Infant deaths in the UK community following successful cardiac surgery: building the evidence base for optimal surveillance, a mixed-methods study

Research Ethics Committee

The study was approved by the London Central Research Ethics Committee on 4 October 2012 (reference 12/LO/1398).

National Information Governance Board

The study was approved by the National Information Governance Board now known as the Health Research Authority (HRA) and permission was granted for the use of identifiable data for the purposes of record linkage within specified limits (reference 12/LO/1398, November 2012).

Healthcare Quality Improvement Partnership

The use of national audit data from NCHDA and PICANet, including the use of identifiable data for the purposes of record linkage within specified limits, was approved by Healthcare Quality Improvement Partnership.

National Congenital Heart Diseases Audit

The use of national audit data for the specified purposes stated in the study protocol was approved by the NCHDA and is subject to a data sharing agreement (reference 12/CONG/03).

Protocol and registration

The protocol and search strategy is registered in the International Prospective Register of Systematic Reviews (PROSPERO) as CRD42013003483. 21

Search strategy

To ensure a comprehensive review of the evidence, and to capture any pertinent risk factors that may not yet have been identified in studies with CHD patients, we used a broad search strategy that included other life-threatening congenital malformations requiring major surgery in the first year of life, for example gastroschisis or diaphragmatic hernia. We used key terms relating to children, congenital abnormalities, surgical procedures, hospital discharge and adverse outcomes to electronically search MEDLINE (1980 to 1 February 2013), EMBASE (1980 to 1 February 2013), Cumulative Index to Nursing and Allied Health Literature (CINAHL; 1981 to 1 February 2013), The Cochrane Library (1999 to 1 February 2013), Web of Knowledge (1980 to 1 February 2013) and PsycINFO (1980 to 1 February 2013). Conference abstracts from the Association for European Paediatric Cardiology, the American Heart Association and the European Surveillance of Congenital Anomalies symposia were searched for the period 2008 to 2012. A forward citation search was carried out on the reference lists of all selected studies to identify additional published studies for review.

Selection of studies

Studies were eligible for inclusion only if they separately reported outcomes for children discharged from hospital and in-hospital surgical mortality. To ensure relevance to infant survival, only studies involving children up to the age of 5 years were included in the review and major surgery was defined as requiring intensive or high-dependency care in the postoperative period. Inclusion criteria for the studies included in the review are summarised in Box 1.

The titles and abstracts for all studies were independently reviewed by two reviewers (J Tregay and J Wray) then full-text papers of selected studies assessed by three reviewers (J Tregay, K Brown and R Knowles) to determine whether or not they met the inclusion criteria. Any discrepancies between reviewers were resolved through discussion with a fourth reviewer (J Wray) (see Figure 1).

Data extraction

Data extraction was independently completed by two reviewers (J Tregay and R Knowles) using a standard pro forma that included information on study design, population, diagnosis, comparison groups, outcomes and risk factors.

Quality assessment

Studies were assessed for methodological quality of study design using levels of evidence rated from one (most rigorous, e.g. randomised controlled trial) to four (least rigorous, e.g. retrospective uncontrolled case series). 22 Within each evidence level, studies were assessed as A (high quality) to C (lowest quality), using predetermined criteria such as confounding, completeness of follow-up and objective measurement of outcomes (see Appendix 2).

Data analysis

The outcomes of interest were unexpected death or unplanned readmission to hospital in the first year of life after discharge following cardiac surgery. Factors associated with increased mortality or readmission risk are presented in a narrative synthesis.

Study selection

There were 17 studies identified through systematic searches and a further six studies through forward citations. Of the 23 full-text papers reviewed, eight studies failed to meet inclusion criteria resulting in 15 studies eligible for review (Figure 1). Despite our inclusive search strategy, no studies of post-surgical outcomes for children with non-cardiac congenital anomalies met the inclusion criteria.

FIGURE 1.

The Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) flow chart showing study selection process. Reproduced with permission from Tregay et al. , 2015. 20 © Cambridge University Press 2014.

The review included eight retrospective reviews of surgical cases,23–30 four retrospective cohort studies,31–34 two case–control studies35,36 and one randomised controlled trial that was reported in two papers. 26,37 Only three studies29,36,37 included a prospective element. Although study designs differed, all studies were rated as good quality (Table 1); studies that were assigned a lower rating failed to address some potential confounding factors. Ten reports were of patients with a functional single-ventricle (SV) diagnosis,23,24,26–28,30,33,35–37 which was most often HLHS. Fourteen papers involved patients who underwent cardiac surgery during the first year of life and the remaining study31 included cardiac patients operated up to the age of 18 years with results provided separately for each age group. Only two studies were not conducted in the USA. 23,27 In total, 29,019 patients were followed up for mortality outcomes, of whom 1113 (4%) died. Of the 3672 children who underwent SV surgery, 452 (12%) died, compared with 661 (3%) of 25,347 children who underwent other types of cardiac surgery. Of the 1639 children who were observed in three studies of unplanned readmission, 173 (11%) were readmitted to hospital during the study follow-up period. Table 2 summarises the included papers.

Adverse outcomes

Reported mortality rates varied markedly and were influenced by the study population at risk as well as the duration of follow-up. Five studies23,26,30,33,35 involved children with SV diagnoses undergoing staged palliative surgery and reported ‘interstage’ mortality between first- and second-stage surgery that ranged from 8% to 16%, with follow-up ending at around 1 year after surgery in most studies. Two further studies focused on ‘unexpected’ interstage deaths, defined as acute events or sudden cardiovascular collapse, and reported 4–5% unexpected deaths in neonates discharged home between stages 1 and 2. 28,37 Carlo et al. 24 and Hansen et al. 27 reported interstage mortality of 9% for children discharged home between second (superior cavopulmonary anastomosis or bidirectional Glenn) and third stage (Fontan) surgery. Two studies investigated postdischarge mortality after all types of cardiac surgery: Chang et al. 31 reported a low mortality rate of 0.62% at 1 year after any cardiac surgery undertaken in children up to age 18 years, whereas Pinto et al. 29 found a higher mortality rate of 8% in the 2 years following discharge after neonatal congenital heart surgery. Mortality rates over 10% were reported in studies focusing on specific higher-risk cardiac defect subgroups32 or patients discharged home on mechanical ventilation. 25

Two studies evaluated unplanned hospital readmissions as distinct from mortality29,34 and one further study36 reported unplanned readmissions as part of a combined outcome measure of mortality and readmission. Readmission rates within 30 days of hospital discharge ranged from 10%34 to 30%36 and at 2 years post discharge were 45%;29 variations were influenced by duration of follow-up and differences in data collection methods, which included hospital records review34,36 and telephone survey. 29

Risk factors associated with adverse outcomes

Although many different factors were investigated, the findings relating to individual factors were inconsistent; this may reflect the heterogeneity of participant characteristics and study designs. Figure 2 summarises the factors investigated by different studies and indicates whether or not these were found to increase mortality risk.

FIGURE 2.

Summary of factors examined in the 15 studies. AV, atrioventricular; ICU, intensive care unit; SES, socioeconomic status; RACHS category, Risk Adjustment in Congenital Heart Surgery category. a, The cardiac surgical procedures defined in these studies were palliative procedures, Norwood stage 1, bidirectional Glenn, shunt construction, thoracic vessel procedures, truncus arteriosus repair, total anomalous pulmonary venous return repair and open valvotomy. Reproduced with permission from Tregay et al. , 2015. 20 © Cambridge University Press 2014.

Context and limitations

Our review confirms the significant lack of research into adverse outcomes after hospital discharge following surgery and highlights the fact that the evidence base to inform postdischarge clinical care and identify infants at high risk for focused support is extremely limited. As many reports derive from North American studies, and the research population is often limited to infants who have severe and complex cardiac diagnoses requiring staged surgery, care must be taken in generalising the findings from these existing studies to the wider UK population of infants with CHDs.

Despite our broad search strategy, which was intended to capture research into other life-threatening anomalies that require surgery during infancy, such as gastroschisis and diaphragmatic hernia,39 the only studies of postdischarge outcomes that met the inclusion criteria concerned CHDs. Of all congenital anomaly subgroups, the highest rate of infant deaths is associated with CHDs, and this may account for the greater interest in monitoring outcomes after hospital discharge. It is also notable that postdischarge outcomes of infants with CHDs came to prominence with the introduction of staged palliative surgery for HLHS, which led to improved early in-hospital outcomes40,41 and highlighted later interstage mortality as an important concern. 15,26

A limitation of our review was the rigour of our eligibility criteria, which excluded any studies that did not clearly differentiate between deaths that occurred before and after hospital discharge, and thus may have excluded from the review some studies that evaluated additional risk factors to those reported here. The relative lack of studies reporting postdischarge surgical outcomes may also simply reflect the limited monitoring of late adverse events and, specifically, of events occurring in the community or primary care setting. 31

Ethnicity and deprivation

In three US studies26,32,34 included in our review, patients of Hispanic ethnicity were found to be at a greater risk of adverse outcomes relative to white patients post hospital discharge. This confirms previous research which has shown that US Hispanic communities are more likely to experience multiple barriers to health care including language and immigration status, and financial barriers, such as lack of health insurance or low family incomes. 42,43 However, the impact of ethnicity and socioeconomic deprivation demonstrated in US-based studies may be influenced by an individual family’s ability to pay for care42–49 and the relevance of these findings for the UK health-care system is uncertain. Nevertheless, there is evidence that lower-income families in the UK also experience a considerable financial burden when caring for their child with CHD and that this may affect care-seeking behaviours. 49 The results of our review therefore add to the growing body of evidence suggesting that patients from minority ethnic and lower socioeconomic groups are more likely to experience barriers to timely and appropriate access to care and underlines the relevance of these factors to the population of infants with CHDs following hospital discharge.

Medical factors

Postoperative feeding and growth were also significantly associated with adverse outcome in several studies33,34,36 identified within the review, consistent with previous research. 50 However, the relationship between feeding difficulties and adverse outcomes post discharge is likely to be complex because of potential confounding with poorer cardiac status and other comorbidities; therefore, it requires further investigation.

Readmission to hospital will depend on both the child’s clinical state and the response to this by parents and medical staff. It is possible that readmission signified a timely response to a child’s deteriorating clinical state in some cases, whereas in others it was a response to a child who became seriously unwell. In our review, we considered an unplanned readmission to be an adverse event indicating that a child deteriorated unexpectedly at home and so did not experience a stable clinical course after discharge. Nevertheless, it is possible that the risk factors for readmission may differ from those for deaths and this may have contributed to the breadth of different risk factors identified.

Conclusion

We identified several key medical and social factors associated with a higher risk of mortality or unplanned hospital readmission for children discharged from hospital after paediatric cardiac surgery. Some of these risk factors, such as feeding difficulties, would be amenable to modification through specific interventions, whereas others enable health professionals (HPs) to identify children who are at greatest risk of adverse outcomes and to offer additional support, such as HMPs, targeted more effectively at vulnerable children and their families within the community setting. Although there were no studies of social and financial factors within the UK health-care context, unequal access to care may disproportionately affect minority ethnic communities and low-income families and should be a focus for future research in the UK. Crucially this review highlights an evidence gap and important need for longer-term studies to investigate the risk factors for out-of-hospital outcomes after surgery separately from in-hospital outcomes. Such evidence would better inform postdischarge care and community-based interventions to improve long-term survival and quality of life of infants with CHDs.

Protocol and registration

The protocol for this review is registered with the International Prospective Register of Systematic Reviews (PROSPERO) CRD42013003484. 51

Search strategy

Selection of studies

Data extraction

Data extraction was independently completed by two reviewers (J Tregay and K Brown) and included information on study design, population, diagnosis, comparison groups, outcomes and the intervention characteristics.

Quality assessment

Two reviewers (J Tregay and K Brown) assessed the methodological quality of studies included for review. Studies were assessed on the basis of study design using levels of evidence (LOEs) rated from 1 to 4; with more rigorous study designs, for example randomised control trials, given a rating of 1 and the least rigorous, such as case series, given a rating of 4. Studies within each LOE were assessed as A (high quality) to C (lowest quality), dependent upon predetermined criteria such as confounding, completeness or follow-up and objective measurement of outcomes (see Appendix 2).

Data analysis

Data were not combined in a meta-analysis and are presented below in a qualitative synthesis. This approach to analysis was selected given the small number of studies found and their methodological quality. Studies were qualitatively reviewed by two researchers (K Brown and J Tregay) attempting to answer the following questions:

1. What were the constituents of the interventions described in terms of inclusion criteria and protocols deployed?

2. Was there evidence in respect of treatment effect: what outcome measures have been used and were there any important biases?

3. Do any important limitations, concerns or constraints exist when considering wider application of the intervention described?

Presentation of results

Studies included in the review are summarised in Tables 3–6. Table 3 contains information about study quality; Table 4 provides a comparison between monitored patients and historical control patients in each of the studies; and Tables 5 and 6 summarise the protocols for the interventions concerned and their outcomes. Studies conducted by the same research group using the same patients are combined together for presentation in the tables.

Studies selected

Title and abstract screening identified 20 studies for potential inclusion in the review. A further four studies were identified through forward citations. Following full-text screening of these 24 studies, 16 studies did not meet eligibility criteria and were excluded, leaving a total of eight studies for inclusion in the review (Figure 3).

FIGURE 3.

Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) flow chart showing study selection process.

Methodological quality

Eligible studies were all prospective cohort studies using retrospective historic controls and were of fair to good quality (rated A or B; see Table 3). Typically, controls were drawn from the time period immediately preceding introduction of the intervention.

Study populations

Despite a broad search strategy designed to encompass a range of congenital conditions requiring surgery, all studies (eight studies from six research groups) that were eligible for inclusion related to patients with CHD, specifically SV diagnoses, such as HLHS. HLHS is complex congenital heart defect necessitating a series of palliative surgical procedures to sustain life, including two in the first year. Infants undergoing these palliative procedures remain fragile, with mortality between their first and second operations a recognised complication for a range of reasons, many of which relate to the underlying heart disease30,35 becoming more stable after the second-stage operation.

The terminology used to describe diagnoses and operative procedures differed slightly between studies: this information is summarised in Table 4, which indicates that four of the research groups studied infants with diagnoses of predominantly HLHS and other related lesions that went down the pathway of a Norwood type stage 1 operation with either MBTS or RVPA15,53–55,58 or a hybrid procedure56 (the hybrid approach is a more recently developed treatment pathway applicable to HLHS and related conditions, which incorporates a series of palliative procedures involving both surgery and interventional cardiology or catheterisation), and two of the research groups studied both infants with HLHS type diagnoses who underwent a Norwood type stage 1 operation and infants with other SV diagnoses who underwent an appropriate operation based on diagnosis including isolated MBTS, Damus–Kaye–Stansel repair and pulmonary artery banding. 52,57 Five of six research groups were from the USA and one was from Germany.

Intervention types

All studies meeting criteria for inclusion involved the evaluation of HMPs, which represent detailed protocols for management from hospital discharge (details in Home monitoring programmes) aiming to enhance patient supervision and respond early to any deterioration. No studies involving any other intervention types focusing on the post-hospital discharge period were identified as meeting inclusion criteria; however, in one study58 home monitoring was only one component of a broader protocol-based intervention that began in hospital prior to surgery. In this study, Srinivasan et al. 58 detail a standardised management protocol for the treatment and follow-up of infants undergoing the Norwood procedure, a complex surgical intervention for patients with SV congenital heart diagnoses incorporating all aspects of in-hospital and early post-hospital management including a HMP.

Outcomes

All studies reported a degree of positive effect linked to HMP:

• Three research groups reported reduced interstage mortality rates in monitored patients versus control patients. 53,55,56 Two studies reported improved initial survival after stage 1 surgery, a non-significant difference (trend to improvement with HMP) in interstage outcome and improved survival at 1 year. 57,58 The latest study by Ghanayem et al. 54 reported improved actuarial survival with HMP in addition to reduced interstage mortality. Interestingly, the only study including patients that underwent a hybrid procedure reports the highest interstage mortality and the highest rate of breaches in both HMP and control groups (non-significant trend towards improvement with HMP). 56

• The three research groups which included data on age at stage 2 operations reported that stage 2 surgery was performed at a younger age in monitored patients than in control patients;15,53–55,57 two of the relevant research groups noted that the earlier, and in their view more optimal, patient-specific timing of stage 2 operations was linked to a reduction in interstage mortality. 53,55

• Three studies reported that monitored patients showed improvements in weight gain relative to control patients. 55–57

• Three studies reported the statistics for breaches of the HMP: these occurred in 31%,55 57%15,53,54 and 62.5%56 of patients (most commonly desaturation) with interventions being enabled for the patients concerned potentially in a more timely manner. One study noted that HMP enabled the timely detection of residual lesions, which could then be treated. 52 The study reporting interstage emergency admissions in patients from the control groups (notably the study on hybrid patients) stated that the number of readmissions was similar in the HMP and control groups;56 however, comparisons between the HMP and the control groups are difficult in this regard because control participants did not have breach criteria.

Outcomes with home monitoring programme

All studies reported benefit in terms of outcome; however, comparisons between patient groups were based on historic control patients in an era during which outcomes for children with CHD have improved. 3,60 Although it seems likely that interventions such as HMP have contributed to improved long-term outcomes over the era in which these have been introduced, the lack of randomised studies means that this cannot be confirmed.

The mortality rate at 1 year is an outcome measure that offsets the effect of earlier stage 2 surgeries in the context of HMP, and hence a shortened period of interstage follow-up, which is a bias to consider when the reduced interstage mortality rate is presented as a primary outcome measure. Three of the research groups presented evidence for improved longer-term survival (to stage 2 in the form of competing hazards over time,58 to 1 year57 and actuarial survival54) in the context of HMP; however, two of the research groups attributed these improvements to the whole patient pathway, in particular immediate post-stage 1 outcomes. These two studies57,58 found non-significant reductions in interstage mortality with HMP; given that these are single-centre studies involving a maximum of 204 patients, and that mortality in the control group was 12% in each case, studies may be underpowered unless there is very large drop in interstage mortality in the HMP group.

Specific patient types

Practical and resource implications

The included studies indicated that HMPs involve the commitment of considerable dedicated resources; for example, one study reported that a multidisciplinary team comprising six paediatric cardiologists as well as several other professionals met weekly to discuss patients. 57 In addition, studies noted that the burden in terms of clinical visits and other patient contacts was increased with HMP. However, conversely, information about the health-care costs of infants who deteriorate in the absence of surveillance, and for example who require intensive interstage care, is lacking. Interestingly, the use of gastrostomy to provide feeding assistance in SV infants and infants with HLHS appears to be higher in the USA than in Europe (26% of infants in the SVR trial were fed by gastrostomy tube between stages);26 in Europe infants may be kept in hospital as inpatients because of feeding problems55 or, alternatively, given nasogastric feeding at home but are very rarely fitted with a gastrostomy tube (personal communication from Dr R Franklin, Royal Brompton Hospital, and L Smith, Great Ormond Street Hospital, both in London, UK, 2013). These alternative approaches may have differing resource implications, and a more detailed evaluation of this aspect is required in order to understand it better.

Studies of interest that did not meet inclusion criteria

Studies that were detected by the search strategy and reviewed as part of the study, but did not meet inclusion criteria included the following.

Conclusion

This review set out to describe and evaluate postdischarge interventions for infants who have undergone major congenital surgery. Surprisingly, only studies involving HMP for infants with CHD (HLHS or SV) met the criteria for inclusion in the review. The prominence of CHD in this review may be a result of the relatively high risk of postdischarge morbidity and mortality in these patients, particularly those undergoing staged palliative procedures. The evidence base for postdischarge interventions in this population would be strengthened by further prospective studies, such as randomised controlled trials, in patients with congenital heart disease and those with other complex congenital conditions; studies from research groups outside the USA; reporting of longer-term outcomes, to at least a year; and more detailed reporting of the content of intervention programmes and of demographic information such as ethnicity and socioeconomic status, which are known to be associated with adverse outcomes in paediatric patients with cardiac disease.

Constructing the data set for analyses

Post-procedural variables

The following post-procedural patient variables known at the point of discharge were considered: LOS for the index admission period; whether or not any surgical or catheter procedures were performed during the index admission in addition to the index procedure; whether or not renal support was required during the index admission (including dialysis and haemofiltration); whether or not extracorporeal membrane oxygenation (ECMO) support was required during the index admission; whether or not the index admission was associated with any intensive care unit events (assigned as such if any admission to PICU during the index admission period contained a Read Code category for collapse or cardiac arrest, acquired injury or complications or a non-cardiac operation in PICU); whether or not the index admission was associated with any acquired comorbidities [assigned if any admission to PICU during the index admission period contained a Read Code acquired category (1–8 in Appendix 7)]; and whether or not the index admission was associated with post-procedural morbidity (assigned if any NCHDA record during the index admission period contained an IPCCC code corresponding to a post-procedural morbidity).

Missing data

When ethnic group was not available from PICANet, the NCHDA ethnic code was used to assign white, Asian or black ethnicity (which showed strong concordance across the two audits) but not to assign Chinese, other or mixed ethnicity (which showed poorer concordance). Sensitivity analyses were performed excluding records without PICANet-derived ethnicity. Index procedure weights recorded as zero (i.e. missing) were replaced by weight recorded within the same index admission (if available). Weight-for-age z-scores outside the range ± 5 z-scores were assumed erroneous and treated as missing.

Outcomes

Two nested outcomes of interest were defined.

• Outcome 1: death within 1 year following discharge from the index admission and not during a planned admission.

• Outcome 2: either death (outside a planned admission) or an emergency unplanned readmission to PICU within 1 year following discharge from the index admission. Outcome 2 was viewed as including ‘near misses’ for death, given that all patients experiencing emergency intensive care unit admission are gravely ill. A decision was made to combine deaths and near misses for this section of the analyses in order to increase the number of events, and to form a useful basis for classification and regression tree (CART) analysis was used to identify patient groups with different ‘profiles of risk’ that could usefully inform the development and prioritisation of interventions aimed at improving outcomes within this patient population.

In order to ascertain these outcomes the following information was used: age at death (if applicable) and life status, which were available in NCHDA; emergency unplanned admissions to PICU were extracted from PICANet (defined as admission type ‘unplanned’ or ‘unplanned after surgery’ AND identified as a retrieval from another unit as an emergency transfer AND not defined as an elective admission). Note that death within 1 year of discharge from index admission that occurred during a planned readmission to intensive care was not considered an adverse outcome in this analysis designed to inform improvements in postdischarge services.

Statistical methods

Data set and headline outcomes

Approximately 20% of records in NCHDA did not have a valid NHS number (the unique patient identifier used in this study to match across the two audits); many of these records pertain to overseas patients as only patients resident in England or Wales are allocated an NHS number. A total of 12,390 infants with a valid patient identifier and meeting the inclusion criteria for the study were identified in NCHDA, of whom 9385 (76%) were linked to at least one record in PICANet. Failure to link a record from NCHDA with one in PICANet could be accounted for by a procedure (in general a catheter) which occurred without an intensive care admission or the lack of an NHS number for a given patient in PICANet. A total of 115 children who underwent only an excluded catheter procedure were excluded from the analysis, as were 765 premature babies who underwent ligation of patent ductus arteriosus (PDA) only and 24 transplant patients. A further 505 patients with unknown life status were also removed.

Of the remaining 7976 patients, 333 [4.2%, 95% confidence interval (CI) 3.7 to 4.6] died during their index admission period and were excluded from our analyses, leaving a final data set comprising 7643 infants discharged alive from their index admission. Of these, 246 (3.2%, 95% CI 2.8 to 3.6) died within 1 year following discharge from the index admission and not during a planned admission (outcome 1). A total of 514 children (6.7%, 95% CI 6.2 to 7.3) either died or were admitted unplanned, as an emergency, to a PICU within 1 year following discharge from the index admission (outcome 2). Finally, 115 children (1.5%, 95% CI 1.2% to 1.8%) died during a planned admission to PICU within 1 year following discharge from the index admission (which was not considered an outcome in this analysis), giving an overall mortality within the year following discharge from index admission of 4.7% (95% CI 4.2% to 5.2%).

Ethnic group was not available from PICANet for 1703 children in the final data set. Of these, the NCHDA ethnic code was used to assign white (n = 1001), Asian (n = 243), black (n = 113) or missing (n = 346) ethnicity. Excluding records without PICANet-derived ethnicity from the analyses did not significantly affect the results. In a total of 528 children weight-for-age was missing or anomalous (assumed erroneous and treated as missing). The variable for which the level of missing data was highest, markedly so, was prematurity [n = 2101 (27.5%)]; sensitivity analysis comparing models with and without prematurity showed a marginally better fit if it was included (based on Akaike criteria) and very little difference in the odds ratio coefficients for all other factors.

Descriptive and univariate analyses

The following variables showed no univariate association with either outcome and so were not considered in further analyses: sex, and whether or not any catheter procedures were performed during the index admission in addition to the index procedure. Both outcome 1 and outcome 2 were significantly associated with all of the other candidate variables in univariate analysis (p-value < 0.05). Table 8 shows the observed numbers of patients and rate of outcomes 1 and 2 in the data set for those parameters significant in univariate analysis (and weight-for-age), along with the results of the univariate analysis.

The relationship between LOS and both outcomes was non-linear and the fractional polynomial transformation was used in the subsequent risk model development analyses; for age we used a log transformation. The relationship between weight-for-age and each outcome did not depart significantly from linearity.

Developing risk models for outcomes 1 and 2

In the multivariate analysis in which continuous variables were treated as such, the significant risk factors for both outcomes were age at procedure (continuous), weight-for-age z-score (continuous), index procedure group, cardiac diagnosis group, non-cardiac congenital anomaly, prematurity, ethnicity and LOS in a specialist centre (continuous). Preprocedure clinical deterioration was additionally significant to outcome 1, whereas neurodevelopmental condition and acquired cardiac diagnoses were additionally significant to outcome 2.

In the final model development, in which ease of interpretation was considered as well as statistical performance, the continuous predictors (age at procedure, weight-for-age z-score and LOS) were categorised as shown in Table 9 (along with observed numbers of patients and rate of outcomes 1 and 2 in the data set). The final risk model for outcome 1 comprises age at procedure (categorical), weight-for-age z-score (categorical), index procedure group, cardiac diagnosis group, non-cardiac congenital anomaly, prematurity, ethnicity, LOS in specialist centre (categorical) and clinical deterioration. The final risk model for outcome 2 comprises age at procedure (categorical), weight-for-age z-score (categorical), index procedure group, cardiac diagnosis group, non-cardiac congenital anomaly, prematurity, ethnicity, LOS in specialist centre (categorical), neurodevelopmental condition and acquired cardiac diagnoses. Details of the regression models (odds ratio, standard errors and 95% CIs) are shown in Tables 10 and 11.

For each patient variable the multivariate odds ratios, standard errors and 95% CIs are presented and the reference category indicated.

We note that, although variable selection was based on > 50% inclusion frequency over the imputed data sets, in practice there was clear discrimination between those factors that were included in the model (≥ 90%) and those that were not included (< 40%). For both outcomes, sensitivity analysis of the models derived on complete case data showed a marginally better fit to the data when prematurity was included, based on Akaike criteria. However, for models with and without prematurity the odds ratios for all other factors remained very similar, indicating robustness of model variable selection. Furthermore, for all models, adjusting for clustering at either the hospital level or regional (primary care trust) level had no significant impact on results so we have presented the unadjusted results.

Risk model performance

The final (categorical) model for outcome 1 gave a combined c-index of 0.78 (95% CI 0.75 to 0.82), indicating good discrimination. This was only marginally less discriminative than the continuous model (c-index 0.80). The final (categorical) model for outcome 2 also showed good discrimination with a combined c-index of 0.78 (95% CI 0.75 to 0.80), compared with a c-index of 0.78 for the continuous model. Calibration of the final categorical and continuous models for both outcomes was also good, with Hosmer–Lemeshow p-values ranging from 0.10 to 0.75 across the models fitted on the 20 imputed data sets, indicating no statistically significant differences between observed and expected number of deaths when calculated in deciles of predicted risk for each of the imputed data sets.

Identifying patient groups differentiated by risk of outcome 2

Figure 4 depicts the final tree generated by the CART analysis along with, for each node, the test set figures for the number of patients and rate of outcome 2. The rate of outcome 2 evaluated across the entire data set is shown below the box for each final patient group. The rate of outcome 2 across the entire analysis data set (total number of patients 7643) was 6.7%. The rate of outcome 2 within a 60% development subset was 7.0%, compared with 6.3% in the 40% test set.

FIGURE 4.

Patient groups derived from classification and regression tree analysis. The final tree generated by the CART analysis along with, for each node, the test set figures for the number of patients and rate of outcome 2. The rate of outcome 2 evaluated across the entire data set is shown below the box for each final patient group. Reproduced from Crowe et al. , 2016. 67 © 2016 The Authors. Published on behalf of the American Heart Association, Inc., by Wiley Blackwell. This is an open access article under the terms of the Creative Commons Attribution-Non Commercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes

Of the 18 variables entered in the analysis, CART identified presence/absence of a neurodevelopmental condition as the best single discriminator between patients experiencing outcome 2 or not. For those without a neurodevelopmental condition, the next best discriminator is whether the cardiac diagnosis is high risk (HLHS, UVH or PA) or low risk (VSD or ‘other’). Of the latter, the next best discriminator is presence/absence of a congenital anomaly followed by the LOS in the specialist hospital (threshold 1 month). This branching creates six discrete patient groups, details of which are set out in Table 12.

Strengths and weaknesses

The national audit data underpinning this study offer a unique opportunity for a population-based analysis, and the study was designed, conducted and reported based on the Strobe Statement for observational studies. 90 First, the data are of high quality (as demonstrated by the results of a regular systematic independent validation process). Second, life-status tracking based on NHS number enables late deaths outside treatment centres to be reliably ascertained. Third, the mandatory and universal nature of the data submission means that all relevant cases are captured. Our findings may therefore be considered more generalisable than those based on single-centre studies or those from a more limited geographical area. That said, our study outcomes inevitably reflect recent historic services provision by the NHS in addition to intrinsic medical risk; hence, findings may differ to some extent between geographical regions. The study also reflects the demographics of the UK in terms of ethnicity, which may in turn have implications for the distribution of CHD types and other congenital anomalies. 91

A further positive feature of the study is that the combined data set used is more valuable for the purposes of this study than either of the individual audit data sets in isolation, since each contains different types of information concerning potential risk and, when variables in the two data extracts overlapped, the item with the highest data completeness and discrimination was selected. The strength of the NCHDA data set includes the detailed cardiac-specific information as determined by IPCCC codes related to CHD and the cardiac interventions undertaken. The strengths of the PICANet data set include the rich medical coding and information pertaining to the type of admission (such as where an emergency occurred), as well as intensive-care supports and interventions.

We note, however, that our results reflect only those elements that are captured within the national audit data and so certain known risk factors, for example postprocedural weight gain33 and selected abnormalities based on echocardiography,36 could not be incorporated. Furthermore, there may be other as yet unknown risk factors that are not captured in the registry data.

It was necessary to group cardiac diagnoses and procedures in order to render them tractable for statistical analysis (to reduce the degrees of freedom in the data set). Although this was performed with a strong clinical sense of what was pertinent, based on literature and clinical experience, we note that such groupings are to an extent subjective and, as in other registry-based studies, there remained a subset of ungrouped procedures. The choice of diagnosis group ‘HLHS’ reflects the body of literature related to ‘interstage’ deaths in this population. A separate group of ‘UVH or PA + IVS’ was chosen because of the high-risk nature of these conditions, although with a view to distinguishing these from HLHS patients. Of the remaining cardiac diagnosis groups, we elected to review the VSD group separately in order for us to evaluate the face validity of comparisons between HLHS or UVH/PA + IVS and the much larger, less homogeneous group of mainly biventricular diagnoses. We note that results for the VSD group and the diverse ‘other’ group were comparable.

We note that the ethnicity group classified as ‘other’ was associated with higher risk: this is discussed in more detail in Chapter 5. The audit data will not have captured the scenario in which babies were discharged to home on a palliative care pathway, and hence our outcomes may have incorporated a subset of babies that were expected to die. However, local audit of cases at two English tertiary centres suggests this occurs in only a minority of cases. 6 Finally, given its focus on the population of infants who have undergone an intervention, the analysis does not include infants awaiting intervention who are also known to be potentially vulnerable.

Concordance of the ethnic classifications used in the National Congenital Heart Diseases Audit and Paediatric Intensive Care Audit Network data sets

During the development of the risk model data set, we explored the concordance of ethnic classification within the NCHDA and PICANet audit databases using a similar methodology to Saunders et al. 107 This determined the ethnic classification subsequently used in the risk model and CART analysis.

The NCHDA database uses a bespoke ethnic classification comprising five ethnic groups: Caucasian, Asian, black, Oriental and ‘other’ (‘other’ in NCHDA also includes children of mixed ethnicity; therefore, we refer to it below as ‘other – mixed’). Ethnicity is recorded in the PICANet database using the detailed (16 category) UK census classification (Table 13), which is often aggregated into six large groups. As ethnicity was recorded for each admission, the recorded ethnic group varied between admissions for some children; in these cases, the most frequently reported ethnic group (mode) for each child was selected as the child’s ethnic group.

As the PICANet and NCHDA ethnicity classifications were not directly comparable, we estimated Cohen’s kappa statistic as a measure of agreement. Furthermore, taking the PICANet ethnic category as the reference standard because of its more widespread use and comparability with national population statistics, we calculated the proportion of patients whose NCHDA ethnic group was concordant with the PICANet-recorded ethnicity (sensitivity) and the probability that a patient with a given NCHDA ethnic group had the corresponding PICANet ethnic group (positive predictive value).

Ethnicity was not recorded for 1713 (22.8%) children in the PICANet data set (see Table 13); however, ethnic group was available from NCHDA for 1438 of these, of whom the majority were Caucasian (n = 1002; Table 14). Using a data set including only children who had a record in both NCHDA and PICANet (n = 5575), we explored the concordance between the two classifications to determine whether or not the NCHDA ethnic group could inform the missing PICANet record. The PICANet categories white, black, Asian, Chinese, other and mixed were matched against the NCHDA categories of Caucasian, black, Asian, Oriental and ‘other – mixed’, respectively. Cohen’s kappa statistic for agreement of PICANet and NCHDA recorded ethnicity was 0.83 overall.

The sensitivity and positive predictive value of each NCHDA ethnic category with respect to the corresponding PICANet category, was evaluated using the pairings white/Caucasian, black/black, Asian/Asian, Chinese/Oriental, other/‘other – mixed’ (Figure 5).

FIGURE 5.

Concordance between NCHDA ethnic group and corresponding PICANet ethnic group. NICOR, National Institute for Cardiovascular Outcomes Research; PPV, positive predictive value.

Around half of the children in the NCHDA Oriental category were classified as Asian (Pakistani, Bangladeshi, Indian or other Asian) in PICANet (see Table 14). The PICANet ‘mixed’ category does not exist in NCHDA; 58 (31.0%) of these children were classified as white in NCHDA and the remainder were distributed equally between the black, Asian and other categories.

The Caucasian, black and Asian categories in NCHDA demonstrated good concordance with the PICANet ethnic groups of white, black and Asian, respectively (see Figure 5), and risk of misclassification was low (positive predictive value ≥ 79.7%). The Chinese/Oriental and other/‘other – mixed’ categories were poorly concordant and the NCHDA category predicted the corresponding PICANet group in < 68% of cases. To facilitate comparison with population denominator data and reduce missing data, children with ethnicity missing in PICANet but recorded in NCHDA as Caucasian, black or Asian were assigned to the ethnic groups of white, black or Asian, respectively, whereas children in the Oriental and ‘other – mixed’ NCHDA categories who had no PICANet ethnicity record remained unchanged in PICANet. Using this method to map children to a PICANet ethnic group, we created a data set in which only 4.6% (348 out of 7529) of children had missing data for ethnic group; further analyses described here, including the risk model, CART analysis and analyses of ethnic variation, use this data set. Of 348 children with missing data, 20 were classified as Oriental and 53 as ‘other – mixed’ in NCHDA, while the remainder had no ethnicity record in either data set.

Incidence of congenital heart disease types by ethnicity

The incidence of CHD surgery in the first year of life in England and Wales from 2006 to 2009 was 2.0 (95% CI 1.9 to 2.0) per 1000 children (n = 5350; Table 15). Although this incidence rate approximates birth prevalence, actual birth prevalence would be slightly higher, as some children will survive the first year of life without an intervention, will remain undiagnosed, or will die without an intervention.

Compared with the white ethnic group, the relative rate of CHD was higher in children of Asian, black and ‘other’ ethnicity and lower in children of Chinese or mixed ethnicity (Table 16). In sensitivity analyses including only children with ethnicity recorded in PICANet (n = 4127), these relationships did not change; however, the CIs around the relative rate for the black and Chinese groups included 1.

The incidence of specific CHD subtypes varies by ethnic group (Figure 6). Compared with the white reference group, the following subtypes were more common: in the Asian ethnic group the relative rate of UVH, TGA + IVS, PA, tetralogy of Fallot, TAPVC, VSD, atrial septal defect (ASD) and PDA; in the black ethnic group HLHS, UVH, AVSD and VSD; and, in the ‘other’ ethnic group, HLHS, truncus arteriosus, TGA, PA + VSD, AVSD, tetralogy of Fallot and VSD (Table 17). Although specific defects appeared to be under-represented in non-white ethnic groups compared with the white ethnic group, as a result of small numbers there was insufficient precision to identify any significant differences.

FIGURE 6.

Distribution of CHD subgroups by ethnicity for children operated between 2006 and 2009. (a) white; (b) Asian; (c) black; (d) Chinese; (e) mixed; and (f) other. AAO, aortic arch obstruction; AS, aortic valve stenosis; ASD, atrial septal defect; AR, aortic regurgitation; IAA, interrupted aortic arch; misc. C, miscellaneous congenital CHD; misc. PC, miscellaneous primary CHDs; MV, mitral valve abnormalities; PS, pulmonary stenosis; TOF, tetralogy of Fallot; Tr, truncus arteriosus; TR, tricuspid valve insufficiency; SaS, subaortic stenosis.

Association between ethnicity and individual characteristics of children operated on (2005 to 2010)

Descriptive statistics are presented in Table 18. Although characteristics of children with no record of ethnicity in either data set (n = 348; ‘ethnicity not stated’) were explored, it is difficult to draw inferences about this group because of its potential heterogeneity and a lack of information about why ethnicity was not recorded. However, it is relevant to note that a higher percentage of children without a record of ethnicity appear to have died during the index hospital admission; therefore, information may be lacking because of an early death or this may represent a particularly vulnerable group.

There were more boys than girls in the data set overall [n = 4189; 55.6% (95% CI 54.5% to 56.8%)] and within each ethnic group except the black and mixed ethnic groups, which each had a preponderance of girls [black ethnic group: boys 48.3% (95% CI 42.8% to 53.9%); mixed ethnic group: boys 48.7% (95% CI 41.6% to 55.9%)].

Children with CHD admitted to the PICU were more likely to be resident in the most deprived areas if they were from non-white ethnic groups than from the white ethnic group (Figure 7). Analysis of distribution across the quintiles by detailed 16-category ethnic groups demonstrated that there was variation within these broad ethnic groups, such that, within the Asian group the proportion of children of Asian-Pakistani and Asian-Bangladeshi ethnicity living in the most deprived areas was higher than the proportion of Asian-Indian ethnicity, more white-British children than white-Irish were found in the most deprived quintile, and children of black-African ethnicity were more likely than those of black-Caribbean ethnicity to be found living in deprived areas.

FIGURE 7.

Distribution of cases by ethnic group and deprivation quintile.

Compared with the white ethnic group, the proportion of children recorded as having neurodevelopmental abnormalities during the index admission was significantly higher within the Asian, black and mixed ethnic groups (white 3.5%, Asian 6.4%, black 5.4%, mixed 6.1%; see Table 18).

Variation by ethnic group in the proportion of children with associated non-cardiac congenital anomalies and acquired comorbid conditions was not significant (see Table 18). The proportion with congenital anomalies ranged from 20.6% to 38.3% between ethnic groups, and we do not currently have a breakdown by type of congenital anomaly between ethnic groups (the range of possible anomalies was very wide).

The proportion of infants born preterm (before 37 completed weeks’ gestation) was 10.2% (n = 766), which is higher than in the general population, but no significant variation by ethnic group was observed (see Table 18). Overall, 2253 children (29.9%) were diagnosed antenatally with CHD and 1504 children (20.0%) showed evidence of clinical deterioration prior to their index admission; no significant differences were found by ethnic group.

Age at index procedure was categorised into four groups: under 10 days, 10–30 days, 31 days to 3 months, and over 3 months. Most children were operated aged either > 3 months or < 10 days and this distribution was similar for all ethnic groups (see Table 18).

Infants were also categorised by their weight z-score at the time of their index procedure: more than –2 SDs of the mean weight for their age (normal), between –2 SDs and –4 SDs of the mean (low) and more than –4 SDs below the mean (very low). Within all ethnic groups, most children (49.7% to 59.7%) were within the normal weight-for-age category (see Table 18).

Ethnic variations in mortality and unplanned readmission outcomes

There were 318 deaths during the index hospital admission; 232 deaths occurred within the first year after discharge home from hospital following the index procedure and were associated with an unexpected collapse in the community or an unplanned readmission to hospital (outcome 1 in the risk model; see Table 18). A further 252 children who went home after the index admission had an unplanned emergency readmission but did not die within 1 year of discharge; therefore, a total of 484 children experienced outcome 2 (see Table 18).

In univariate analyses (Table 19), children of Asian ethnicity had a higher mortality risk [RR 1.6 (95% CI 1.2 to 2.2)] than the white reference group during the period of the index hospital admission. Following discharge, the ethnic group that experienced a higher rate of unexpected deaths (outcome 1) than the white ethnic group were children of ‘other’ ethnicity [RR 3.4, (95% CI 1.9 to 5.9)]. In the first year following discharge from hospital, children of Asian, black and other ethnicity had a significantly higher univariate risk of outcome 2 [Asian RR 1.3 (95% CI 1.0 to 1.7), black RR 1.6 (95% CI 1.1 to 2.3), other RR 2.1 (95% CI 1.3 to 3.4)] than the white reference group.

Data in context

In the UK, infant mortality (death in the first year after birth) is highest in the black Caribbean, Bangladeshi and Pakistani ethnic groups (at over 6 deaths per 1000 live births); this is twice as high as in the white ethnic groups. In 2012 UK data from the Office for National Statistics, the most common cause of death reported for most ethnic groups (black 54%, mixed/Chinese/any other ethnic group 44% and white 43%) was ‘immaturity-related conditions’, whereas for Asian infants the most common cause of death was congenital anomalies (41%). 108

Analysis of UK infant mortality statistics2 has also shown ethnic variation in mortality risk for babies born with congenital anomalies. Compared with affected babies born to mothers of white British ethnicity, there is evidence for a fourfold increased risk of death in the first year for infants born to mothers of Pakistani ethnic origin and a 45% higher infant death risk for babies born to mothers of ‘all other’ ethnicity (a group that includes other, Chinese and mixed ethnicity in our analysis). Some authors have suggested that excess infant mortality in the Pakistani population may be because of consanguinity91 and, based on a meta-analysis across different populations, Bittles109 suggests that children of closely related parents have a 3.3% median excess risk of congenital anomalies compared with children of unrelated parents. Nevertheless, Bittles also emphasises noticeable variations in the baseline prevalence of congenital anomalies between different populations.

It is possible that our risk model did not demonstrate a higher infant mortality or emergency admission rates associated with Asian ethnicity, as has been demonstrated in some previous studies,19,110 because the higher death rate among Asian babies during the index hospital admission meant that many infants did not survive to hospital discharge. These early in-hospital deaths are likely to have disproportionately affected babies with more severe cardiac defects or associated comorbid conditions, such as more severe types of non-cardiac congenital anomalies or prematurity. This aspect has not to date been explored in detail and goes beyond the scope of our research project.

Different mortality rates after birth will also be influenced by the likelihood of prenatal detection for different defects, and views about the acceptability of pregnancy termination will determine whether or not the diagnosis of a severely affected infant leads to a live birth. In 1999, Bull reported that a serious CHD was diagnosed in 23% of all pregnancies, but only 12% of live births were affected by a CHD; pregnancies affected by CHDs were terminated in around 50% of cases, and this varied by gestation at diagnosis but not by region of birth. 111 Although analyses of termination rates for pregnancies affected by CHDs by ethnic group have not been reported in the UK, a recent analysis of pregnancy terminations for neural tube defect reported to two English regional data sets suggested that UK mothers of Pakistani, black African and black Caribbean ethnicity were less likely to terminate an affected pregnancy than mothers of white, Indian or Bangladeshi ethnicity. 112

It is more difficult to determine the underlying reasons for the higher postdischarge mortality of babies within the ‘other’ ethnic group, because this group is heterogeneous and has not often been described in detail. An analysis of UK Census data,113 which reviewed the country of birth of individuals identified within this group, found that over half were born in the Far East (including the Philippines, Japan, Thailand and Vietnam), 10% were born in the Middle East (mainly Iran and Iraq) and 7% were born in North Africa (including Egypt, Morocco, Algeria and Libya). The ethnic group classed as ‘other’ warrants further evaluation and exploration in order to better understand what factors are at play, given that we do not know how to explain the finding with any certainty. An unproven hypothesis is that a proportion of families of patients in the ‘other’ ethnic group within this study may have recently travelled or migrated to the UK, in some cases for treatment, although all those included in this data set had an NHS number and hence were residents in the UK at the time of surgery. The particularly high incidence of CHD (highest incidence of all ethnicities) in the ‘other’ ethnic group supports this hypothesis. Recent migrants to the UK may have less awareness of and access to services important post discharge, and this may have an adverse effect upon outcome. Further qualitative assessment of families in the ‘other’ ethnic group is required to address these questions.

Conclusion

Although this detailed analysis of routine audit data has highlighted a lack of consistency and completeness in ethnic coding, it has nevertheless provided very valuable insights into the influence of ethnicity on the distribution of CHD in the UK infant population and on outcomes for affected infants. Further investigation of the impact of ethnicity on pregnancy termination rates and infant mortality for babies diagnosed with CHDs is merited. Further investigation of the links between ethnicity, other aspects of case mix and outcome incorporating different stages of the patient journey over the first year of life (incorporating multivariable modelling) is warranted.

Data analysis

Responses were collated into a single transcript and thematic analysis was used to analyse forum responses. Thematic analysis involves:

• familiarisation with the data

• generating initial codes

• searching for themes among these codes (but not by counting them)

• reviewing themes

• defining and naming themes.

Analysis was conducted by members of the research team working as a group.

Participants and participation

The forum ran from February to May 2013. During this period, a total of 91 participants (mean age 35 years; range 23–58 years; 89 female; 89 parents, two grandparents) submitted demographic information and were given access to the closed forum group. Participants came from all over the UK and were predominantly of white British ethnicity (85 out of 91). Of these, 73 parents participated in the forum discussion and most responded to between one and five questions. Neither of the two grandparents contributed to the forum discussion. The questions with the most responses were those about information provided about caring for their baby at home (symptoms and how prepared they felt); community support [particularly experiences with general practitioners (GPs) and health visitors (HVs)]; and support from cardiac liaison nurses (CLNs). Parents also responded to others’ posts, offering support and sharing experiences, and towards the end of the forum parents became more open in identifying with other parents’ experiences: ‘After reading this it was like reading my own story’; ‘It is so good to read about someone who has done so well’. Selected quotations are included in Boxes 4–9.

Findings

The core theme emerging from the data was the family experience of isolation, epitomised by the following quote: ‘It’s a pretty lonely place’. This was described in terms of the way in which that isolation was experienced (physical, social and knowledge) and the resulting psychological impact, together with the factors that made that worse or better. There was also a theme of time, as parents moved from feeling overwhelmed and lacking in knowledge and skills to becoming ‘expert parents’ with a corresponding increase in their knowledge and skills. Figure 8 is a schematic representation of this.

FIGURE 8.

Themes from the online forum: ‘It’s a pretty lonely place’.

Limitations

There were a number of limitations with this element of the study, related to the method of collecting the data. The sample was predominantly white British, most of whom were mothers, corroborating research which has identified that the majority of internet users are women and participants in online interactions tend to be predominantly white, younger and highly educated. 134 In order to participate they required access to the CHF website and familiarity with Facebook, which meant that parents who did not speak English and were not able or willing to use social media could not participate. Furthermore, participants required a certain level of computer skills because of the requirement to register and login to the site. It is probable that parents from a more ethnically diverse population face additional challenges and have different experiences, which this method of data collection prevented us from capturing. Furthermore, families who are in contact with charities offering support may be more likely to engage with this type of research, thus limiting the representativeness of the sample. Finally, although questions posted on the forum related to discharge after infant cardiac surgery, some of the parents provided information and views about other stages of their journey (e.g. after surgery in later childhood). Although we did not include these posts in our analysis when this was made explicit, it is possible that some posts were included when the time that parents were referring to was not identified as being other than following cardiac surgery in infancy.

Online forums also have some inherent limitations compared with more traditional face-to-face methods of data collection. For example, non-verbal and contextual cues cannot be picked up on and specific participant comments cannot be probed, resulting in the potential loss of some richness of the data. Moreover, responses were generally much shorter than would be elicited in an interview. The automatically generated transcript, while being a benefit of this research method, was also not perfect, as has been identified previously in online forum research. 135 We did not collect information about the time spent on the site or number of visits participants made while the forum was running and we also did not know if any participants had technical problems accessing the site at any time or whether or not potential participants failed to join the forum at all. However, it is also important to consider the aims of this part of the study and the fact that we were not conducting a qualitative study requiring data saturation but wanted to elicit information about parental experiences to inform the development of a topic guide for the in-depth interview in the next phase of the study.

Conclusion

Use of an online forum provided a means of eliciting a large number of parents’ experiences of caring for their child after discharge from hospital following cardiac surgery. Parents engaged with the forum and were able to articulate what went well and what went less well, in addition to sharing their stories and supporting each other through doing so. Information gained from the forum was used to shape the questions for the parent interviews in a subsequent phase of the study and has been used to inform the development of an intervention, in conjunction with other findings from the study, for parents of high-risk babies following discharge after cardiac surgery.

Framework analysis

Framework analysis137 is a structured approach to managing qualitative data that allows researchers to organise and extract themes from the data more easily. It is a systematic approach that aims to reduce bias and make analysis of large data sets more manageable and involves entering qualitative data (e.g. quotes and summaries) into charts to aid interpretation, ensuring key themes are systematically searched for across each transcript.

Analysis using the framework approach consists of five stages.

• Familiarisation: listening to audiotapes, reading transcripts, becoming aware of key ideas/recurrent themes and noting them.

• Identifying a thematic framework: recognising emerging themes in the data set; may be guided by a priori issues but must also be open-minded to new themes; making judgments about meaning, relevance and importance of issues; may be tentative at this stage.

• Indexing: identifying data (e.g. quotes) that correspond to a particular theme.

• Charting: data identified during indexing are now arranged in charts of themes.

• Mapping and interpretation: analysis of key characteristics laid out in the charts; should be able to provide a schematic diagram.

Members of the core research team worked collaboratively and iteratively on the development of the frameworks, mapping and interpretation of the data for each of the participant groups.

Descriptive information

Specialist nurses contacted 25 families, 21 of whom agreed to be interviewed. One family was excluded, as they did not meet inclusion criteria, leaving a total of 20 families who were interviewed for the study. Of these families, 12 were bereaved. Fourteen interviews were conducted with one parent alone (n = 14 mothers), and six with both parents together. A range of ethnic, educational and socioeconomic backgrounds was represented in the sample (Table 20). One parent did not speak English as a first language and two were bilingual.

All children had their first surgery between September 2009 and October 2013. Following their initial surgery, 12 children were discharged home directly from the specialist surgical centre; the remainder were discharged to their local hospital in a ‘step-down’ arrangement. A case-by-case summary of complications and symptoms for each patient is provided in Table 21.

Data are presented in three main sections. The first describes symptoms of deterioration noted by families and is divided into five subsections: no symptoms, feeding, respiratory distress, appearance and behaviour. The second section focuses on decision-making about symptoms and the final section describes families’ experiences of seeking help. Additional quotes are presented in Boxes 10–16.

Symptoms

Decision-making about symptoms

Although many, but not all, parents recalled being given information about signs and symptoms during their child’s hospital admission this was not always sufficient to enable them to recognise these symptoms out in the community. Even when symptoms were recognised, parents sometimes struggled to describe these and to make decisions about a course of action. This was particularly true if symptoms appeared very subtle or had a gradual onset. One parent also commented that as this was her first child she found it difficult to evaluate symptoms, as she did not know what was ‘normal’ for a healthy baby. Several parents spoke of the burden of completing monitoring forms at home, particularly in relation to feeding, with one parent explicitly stating that she felt she may have missed early warning signs in her child as a result. There was also some difficulty identifying change when the child’s baseline was atypical. A small number of parents said that they did not recognise the symptoms on the first occasion but that once these had been pointed out to them on their own child they found it much easier to identify them on subsequent occasions.

Often decision-making about symptoms took place in the context of local services that were relatively unfamiliar with CHD in comparison with the specialist centres where their child had their surgery and, in some cases this had a detrimental effect on parent’s trust in their local hospital. Despite this, some families still recognised the need for their child to be known to local services.

Experience of seeking help

For symptoms that parents judged to be non-urgent, their first point of contact would typically be the HP with whom they felt they had the best relationship, often the CLN at their specialist centre or the community nurse. However, in some cases parents waited until their next follow-up appointment to discuss their concerns with their local paediatrician or their cardiologist, resulting in a delay in their child receiving treatment. Several parents mentioned fear of appearing ‘silly’ or ‘paranoid’, particularly to more senior HPs, although this could be countered by positive experiences of seeking help and reassurances at an early stage, typically from the liaison nurse, that they should telephone with any concern no matter how small.

Parents reported an overwhelmingly positive experience of the support they received from their CLN with this link being described by one family as a ‘lifeline’. This was particularly true if this was someone they had met during their hospital admission. In some cases the liaison nurse was able to intervene with a family’s local hospital to facilitate more rapid access and treatment in an emergency, discuss treatment plans with local HPs and arrange transport back to the specialist centre when required.

Not all families had a good experience of seeking help when they were concerned about their child, and several families said they felt that their concerns were not taken seriously by their local services. In some cases, parents were falsely reassured about symptoms or told to wait until their next follow-up appointment to discuss it with their cardiologist. Several parents described occasions when they had to be particularly assertive with HPs in order to get the right care for their baby.

In a number of cases parents felt that their local accident and emergency (A&E) staff were unprepared to manage a child with CHD in an emergency, and in some cases there were also problems with A&E staff gaining access to the information they needed to treat the child. Out-of-hours service at A&E was also raised by some parents, with many detailing long wait times and one parent describing having to make decisions about the severity of her child’s symptoms within working hours in order to get the best care.

Limitations

Our study has a number of limitations. First, parents approached to take part in the study were those known to the specialist nurses who assisted with recruitment. As this meant that parents were approached by someone familiar to them, resulting in a high opt-in rate, it is possible that those families who opted into the study are those who had a better relationship with the specialist nurses at their hospital. An important consideration of qualitative research methods is to describe, rather than quantify, the views held by a population of interest; therefore, it is important to ensure that the study sample represents the diversity present in the population being described. Our study included parents of children with a range of diagnoses, outcomes and discharge pathways. We also attempted to achieve diversity in our sampling of ethnicity and parent educational level, although, as is often typical in UK research, our sample included parents of predominantly white British children. We also struggled to recruit parents’ whose first language was not English, despite offering access to interpretation, and it is likely that these parents face additional challenges not captured by this study.

Conclusion

Many of the complications that can arise following congenital heart surgery may lead to relatively rapid deterioration in a small infant, thus leaving a small window of opportunity within which to intervene. This makes it particularly important for parents to be supported to recognise symptoms in their child and for them to be able to summon help quickly when there is a concern. The family burden of caring for a child with complex health needs is well known,138–144 and as families responding to their infant’s symptoms are likely to be acting under stress, it is important that the information they are provided with is straightforward and that help is readily accessible. Therefore, our study has implications for HPs involved in the discharge and follow-up of babies after congenital heart surgery, in relation to both interpretation of reported symptoms and the processes they follow in response.

Descriptive information

Information in respect of participating families is reported in Chapter 7.

Qualitative data from interviews

Data are presented below in sections corresponding to the patient journey, with illustrative quotations. Each section first comments on the areas of concern that were raised by respondents, and then highlights aspects of service provision they perceived to be effective. Illustrative quotes for each section are shown in Boxes 17–19.

Limitations

The study has several limitations. The parents approached were already known to the specialist nurses assisting with recruitment so potentially represent those with a good relationship to the local team. Despite our attempts to achieve ethnic diversity, three-quarters of our sample were ‘white British’ families. Professionals approached were purposively sampled so may not be typical of all clinicians involved in the care of cardiac infants; furthermore, we were limited to one or two professionals per centre. Although we interviewed professionals from secondary and primary care and from rural and urban settings, given the small number of subjects and the large extent of the services they represent, we may not have captured a complete range of views.

Implications of our findings

Our study demonstrates certain ‘system problems’ within the discharge and follow-up pathways for infants going home following cardiac interventions.

Paediatric cardiac network services incorporate multiple team interfaces with corresponding steep knowledge gradients and opportunities for information loss. Infants with CHD may be medically fragile and subject to dangerous deterioration; many non-specialist HPs and parents find this responsibility challenging and extremely stressful. The pressures resulting from the system problems that our study identified fall particularly on parents, CLNs and PECs.

Our study suggests that implementation of HMP (see Chapter 3) in the UK is variable and the complex regimens of feeding, medications, weights and saturation monitoring with ‘breach criteria’ place considerable burdens on those responsible for them.

In practice, infants with heart disease who become acutely unwell at home are likely to present to a GP, to a local hospital via A&E or through ‘open access’ to the paediatric department. Our study identified problems with correct identification of the deteriorating child, and difficulties determining what appropriate steps to take from the perspective of both primary and secondary health-care professionals. Within this context, interview participants reported the potential benefit to both parents and local medical personnel of local ‘step-down’ care before an infant’s final discharge home after cardiac surgery, such that the child is known when in a stable condition.

Our findings also highlighted the effective role that named and informed PECs with responsibility for local children with heart disease could play in strengthening networks. PECs are well placed to recognise postpericardiotomy syndrome or discriminate a respiratory infection from decompensated ventricular function, perhaps with support from a regional cardiologist. PECs are also well placed to address the less acute but nevertheless important aspects of managing the postoperative infant, particularly other comorbidities and dealing with feeding difficulties, reflux or lung disease in context.

Paediatricians with expertise in cardiology, however, may not be the first point of contact for the deteriorating infant with CHD, and there may be potential for learning from related secondary care examples, such as scoring tools that have been deployed in A&E and utilised by non-specialist HPs to detect signs of deterioration in presenting children. 146

Finally, at present the focus of national and international audit remains on 30-day mortality rates for paediatric cardiac surgery68,69,147 and we note that additional audit metrics focusing on the postdischarge stage of the patient journey may be a useful lever for quality improvement in the future.

Key findings from the helpline interviews

The core theme identified from the helpline interviews was communication and knowledge (Figure 9), which was described in terms of patient and family factors, sources of support and systems. Quotes for each theme are presented in Boxes 20–29.

FIGURE 9.

Themes from the helpline interviews.

Patient and family factors

Systems and processes

The third component of knowledge and communication focused on the systems and processes, in terms of the content, format and quantity of information provided, knowing the network and people who should be providing support to parents, together with barriers to obtaining support and the consequences of that. The overarching theme was consistency and continuity – or, conversely, the individual approaches in each centre, each community and of each professional, with the result that helpline staff perceived families to have very different experiences of ‘the system’ and how it works. This lack of consistency was perceived to extend across the network, from the tertiary centres and into primary and secondary care, and staff identified the need for a more joined-up and co-ordinated approach to care for individual families.

Participants also talked about their perceptions of what happens to families at discharge, in terms of the planning that happens prior to discharge, the practicalities at the time of discharge and communication with local HPs. In particular, staff described the lack of consistency and the impact on families of a poorly planned discharge. A number of staff identified processes that they thought worked and did not work for families once they were back home, which were primarily focused on the processes of communication and keeping the family engaged with services.

Limitations and strengths of the study

There are some limitations to this part of the overall study. First, although CHF and the three affiliated charities providing support to families were involved, the overall number of participants was small. The three organisations provide support in different ways and their access to specialised medical knowledge varies. Participants from three of the groups provide support to specific patient groups, whereas the remainder are there for a generic heart population covering the whole CHD spectrum. Participants could also comment only on the support and information needs of families who contacted the helpline, which is likely to result in the needs of some parents not being represented, such as those of parents from non-English-speaking populations or parents who were less able, for financial, social or other reasons, to contact the helpline. A further limitation was that not every participant had experience in each of the areas covered by the interview; for example, one helpline staff member had not taken any calls in relation to discharge after surgery. Finally, parents invariably call a helpline because they have a problem so the views expressed by the helpline staff are inevitably skewed towards the things that do not work or have gone wrong for families, rather than reflecting a more balanced view.

Despite these limitations, there are a number of unique elements with the approach adopted in this part of the study. The helpline staff are part of national charities and are therefore in contact with families from all over the UK, rather than only being exposed to families from a limited number of specialist centres. They do not have affiliations with any individual centre and, although individual helpline staff will inevitably hold opinions about aspects of care in specific centres, they were providing us with a national overview. Furthermore, charities do not provide clinical care and in most cases helpline staff are not medically trained, thus reducing the likelihood of medical or clinical bias influencing their responses.

Previous research into the role of helplines in health care has focused on the user perspective, involving interviews or surveys with callers rather than with the helpline staff. 151,152 We focused instead on accessing information about families’ needs and their experiences of using services by tapping into the staff’s body of experience. In common with studies of cancer patients’ use of helplines,150 interviews with helpline staff in our study indicated that parents’ reasons for calling charity helplines were multifaceted, with their psychosocial needs being intertwined with their needs for information and advice. Thus helplines clearly have a unique and important role in the support of families of children with CHD. Furthermore, the views of the helpline staff we interviewed provided important and previously uncaptured evidence to help inform and develop interventions to better support children and families after hospital discharge following infant cardiac surgery.

Establishing the final set of suggestions for health care

A working group comprising selected members of the expert advisory group and invited additional representatives from the community and charitable sector was convened to assess the draft suggestions for health care and propose a final set for endorsement by the IHS expert advisory group (see Appendix 10 for the working group’s terms of reference and list of members). A facilitated all-day workshop was held in September 2014 (tape-recorded and with live minutes), in which the group was tasked with:

• reviewing the draft suggestions for health care: assessing the feasibility and acceptability of each for service improvement

• assessing the set of draft suggestions for health care as a whole within the context of patient risk groups and targeting these (setting priorities)

• agreeing a final set of suggestions for health-care improvement to circulate among the IHS expert advisory group for comments and endorsement.

On the second point, the working group explicitly considered both the size and nature of the risk associated with each of the patient groups identified in our analysis of national audit data reported in Chapter 4 (summarised in Table 24). The groups are characterised by patient risk factors and stratified by occurrence of outcome 2 (death or emergency readmission to paediatric intensive care within the first year post discharge from infant cardiac surgery). Any decision about the types of interventions supported for different patient groups will have implications for resource use. Given the limited resources available for providing these services, the working group therefore agreed that it was important to consider the relative size of the patient groups alongside their relative risks and the nature of those risks when thinking about how limited resources might be targeted most effectively.

Strengths and limitations

A limitation of our research project was that the development of optimal guidelines for services across sectors is challenging, not least because there is often limited or disparate evidence that is difficult to synthesise (for example with no established methods equivalent to the meta-analysis of randomised controlled trials), and inevitably involves an element of subjectivity. In this context, our study had two key strengths. First, developing and applying a systematic and transparent process for synthesising and incorporating a broad range of available evidence covering multiple aspects of the problem enhanced the richness and breadth of the guidelines. For example, the qualitative evidence was very useful in specifying what the problems in services were and how they might be improved, and the quantitative evidence enabled prioritisation of patient groups according to their risk. Second, representatives from across the entire patient pathway critiqued the feasibility and acceptability of the recommendations, and the needs of service users remained of central importance through incorporating findings from an OF and interviews, views captured in a family workshop and involving a parent representative on the expert group.

Implications for practice

Our proposed guidelines for services are of direct relevance to all health-care professionals caring for infants with CHD including GPs, community nurses, HVs, secondary care paediatricians and clinicians in specialist surgical centres, as well as patients, their families and support groups. The proposals will resonate with clinicians, patient families and user groups from other geographical areas when later postdischarge CHD deaths have been reported as a concern in certain diagnostic groups, for example Germany55 and the USA. 26

As discussed in more detail in Table 25, there is a need for reduced variability and improved overall standards in respect of the training and information for families with infants affected by CHD predischarge, the process for discharge and transfer of infants with CHD to non-specialist services and their medical follow-up services, the non-medical support for patient families, the provision of patient information, the access to support when a baby is sick, and in reducing knowledge gaps and strengthening communication between HPs. It is likely that such interventions could have a material impact in terms of reducing postdischarge deaths and readmissions to intensive care, as well as improving the perceptions and experiences of services by patients’ families.

There is a role for the relevant professional groups (e.g. the British Congenital Cardiac Association and the PEC Special Interest Group), research funders, the national audit body and service users in taking forward certain aspects that require further research and development (see Recommendations for research). CHD services in the UK are currently under national review by NHS England and the findings and conclusions reported in this article fed into the review’s public consultation on the care standards and service specification to be used in commissioning specialist CHD services (see Appendix 11).

Suggestions for audit metrics

The evidence and suggestions from the IHS would support monitoring of the process and outcome measures, in order to increase the focus on this neglected phase of the patient journey and hence, it is hoped, to drive up standards of service provision. One potential option for such monitoring is the Quality Dashboard, which is guided by the Clinical Reference Group. Please note, however, that while the IHS team suggests the monitoring of out-of-hospital outcomes was highlighted as important within the IHS data, the analytical steps and processes required to do this are beyond the scope of the IHS team and would need to be considered further by, for example, the NCHDA steering committee.

The IHS findings would first need to be disseminated more widely within the community and agreement reached about the steps required for implementing the recommendations; for example, a number of metrics relate to standardised documents/protocols that would need to be developed and appropriately piloted beforehand. Others require further consideration as to how the information could be collected and/or aggregated. Furthermore, some findings may be considered more appropriate for local monitoring for improvement purposes, rather than for inclusion on the national Quality Dashboard.

The IHS therefore suggests that the feasibility and appropriateness of including any or all of the following metrics from Box 30 on the Quality Dashboard should be revisited in the next 1 to 2 years and, in the meantime, steps taken to develop these areas.

Recommendations for research

The study highlighted areas where further specific research is needed, these being as follows.

Level of evidence P1: quality assessment for inception (prospective) cohort studies (or meta-analyses of inception cohort studies), or validation of a Clinical Decision Rule (CDR)

Level of evidence P1: quality assessment for meta-analyses of inception/prospective cohort studies

A meta-analyses of these types of studies is also allocated a LOE P1.