTOR lenses, intraocular EXPLODE ALL TREES 9

3. MeSH DESCRIPTOR lens implantation, intraocular EXPLODE ALL TREES 9

4. (phacoemulsificat*) OR (capsulorhexis) 19

5. (extract* adj3 cataract) OR (surg* adj3 cataract) 19

6. #1 OR #2 OR #3 OR #4 OR #5 45

7. (posterior adj3 capsule* adj3 opaci*) 8

8. (PCO) 5

9. (aftercataract) 0

10. (secondary cataract) 0

11. #7 OR #8 OR #9 OR #10 10

12. #6 AND #11 10

13. #6 AND #11 10

Initial preamble: background of participant

Could we start by you just telling me a little bit about yourself, and what you do?

Probe: what’s their educational background/what roles have they had in the past/how did they become a member of the commissioning group/what is their role within the commissioning group.

The role of the commissioning group, and the impact of funding constraints

Can you explain what the CAF [Commissioning Advisory Forum]/CPG [clinical priorities group] does? Explore their definitions of priority setting and the processes involved.

Are there any challenges involved in this? Explore: can these be overcome? Do they overcome these? How could they be overcome?

(If not yet raised) Has the current economic climate had an impact on the way funding decisions are made? Explore: how; if answer ‘no’, probe further about NHS funding constraints, and what it has meant for decision-makers/how they have responded.

Do you think there will be any commissioning challenges in the coming years? Explore: what, how, why?

Will commissioners need to do things differently over the coming years? Explore: why.

[If so] What do you think they/you should do?

Previous experiences of/attitudes towards disinvestment

What does disinvestment mean to you?

How do you feel about disinvesting from health procedures of low clinical value? Explore: is this a viable solution to being able to fund new treatments? What other methods could be used?

Do you have any experience of disinvesting? If so, what processes/methods have you used? Explore methods: what do they involve, etc. (Partial disinvestment, full, etc.)

What factors/which groups did you need to consider whilst engaging in disinvestment? Probe: patient perspectives sought? Where was evidence obtained from? Other stakeholders, etc.

Did the methods work? Explore: why/why not/what should be done differently?

What do you think are the best ways of saving money? Explore alternatives; probe for other ways of saving money (e.g. changing the way services are delivered, where delivered, by whom, etc.)

Expectations of the proposed disinvestment method

Have you heard of the disinvestment method suggested by the UoB [University of Bristol] research team? If no, introduce the method in lay terms

What do you understand from it? If necessary, explain the method in lay terms

Do you think variations in procedure rates are a suitable tool for identifying candidates for disinvestment? Explore: why/why not.

How would you define a successful disinvestment method? Can they give examples? Explore: ideas surrounding saving money, ‘fairness’, quality of healthcare, clinical consequences.

Do you expect this disinvestment method to succeed? Explore: why/why not?

What do you think the barriers to success might be? Probe: difficulties of changing established practice, etc.

How do you think we can overcome these barriers?

What impact do you think this disinvestment process will have on patients, GPs/other stakeholders/the PCT?

Wind down and closure

Researcher thanks informant for their time, asks if there are any additional comments they would like to make. Some of the main points of the interview may be summarised, if appropriate. The researcher may use this opportunity to check and confirm certain ideas/impressions built up during the interview.

Initial preamble/background

Could we start by you just telling me a little bit about yourself? Professional role.

Could you give me a brief overview of what surgery for carpal tunnel syndrome (CTS) involves?

Referral

How do you think GPs reach decisions about which CTS patients to refer to secondary care? Are referrals well thought out? Guidelines/policies? Any influence from pat preferences/demands?

What happens after a GP refers a patient to the outpatient clinic?

Do you find that patients have expectations of what may happen? Is this common? Do they expect surgery?

Accessing treatments

How often do patients go on to receive surgery for CTS?

How do you decide on which patients are admitted for surgery?

What happens to patients who don’t get admitted for surgery?

Are there any limitations on your ability to provide this service for patients who need it (in PCT 1/2)? Explore.

What might affect whether a patient gets access to surgery? Explore.

Clinical variation

Do you, or your clinical team, ever look to other trusts to compare clinical practice?

Have you/your team, everlooked at activity rates for CTS surgery? Explore: internal benchmarking or external? Trends over time? Differences btw [between] trusts? Where do they think PCT1/2 might rate in terms of level of activity?

(LR:) The study I am working on has done some benchmarking work to identify secondary care elective procedures that show variation in activity, where this variation is unlikely to be a result of differences in clinical need. We are not sure why variation exists, and do not know what the optimum level of activity is, but we found that the rates of CTS surgery in PCT1/2 are higher than the national average. I was interested in exploring some of your thoughts on this . . .

Does this surprise you? Do you feel there are any explanations as to why PCT1/2’s activity rates for CTS surgery appear to be higher than the national average? Is this acceptable?

Do you think variations in clinical activity matter? Should we be trying to level out activity to a ‘norm’? Explore why/why not.

Role of commissioners

Do you feel commissioners influence demand for CTS surgery in any way? (If yes) How? Why do you think they do this?

If threshold policies mentioned, continue. If not mentioned, ask if clinician is aware these exist. If yes, proceed with following questions. If not, ask if they have heard of threshold policies for any procedures, and adapt following questions.

When did you first become aware of these policies?

What do the policies say?

What do you think the policies are designed to do?

How are the policies enforced? Explore auditing.

Do you believe these policies have impacted your clinical practice in any way? Explore. Also, do policies influence rates of activity?

Do you believe these policies are appropriate? Clinically appropriate? Fair? Would clinician change the policies in any way? Explore.

Do you ever come across situations where the parent wants the child to have surgery, but you have had to say no for any reason? Examples. Explore how clinician felt. Patient reaction. How clinician ended consultation.

(If no to the above) What about any other clinical procedures that may have policies attached?

General views on gate-keeping

Who do you think should have a role in determining which elective surgical treatments NHS patients receive?

On what factors should the decision to have surgery be made?

Draw on some of the issues raised, and ask whether these issues will change with NHS reforms.

Wind down and closure

Researcher thanks informant for their time, asks if there are any additional comments they would like to make. Some of the main points of the interview may be summarised, if appropriate. The researcher may use this opportunity to check certain ideas/impressions built up during the interview.

Background

When did you first start to have problems with your wrist/hand(s)?

How did you deal with this initially? Did they know what was wrong? How? Did they seek any information or advice? Did they try any treatments?

At what stage did you decide to seek medical care for your symptoms? Explore triggers to consult.

Pathway leading to outpatient appointment

Do you remember your first consultation with a GP? Can you talk me through what happened?

If not covered above:

- What did you expect the doctor would do?

- What were you hoping the doctor would do?

- What was the outcome? (including recommendations/treatments)

- Did you feel this decision was made WITH you? (Probe for their involvement/did they express what they wanted/preferences? Why/why not?)

- Did you understand why this decision was made?

What happened after this consultation?

- Did you follow recommendations/advice/take prescribed treatment? Did this help?

- Did the types/intensity/frequency of symptoms change? If so, how?

Thinking about the GP consultation that resulted in your outpatient referral to hospital . . .

- What triggered this consultation?

- Can you talk me through what happened during the consultation? Allow participant to describe the consultation feely.

If not covered above:

- What were you hoping for?

- Why do you think you were referred? Why now? Why not before? (if relevant)

- Did you expect to be referred? Explore what may have created these expectations.

- How did you feel about being referred? Explore.

- Did you feel your views and preferences were taken into account before reaching this outcome?

Outpatient appointment

How many outpatient appointments have you had?

What did you expect from the (first) hospital outpatient appointment? Distinguish between what they expected would happen during appointment, and what outcome they expected. Explore what might have influenced these expectations.

Can you talk me through what actually happened at the hospital outpatient appointment?

Did the hospital doctor discuss with you any tests on your wrist which might help decide which treatment was best?

Did the hospital doctor discuss with you what types of treatment might be best for you? What was discussed? Probe: Surgery? Different types of surgery? Injections? Hand exercises? Other?

What was the outcome? If no surgery, explore if consultant made recommendations. Did patient understand reasons behind decision, and what was going to happen?

Were you happy with this decision? Why/why not?

Did you feel your views and preferences were taken into account?

If multiple outpatient appointments, repeat.

Post-outpatient appointment

Views surrounding access to care

Who do you think was the most influential when it came to deciding whether you would receive surgery? Probe: acceptable? Who should have this responsibility?

If you were to decide who should be offered carpal tunnel surgery, how do you think you would decide?

Generally speaking, do you think there is a need to make decisions in the NHS about who is eligible for treatment? Why? Explore if patient is aware of finite NHS resources.

(If yes) Who do you think should make these decisions? What kinds of criteria should be considered?

Wind down and closure

Researcher thanks informant for their time, asks if there are any additional comments they would like to make, and ensures the conversation has returned to a topic the patient seemed relatively comfortable discussing prior to interview closure.