Effectiveness of septoplasty compared to medical management in adults with obstruction associated with a deviated nasal septum: the NAIROS RCT

Evidence review of septoplasty effectiveness

To our knowledge, van Egmond et al. ,58 published the only RCT on the clinical effectiveness of septoplasty in 2019. This study was based on analysis of septal surgery, compared with non-surgical management, across 206 patients in the Netherlands. They concluded that septoplasty was more effective than non-surgical management for nasal obstruction. Those patients randomised to non-surgical management did not receive a standardised treatment, and they received a variety of medical interventions considered appropriate by the treating clinician. Therefore, it is difficult to conclude whether or not a positive trial is the result of an effective intervention or low-efficacy standard care. 59 Furthermore, the primary outcome measure of the trial, the Glasgow Health Status Inventory, a health-related quality-of-life measure, demonstrated an improved score among those patients undergoing surgery. However, the Glasgow Health Status Inventory is known to have limited sensitivity to septal surgery outcomes,60 and, when compared with the pragmatic medical treatment comparator undertaken by van Egmond et al. ,58 this effect may have been exacerbated. Over 75% of the surgical cohort underwent concomitant inferior turbinate surgery, bilateral in 67% of patients, even though inferior turbinate enlargement was noted in < 50% of patients at baseline. This additional surgery is likely to exaggerate the beneficial impact of turbinate surgery over septoplasty alone. 2 In addition, there was a 30% crossover from the non-surgical to the surgical arm, introducing a further potential element of bias or dilution of the full effect of septoplasty.

In a 2018 systematic review of the evidence for septoplasty with or without inferior turbinate reduction as a treatment for nasal obstruction, van Egmond et al. 61 first compared septoplasty (with or without concurrent turbinate surgery) with non-surgical management and, second, compared septoplasty alone with septoplasty with turbinate surgery. In their review, there were no RCTs comparing septoplasty with non-surgical management, and five RCTs and six controlled trials comparing septoplasty alone with septoplasty with turbinate surgery. Included studies demonstrated substantial heterogeneity in study population, outcomes measured and time points of outcome assessment. The risk of bias was considered high in most reports. However, although subjective and objective assessment improvements did not necessarily mirror one another, studies demonstrated an overall improvement in nasal obstruction symptom. No additional benefit of turbinate surgery was reported in 8 out of 9 studies using subjective assessments, and 5 out of 7 studies using objective measures. Complications were rare, and were reported in only three studies. There were significant limitations related to short follow-up periods, which in some cases were only 9 months.

In a second systematic review of septoplasty alone, Tsang et al. 62 found six studies assessing patient satisfaction; rates varied from 69% to 100% in three studies in which patients were asked if they were satisfied or dissatisfied with outcomes at 6 months. Two studies assessed the degree of patient satisfaction, with one study indicating that 88% of patients were moderately satisfied or better at 1 year post operation, and the other reporting that 50% of patients were satisfied at 5 years post operation. There was significant heterogeneity in the method of assessment of nasal obstruction, which may have led to the finding that, in general, patients with more severe symptoms of nasal obstruction had a better outcome. The authors noted that there were high risks of bias as studies were only observational, with significant variation across multiple categories including patient population, outcome measures and follow-up duration. Overall, the authors concluded that there was insufficient evidence that septoplasty alone had good long-term patient-reported outcomes in the management of septal deviation, based on the heterogeneity of existing data and lack of RCTs. They recommended further research to define which preoperative characteristics are predictive of both subjectively and objectively positive septoplasty outcomes. Neither systematic review undertook a meta-analysis because of the substantial heterogeneity across the studies included.

Surgical

Septoplasty has level III evidence of effectiveness, and observational studies confirm good levels of patient satisfaction. The operation, if successful, requires a single procedure without the requirement of ongoing medical therapy. However, there is variation in criteria for surgery, surgical technique and postoperative follow-up. Selection criteria for surgery variance was discussed previously, but there is also variance in operative techniques among surgeons and the need for patient follow-up postoperatively. In addition, there is no good-quality evidence on either the clinical effectiveness or the cost-effectiveness of such surgery. Septal surgery also carries an economic cost, in terms of time off work or normal duties, and can be associated with side effects or complications that delay recovery and potentially necessitate additional treatment.

Following surgery, it is normal to have some symptoms, lasting between 48 and 72 hours, of minor bleeding, congestion and nasal discomfort. In our own group’s early publication63 of the outcomes among 121 septoplasty patients at 6 weeks, two postoperative complications were noted: septal perforation (1.7%) and nasal septal adhesions (3.3%). A Chinese study of 54 patients reported nasal septal adhesions in > 7% of patients. 64 Adhesions and septal haematoma may necessitate re-admission for corrective surgery. In addition, minor cosmetic change occurs in up to 30% of patients and more major change in > 4% of patients. 65 Dabrowska et al. ,66 in a retrospective series of 5639 patients undergoing septoplasty with or without turbinate reduction, reported excessive bleeding in 3.3%, septal perforation in 2.3%, infection (prolonged healing) in 3.1%, reduced smell acuity in 3.1% and dental anaesthesia in 0.1% of patients.

Septoplasty is routinely performed under general anaesthesia as a day-case procedure. An initial incision is made on one side of the septal lining allowing for the straightening or removal of areas of twisted cartilage and bone. It may not be possible to fully straighten the septum without risking the cosmetic appearance of the nose; therefore, the surgeon must use careful judgement to minimise this risk. The procedure typically takes between 45 and 60 minutes to complete. Nasal packing can be used following surgery; in the NAIROS it was not to be undertaken, if possible, because of the associated discomfort to the patient. Instead, suture repair of the septum was to be performed at the end of the procedure, as this seems to have equal efficacy and fewer disadvantages. 67–69

Medical management

Nasal steroid sprays have potent anti-inflammatory and antiallergic properties, inhibiting the release of inflammatory mediators produced by the nasal lining, thereby reducing swelling and nasal mucus in the nasal passages. Two sprays of 100 μg of mometasone furoate in each nostril twice daily for 6 weeks, followed by 100 μg per day in each nostril, for the remainder of the 6-month treatment period, was chosen based on the maximum recommended standard nasal steroid dose. This spray was identified in patient and public involvement (PPI) discussions with GPs as being restricted by formulary protocols in primary care, and therefore unlikely to have been prescribed previously. Mometasone is licensed for use among patients with reduced nasal airway and has marketing authorisation in the UK.

Medical therapy has the obvious advantage of the avoidance of general anaesthesia and surgery and their associated risks. Nasal steroid sprays are standard treatment for nasal obstruction symptoms and are deliverable in primary care. However, they may not be an effective treatment for patients with a deviated nasal septum70 and, even when they are, patients may require indefinite medical therapy, with its associated costs. Intranasal steroid sprays also have potential risks to the patient: they can cause nasal bleeding (odds ratio 1.56 in a 2020 systematic review, compared with placebo71), crusting, pharyngitis and headache, all of which may affect patient compliance, and there are reports which associate increased intraocular pressure and adrenal suppression with such sprays. Both hypersensitivity and anaphylaxis are rare complications. 72 There are few data comparing different intranasal steroid preparations. 72

Proprietary saline irrigation of the nasal cavity may be performed with either isotonic or hypertonic solutions. 73 These may either be low-positive pressure (from a spray or pump), or gravity-based pressure (using a container with a nasal spout). Saline nasal irrigations have been recommended in the management of various nasal and sinus mucosal disorders such as chronic sinusitis74 and allergic rhinitis,75 although this recommendation is based on mostly low-level evidence. Similarly, saline irrigations have been recommended following endoscopic sinus surgery, but uncertainty remains in the literature around the optimum method of delivery and composition of the saline solution. 73–79 Saline sprays have not been trialled specifically in nasal septal deviation. In consultation with GPs and the PPI group, the Stérimar isotonic spray (Sofibel SAS, Paris, France) was chosen for the NAIROS. Mometasone and Stérimar sprays, in combination with the option of deferred surgery for those in the medical management arm, was noted at initial PPI discussions to be an acceptable alternative to surgery by patients.

Primary objectives

The primary objective was to compare the clinical effectiveness of nasal septoplasty (with or without unilateral turbinate reduction) with medical management over a duration of 6 months among adults with a nasal septal deviation who have been referred to otolaryngology outpatient clinics with nasal airway obstruction.

Secondary objectives

The secondary objectives are split into three different aspects: clinical effectiveness, an economic evaluation and a mixed-methods process evaluation.

Feasibility phase

The NAIROS researchers undertook an extensive assessment of the feasibility of the study by triangulating the views of potential patients, GPs and consultant ENT surgeons. This work assessed the willingness of patients to participate in randomisation, clinicians’ willingness to refer patients and randomly allocate to trial groups and the acceptability of the deferred surgery treatment arm.

Internal pilot

We originally intended for the NAIROS to include a 5-month internal pilot involving 10 sites. However, delays in study set-up expedited an agreement with the funder to remove the internal pilot and open all 17 sites simultaneously.

The pilot objectives of identifying, understanding and addressing any barriers to recruitment, retention or compliance with protocol were incorporated into the main trial, but extended to the full first year of recruitment. Areas for particular scrutiny during this period included patient recruitment, patient discontinuation of allocated treatment and compliance with surgery window. Monitoring of recruitment, retention and compliance with the protocol continued throughout the whole trial.

Collecting primary and secondary outcome measures

All trial interventions had been administered before the onset of the COVID-19 pandemic in March 2020. Participants in the trial follow-up were invited to complete the PROMs remotely via e-mail or post [i.e. SNOT-22, NOSE, healthcare utilisation questionnaire (HUQ), time and travel questionnaire and SF-36; see Report Supplementary Material 2, 4, 5 and 6]. The SNOT-22 (primary and secondary outcome measure) could also be completed by participants using a validated online platform hosted by Castor (Castor EDC, Amsterdam, the Netherlands); the method of completion of the SNOT-22 was noted in the database. Owing to suspension of all face-to-face clinic visits from 30 March 2020, no other clinical outcome measure scheduled after this date was collected [i.e. DOASS (see Report Supplementary Material 3), clinical examination, symptoms review, measures of nasal patency]. AEs and concomitant medications were collected remotely over telephone at the 6- and 12-month visits from 30 March 2020 until the end of the trial.

Inclusion criteria

• Adults aged ≥ 18 years.

• Baseline NOSE score of ≥ 30.

• Septal deflection at baseline visible via nasoendoscopy.

• Capacity to provide informed written consent and to complete trial questionnaires.

Exclusion criteria

• Any prior septal surgery.

• Systemic inflammatory disease or the use of oral steroid treatment within the previous 2 weeks.

• Granulomatosis with polyangiitis.

• Nasoendoscopic evidence of unrelated associated pathology, for example adenoid pad, septal perforation, chronic rhinosinusitis indicated by polyposis or pus.

• Any history of intranasal recreational drug use within the previous 6 months.

• Breastfeeding, pregnancy or intended pregnancy for duration of involvement in the trial.

• Bleeding diathesis.

• Therapeutic anticoagulation (warfarin/novel oral anticoagulant therapy).

• Clinically significant contraindication to general anaesthesia.

• Known to be immunocompromised.

• Presence detected of an external bony deformity likely to make a substantial contribution to the nasal obstruction (determined by clinical opinion).

The NOSE score is a validated five-item, unifactorial self-report of nasal-block severity that has been applied in previous research and audit studies. 36,81 The three recognised NOSE-derived categories of baseline severity used were 30–50 (moderate), 55–75 (severe) and 80–100 (extreme). The NAIROS aimed to recruit participants with at least moderate nasal obstruction symptoms.

Septoplasty

Participants randomised to septoplasty received surgery up to 8 weeks (+ 4 weeks) after randomisation. The additional 4-week window was to allow for extenuating circumstances such as pressure on surgical waiting lists or patient cancellation, with reasons for delays to surgery collected and reported. The NAIROS stipulated that experienced surgeons who were not in training should perform the procedure. They had the option to carry out unilateral turbinate surgery on the wider side, according to their assessment of the individual patient airway. As a pragmatic study, the NAIROS did not ask surgeons to change their usual practice in this regard, reflecting the considerable variation in current UK surgical practice. Both the intention to reduce one turbinate prior to randomisation and details of the actual surgery performed were collected.

Clinical Trial of an Investigational Medicinal Product intervention: medical management

Participants randomised to medical management were supplied with 6 months of Stérimar isotonic nasal saline spray and mometasone furoate nasal steroid spray at the time of randomisation. Use of mometasone, an investigational medicinal product, was as indicated by the marketing authorisation. Twice per day for 6 weeks, each participant sprayed a daily metered Stérimar isotonic nasal saline dose into each nostril, followed by a twice-daily dose of mometasone furoate steroid spray. For the remainder of the 6-month period, the same dose of saline spray applied, and the steroid dose was reduced to 100 μg per day; participants could administer this either by spraying either two 50-μg doses into each nostril once daily or one 50-μg dose into each nostril twice daily.

Discontinuation of allocated treatment

Formal crossovers between trial arms were not permitted, but the NAIROS was designed as a pragmatic RCT; therefore, participants could discontinue allocated treatment and explore other options in standard NHS care while remaining in the trial. Medical management arm participants opting for surgical treatment were invited to defer surgery until after the 12-month follow-up visit and received standard non-trial septoplasty in line with current local waiting times (taking into account the time that they had already spent in the medical management arm). Discontinuation of allocated treatment did not constitute withdrawal from the trial.

Participants randomised to the surgical arm, or medical management participants who wished to continue using nasal sprays beyond the initial 6-month intervention period, could be prescribed mometasone furoate nasal spray and Stérimar isotonic spray (or an alternative at the discretion of the clinician) as per standard practice of the local NHS team and without the need for a NAIROS trial prescription.

Participants who wished to discontinue their allocated treatment but remain in the trial were followed up as scheduled for their allocated arm, with data analysed on an intention-to-treat (ITT) basis.

Funding of the trial intervention

As the surgical intervention was part of the standard pathway of NHS care for this cohort of patients, it was funded from the standard NHS tariff. The 6-month medical management intervention was categorised as an excess treatment cost and was funded by NHS England, Scotland and Wales.

Primary outcome

The primary outcome was defined as patient-reported assessment of nasal and general symptoms assessed using the SNOT-22 at 6 months (−2 weeks/+ 4 weeks). The SNOT-22 is scored from 22 questions with each item scored from 0 to 5. The final total score can range from 0 to 110, with a higher score indicating worse symptoms. The SNOT-22 was also assessed at baseline and 12 months.

Secondary outcomes

• Objective assessment: nasal airflow was measured using a PNIF meter and an NV1 rhinospirometer (GM Instruments Ltd, Irvine, UK); both rhinospirometry (i.e. tidal volume and MIV) and PNIF measurements were taken at baseline and at 6 and 12 months. Additional exploratory analysis was undertaken using the rhinospirometer.

• Quality of life was measured using the SF-36 quality-of-life questionnaire (acute/1-week recall) at baseline and at 6 and 12 months.

• Healthcare resource use (primary and secondary care) was measured using a HUQ at baseline and at 6 and 12 months.

• Patient-reported outcome measures: subjective – SNOT-22 subscales (rhinologic, sleep, ear/facial pain, psychological) at 12 months, NOSE scale at baseline and at 6 and 12 months, and DOASS at baseline and at 6 and 12 months were used to measure longer-term outcomes.

• Safety measures: number and characteristics of any AEs and surgical complication/failure and reintervention within 12 months.

• The economic evaluation compared the following between the two intervention arms:

  • costs [the average total cost per participant from the perspective of the NHS and Personal Social Services (PSS)]; sensitivity analyses included participant costs

  • QALYs, based on responses to the SF-36 converted to SF-6D scores, were derived using the area under the curve method82,83

  • improvement (of ≥ 9 points) in SNOT-22 scores

  • number of AEs

  • incremental cost per improvement (of ≥ 9 points) in SNOT-22 score

  • incremental cost in number of AEs avoided

  • incremental cost per QALY gained.

• As part of the economic evaluation, costs and effects were extrapolated beyond 12 months using an economic model.

Identification, screening and recruitment of participants

Although primary care clinicians were encouraged to refer patients with a deviated nasal septum directly to nasal research clinics, the majority of eligible patients were proactively identified by researchers from general ENT primary care referrals. Triage of paper referrals and scrutiny of ‘choose and book’ referrals were used to populate research clinics, or alternatively, research slots in general rhinology/ENT clinics. At the majority of sites, clinicians did not have the resources for dedicated research clinics. Nasal septal deviation can be challenging to diagnose in primary care; therefore, the participants selected to attend research clinics were chosen at the clinicians’ judgement.

Screening for study eligibility was maximised by training all staff involved. Potential participants, whenever possible, were sent the participant information sheet (PIS) with their clinic appointment details and were directed to the patient information video available on the trial website. 85 All patients were given a minimum of 24 hours after receiving the PIS to decide whether or not they wished to participate.

Patients were invited to provide written informed consent for the study in three stages, after they had been given sufficient time to consider the trial and had the opportunity to have any questions addressed by the local clinical team.

First, patients were invited to provide consent to undergo screening assessments to determine eligibility for inclusion in the trial.

Participant allocation

At the baseline visit, consenting, eligible patients were randomised on a 1: 1 basis using the centrally administered Newcastle Clinical Trials Unit (NCTU) web-based system. Randomisation was by random permuted blocks of variable length, stratified by gender and three recognised NOSE-derived categories of baseline severity as defined previously in Eligibility criteria. The treatment allocation was open-label, with the randomisation system providing a unique trial identifier for each participant.

Septoplasty participants

The surgeon’s intention to reduce the inferior turbinate was recorded prior to septoplasty and the following information was recorded at the time of surgery:

• the date of surgery

• duration in theatre

• grade of senior surgeon and senior anaesthetist present

• whether septoplasty, with or without unilateral inferior turbinate reduction, was carried out

• technical aspects of the surgery

• discharge medication

• whether or not any complications occurred

• whether or not there was any overnight hospital admission.

Medical management participants

As the NAIROS was a pragmatic trial of standard treatment, formal participant compliance with the medical management intervention was not formally assessed. However, quantities of bottles used per participant and reasons for ceasing treatment were recorded.

Follow-up

Participants were contacted by either telephone, e-mail or text 2 weeks after randomisation (medical management) or 2 weeks after their septoplasty, to record any AEs and concomitant medication. Medical management participants were reminded to reduce their dose of mometasone nasal spray at 6 weeks.

Six months after randomisation (− 2 weeks/+ 4 weeks)

Assessments performed are detailed in the trial schedule of events (Table 1). The 6-month follow-up visit was scheduled to allow a minimum of 12 weeks’ recovery from septoplasty surgery.

For participants allocated to the surgical arm, any complications from the septoplasty were recorded.

Twelve months after randomisation (± 2 weeks)

Assessments at the 12-month follow-up visit are detailed in the schedule of events (see Table 1).

At both the 6- and 12-month visits, participants were given the option to discontinue allocated treatment and explore treatment options available within standard local NHS care.

Sample size calculation

The sample size calculation was based on a t-test for superiority assuming equal variance across groups, a conservative approach given the primary analysis was based on adjustment for the baseline values of SNOT-22, which generally increases statistical power. The target recruitment number of 378 participants allowed for a 20% drop-out rate (based on experience from two prior septal surgery audits). 86,87 The remaining 302 participants (151 per arm at completion) would be sufficient to deliver 90% power to detect a 9-point difference in the overall SNOT-22 score between arms. This assumed a 5% type I error rate and a standard deviation (SD) of 24 points. The MCID of 9 units was informed by relevant literature; it was felt to be the most relevant and a change of 8.9 units had been identified as being clinically meaningful. 43,45,46,88,89 The same literature was used to guide assumptions about what would be a reasonable value for the SD for the design parameter. We took a conservative approach as we assumed that the SD would be 24 points, which was the largest value reported. 43,45,46,88,89

Primary outcome measure

The primary outcome measure was the SNOT-22 score at 6 months. SNOT-22 scores were recorded at baseline and at 6 and 12 months post randomisation. Baseline and follow-up data were summarised using appropriate statistics and graphical summaries. Box plots (e.g. Figure 6) show summary statistics of the measurement they represent. The box represents the middle 50% of the data (lower quartile to upper quartile), the line within the box shows the median (50th percentile), the whiskers show data that fall within 1.5 × the interquartile range (IQR) and the points show data that fall outside these limits. SNOT-22 questionnaires with up to 20% of items missing were imputed, with the average of the completed questions used for missing items.

Summary statistics of overall scores, including means with associated 95% confidence intervals (CIs), are presented by treatment group and overall in Table 6.

Defining the populations for analysis

The following analyses by population were undertaken:

• Intention-to-treat group – all ineligible and protocol-violator participants included in the analysis on an ITT basis, with participants kept in their randomised treatment group. This included outcome measures completed at any time.

• Compliant ITT group – all participants in the ITT group, but complying with questionnaires completed within the 6-month (− 2 weeks/+ 4 weeks) return window with no consideration given to septoplasty status.

• Per-protocol group – all participants who received the treatment they were randomised to and complied with protocol in terms of timings and compliance windows for the surgery and primary end point (6-month visit/SNOT-22 completion). This excluded participants randomised to septoplasty but who did not receive septoplasty within 12 weeks of randomisation, participants randomised to medical management who had actually received standard care septoplasty before the 6-month + 4-week primary end point, and participants whose primary end point was completed outside the compliance window.

• Per-treatment group – this was similar to the per-protocol group, with some additional participants included. The medical management group was exactly the same as in the per-protocol group. Any participant who was randomised to septoplasty and received their septoplasty at least 10 weeks before the primary end point was included in the septoplasty group. In addition, any participant randomised to medical management but who received septoplasty at least 10 weeks before the primary end point was completed was included in the septoplasty group.

• Non-randomised group – those eligible to be included in the NAIROS trial but who declined to take part. We planned to compare the non-randomised group with those consenting to take part in the trial; 45 patients agreed to join the non-randomised group and allow their data to be collected and analysed. However, owing to the small number of patients who agreed to join the non-randomised group (only 19 of whom provided NOSE data), a meaningful comparison between this group and those consenting to the trial could not be made and these data were not included in this report.

Analysis of the primary outcome

Secondary outcome measures

The analysis of secondary outcomes followed a broadly similar strategy to the primary outcome measure. Secondary outcomes included data at the 6-month follow-up from the other outcomes [i.e. NOSE, DOASS, PNIF (maximum of three measurements), NPR from MIV (using mean volumes from three measurements) and tidal breathing] and data for all outcomes at the 12-month follow-up. SNOT-22 subscales (rhinologic, sleep, ear/facial pain, psychological) at the three time points are presented.

Summary statistics and graphical representation of subjective scales were tabulated at randomisation and at 6 and 12 months’ follow-up, both by intervention arm and collectively. Multiple regression was used to compare the outcome scores between treatment groups at follow-up time points. Variation between sites was included as a random effect with an assumed normal distribution, with analysis including the stratification factors of baseline severity and gender. Further adjusted analyses included terms for baseline values of the scores and key demographic and clinical covariates.

Adverse events were tabulated according to the World Health Organization Common Terminology Criteria for Adverse Events, version 4.03, with the number of severe AEs (Common Terminology Criteria for Adverse Events grade 3, 4 or 5) reported as a proportion of all AEs. The number of participants experiencing at least one severe AE according to the Common Terminology Criteria for Adverse Events was reported as a proportion of all participants. Surgical complication/failure and reintervention were described and not subjected to statistical testing. Technical failures (defined as occasions when the participant self-reported symptoms remaining the same or worsening, along with surgeon opinion on whether revision surgery was required) from operations in which widening of the nasal airway was achieved were reported. Complications experienced as a result of the septoplasties are also presented in the results chapter.

We carried out a subgroup analysis for participants who were recommended to receive the inferior turbinate reduction. This led to four groups: as randomised, recommended for turbinate reduction or not. This broadly followed the primary analysis, but analysis was carried out separately for each subgroup. Subgroups by gender are also reported.

Overall summary following primary analysis and sensitivity analyses

• Marked improvement seen in the septoplasty arm, with an average difference of −20 units (95% CI −24 to −16; p < 0.0001), compared with the medical management arm, at 6 months.

• The lower limit of the 95% CI for SNOT-22 score improvement is −16.40 units, which is substantially greater in magnitude than the MCID of −9 units used in the sample size justification for the trial.

• All sensitivity and secondary analyses show very similar results. There is strong pragmatic evidence that septoplasty is effective at reducing SNOT-22 score at 6 months, compared with medical management.

Double Ordinal Airway Subjective Scale

The DOASS is a PROM; it collects data (a score) for each nostril. Appendix 1, Figure 25, shows the raw data collected from randomised participants at each time point, by nostril (better and worse). The summary statistics of the raw data are tabulated in Appendix 1, Table 53.

The raw data are presented in terms of worse and better nostrils. These data are paired (both nostrils measured) at each time point. The nostril with the lower score at baseline is defined as worse, and the nostril with the higher score is defined as better (range 1–10). The nostrils defined as worse and better at baseline are presented at the 6- and 12-month follow-up time points. We used the scores taken at the 6-month follow-up to define which nostril was better or worse for participants whose nostrils both had the same score at baseline.

Appendix 1, Figure 25 shows that, for participants randomised to septoplasty, the score for the worse nostril had improved by the 6-month follow-up. The difference in scores between the worse and better nostrils reduced dramatically among septoplasty patients at 6 months. The differences between the worse and better nostrils are less evident in the medical management group follow-up scores, although improvement over time is seen in both groups.

Subjective Double Ordinal Airway Subjective Scale

One method to present the DOASS as a single summary ratio is the subjective DOASS. This representation of the magnitude of the differences between the two sides is generally presented as the NPR, calculated as (left score − right score) ÷ (left score + right score). Using the subjective DOASS was advisable as this same formula is also used to combine rhinospirometry paired nostril data. As we consider that laterality is irrelevant, we present absolute subjective DOASS, which is the modulus of the subjective score (i.e. sign is ignored). Scores close to zero mean that there is little difference between the nostrils. Summary statistics for absolute subjective DOASS NPR can be found in Appendix 1, Table 50.

Figure 10 shows that scores tend to be closer to zero (symmetrical nasal passages) for the septoplasty arm, which shows a reduction in subjective severity at 6 months. There is evidence of a more modest improvement in the medical management arm at 12 months.

FIGURE 10.

Absolute subjective DOASS.

Sino-nasal Outcome Test-22 items subscales

Summary statistics for the SNOT-22 subscales at all three study visits are presented in Appendix 1, Table 51. The corresponding box plots are shown in Figure 11.

FIGURE 11.

The SNOT-22 subscales. (a) Sleep score; (b) nasal score; (c) otologic score; and (d) emotional score. (continued)

The sleep and nasal subscales comprise 8 of the 22 items in the questionnaire; the otological subscale comprises four items and the emotional subscale comprises the remaining two items.

Appendix 1, Table 51, and the associated box plots (see Figure 11) show that improved and maintained scores in the septoplasty arm, and modest improvement in the medical management arm, are found in all subscales at both 6 and 12 months. The subscale scores range from 0–40 (sleep and nasal subscales), 0–20 (otologic subscale) and 0–10 (emotional subscale).

Nasal Obstruction Symptom Evaluation

The PROM NOSE scores were collected at baseline and used as a randomisation stratification variable. NOSE scores were also collected at 6 and 12 months. Continuous NOSE scores at baseline are summarised in Table 7 by allocated treatment group and overall using descriptive statistics.

As demonstrated by the overlapping CIs (and as expected for a randomised trial) there is no evidence of a difference in baseline scores between the two arms.

However, as evidenced by non-overlapping CIs, there are significant differences at the 6-month (primary end point) and 12-month follow-ups.

The NOSE score can range from 0 to 100, with higher scores representing worse symptoms. NOSE scores are similar at baseline across both treatment groups, as would be expected with random allocation (see Appendix 1, Figure 26). Figure 12 shows that the shift in NOSE scores at 6 and 12 months shows a pattern similar to that observed for SNOT-22 scores, that is, the scores tend to improve modestly for medical management, but more markedly for septoplasty at the 6- and 12-month time points.

FIGURE 12.

The NOSE scores at baseline, 6 months (primary end point) and 12 months, by allocated arm.

Measurements of nasal patency

Objective measurements of nasal patency were collected at all three trial visits. The measures collected were PNIF and rhinospirometry measures of MIV and tidal breathing. Measurements were taken both pre and post decongestant, with only the post-decongestant measures presented here. The absolute NPR for MIV and tidal breathing are presented in this section. Summary statistics for the following data are presented in Appendix 1:

• PNIF (post decongestant, maximum of three measurements) (Figure 13) (see Appendix 1, Table 52)

• inhaled volume by worse (lower-score) side and better (higher-score) side from post-decongestant MIV rhinospirometry (mean volume from three measurements) (see Appendix 1, Table 53 and Figure 27)

• absolute NPR from post-decongestant MIV rhinospirometry (using mean volume from three measurements) (Figure 14) (see Appendix 1, Table 54)

• inhaled volume by worse (lower) side and better (higher) side from post-decongestant, tidal breathing rhinospirometry (one measurement) (see Appendix 1, Table 55 and Figure 28).

FIGURE 13.

Post-decongestant PNIF rate (maximum of three measurements) (ITT population).

FIGURE 14.

Absolute NPR from post-decongestant MIV (using mean volume from three measurements) (ITT population).

Absolute NPR measures are defined as the modulus (ignoring the sign) of the NPR, which is calculated in the same way as for the DOASS. The COVID-19 pandemic resulted in the suspension of face-to-face visits, which had a major impact on data collection. Thus, there are limited follow-up data, particularly at 12 months. It is notable that 37 medical management participants underwent NHS surgery after 6 months, whereas only five underwent surgery before the 6-month time point.

Summary statistics of the absolute MIV NPR can be found in Appendix 1, Table 54. Summary statistics showing raw (volume) data for post-decongestant MIV by better and worse nostril can be found in Appendix 1, Table 53, with the accompanying box plot in Appendix 1, Figure 27.

Figure 15 shows the absolute tidal breathing NPR for the ITT population. Summary statistics for the absolute tidal breathing NPR can be found in Appendix 1, Table 56. Summary statistics showing raw data for post-decongestant tidal breathing by better and worse nostril can be found in Appendix 1, Table 55, with the accompanying box plot in Appendix 1, Figure 28.

FIGURE 15.

Absolute tidal breathing NPR from post decongestant (ITT population).

During the COVID-19 pandemic, face-to-face visits at trial sites were suspended, which meant that fewer participants completed the objective nasal patency outcome measures than had completed the PROMs. Despite the reduction in available data for the objective nasal patency measures, a marked improvement is seen by the 6-month measurement in the septoplasty arm, with a more modest improvement seen in the medical management arm.

Analysis of secondary outcomes

The 12-month SNOT-22 scores, and both the 6- and 12-month NOSE and DOASS scores were analysed in a way similar to that in model 1 (see Efficacy analysis: primary outcome). All analyses are for the ITT population, but numbers analysed are reduced owing to lower rates of completion for secondary outcomes.

Table 8 gives the treatment group coefficient of the regression analysis for each secondary outcome placed as the response variable, adjusted for stratification factors (gender and baseline NOSE categories) and the appropriate baseline measure for each secondary outcome, as well as a summary comment highlighting the direction of the effect.

Participants’ opinions of success of the treatment provided

One of the clinical effectiveness aims of the NAIROS trial is to use the results in the surgical arm to explore a possible definition of ‘technical failure’ in experienced hands, that is experienced surgeons, (i.e. consultants or non-consultant career clinicians, but not trainee otolaryngologists). To explore this, the investigator recorded the participant’s satisfaction with the surgery outcome in the post-surgery case report form. Responses were received from 133 out of 166 (80%) of those randomised to receive septoplasty; 116 of the 133 respondents (87%) were recorded as being satisfied and 17 (13%) were recorded as being not satisfied. In addition, for 11 of the medical management participants who requested to receive septoplasty, 10 were recorded as being satisfied and one was recorded as not being satisfied.

To explore the possible relationship between participants’ views and surgeons’ impressions as to the need for revision surgery, Table 9 combines responses to the following questions with surgeon opinions:

• At 6 and 12 months, participants were asked by the investigator, ‘In the past 6 months do you feel your nasal symptoms have been better than/about the same as/worse than before treatment commenced?’.

• At 6 and 12 months, investigators were asked, ‘Will septoplasty or revision septoplasty be required?’. A response of yes/no was recorded. Surgeons could recommend revision surgery if they felt that it was needed.

Only six revision surgeries were recommended for participants randomised to receive septoplasty, four at 6 months and two at 12 months. In one case, the recommendation for revision surgery by the investigator contradicted the participant’s view that their symptoms had improved.

Turbinate reduction

At the baseline visit, participants were assessed to determine whether or not it would be appropriate to reduce the inferior turbinate. These assessments are summarised in Appendix 1, Table 57. The relationship between this recommendation and the occurrence of turbinate surgery is shown in Appendix 1, Table 58. A subgroup analysis for participants in the septoplasty arm only was carried out.

Of the 155 septoplasties for which information on turbinate reduction was available, 88 (57%) included a turbinate reduction.

Of the 166 septoplasties carried out in the septoplasty arm, 148 have both baseline and 6-month SNOT-22 data, and so could be included in this analysis.

We analysed the data using linear regression adjusting for baseline SNOT-22 score and the stratification factors of severity and gender (Table 10). The dependent variable was SNOT-22 score at 6 months, and the model was fitted with the additional binary variable that indicated whether or not the turbinate was reduced in surgery (see Appendix 1, Figure 31).

The regression analysis shows that, on average, adjusted scores were 2.79 points higher (95% CI −2.78 to 8.35 points; p = 0.324) for those who received a turbinate reduction than for those who did not. As turbinate reduction was not a randomised intervention, but a clinical/surgical decision made within the randomised septoplasty, no conclusion can be drawn as to whether this was an effective part of the surgery.

Subpopulation treatment effect pattern plot by baseline Nasal Obstruction Symptom Evaluation score

Figure 16 shows the STEPP for the ITT population, demonstrating the impact of baseline NOSE score on the primary outcome (i.e. SNOT-22 score at 6 months).

FIGURE 16.

The STEPP of all participants in the ITT population.

The green line with shaded 95% CI limits in Figure 16 shows the average effect of being randomised to septoplasty for those with specific NOSE scores at baseline. For those with moderate NOSE scores at baseline, the average improvement in SNOT-22 score is around 5 units. Those with scores of around 60 have an average improvement of around 15 units, and those with extreme scores at baseline can improve by as much as 30 units. All of the values covered by the green line show improvements in SNOT-22 scores. However, it is clear that improvement increases as baseline severity increases.

It is important to consider the ‘floor’ effect of the SNOT-22 variable, which means that participants with SNOT-22 scores of < 20 at baseline have limited scope for improvement. 96 Baseline SNOT-22 scores ranged from 6 to 104 (see Table 5), with 75% of scores being ≥ 27 and hence well above the impact of the floor effect.

A STEPP analysis of SNOT-22 scores at 6 months by DOASS (worse side at baseline) did not show a reportable trend between the two variables (see Appendix 1, Figure 32).

Analysis by gender

A subgroup analysis of the primary outcome by gender was performed. This is reported using paired data to assess individual changes in SNOT-22 scores from baseline to 6 months, and baseline to 12 months (presented in Appendix 1, Table 59; see Figure 33 for corresponding box plots). As part of the subgroup analysis by gender, STEPPs were produced separately for males and females [see Appendix 1, Figures 34 (male) and 35 (female)]. No major differences between gender were observed.

Adverse events

A total of 227 AEs were reported across 123 unique participants. Differentiated by arm, this equates to 95 AEs reported by 62 unique participants in the medical management arm and 132 AEs reported by 61 unique participants in the septoplasty arm. The severity of reported AEs is tabulated in Appendix 1, Table 60. The action taken to address the AEs is tabulated in Appendix 1, Table 61, which shows that 61% of AEs required no action, 11% required that treatment be interrupted or discontinued and 28% were treated with concomitant medications. The status of the AEs (ongoing or resolved) at the end of the trial is tabulated in Appendix 1, Table 62. This table shows that 66% of AEs were categorised as resolved, 26% were categorised as ongoing and the other 8% were categorised as ‘resolving’. Appendix 1, Table 63, summarises causality and severity for the AEs by randomised arm.

Three participants randomised to septoplasty had an AE date after the end of follow-up in the ITT population. Two were within the compliance window for the 12-month visit and one was beyond the compliance window (see Appendix 1, Table 64).

AE start dates were not recorded for six participants; however, their AE completion dates are provided (see Appendix 1, Table 65). Line listings of the 227 AEs for all participants can be found in Appendix 1, Table 66. As the AE event terms were reported as free text in the database, these subsequently underwent Medical Dictionary for Regulatory Activities (MedDRA) coding. Table 11 presents a summary of the AEs, categorised according to approximate lowest-level MedDRA term.

Serious adverse events

There were 23 SAEs reported by 16 unique participants. Differentiated by arm, this equates to nine SAEs reported by five unique participants in the medical management arm and 14 SAEs by 11 unique participants in the septoplasty arm.

Serious AE causality and severity by randomised arm is summarised in Appendix 1, Table 67; Appendix 1, Table 68, groups the SAEs by both category and severity. Figure 17 shows the timings of SAEs; many of these correspond with the timing of septoplasty.

FIGURE 17.

Time from randomisation to SAE start date.

Table 12 shows the SAEs recategorised by the clinical team. A line listing of the 23 SAEs is presented in Appendix 1, Table 69. The median time from septoplasty to SAE is 0 days and the maximum time is 8 days.

There were four reported SUSARS: four mild events in the septoplasty arm that were related to septoplasty. These are shown in the last column in Appendix 1, Table 69.

Surgical complication/failure and reintervention within 12 months

Participants allocated to either trial arm who received trial or non-trial septoplasty at any point during the trial were asked to report AEs at 6 and 12 months; 174 participants responded. The numbers of participants who responded (yes/no) to the specific questions about complications at either or both of these time points are summarised in Table 13. Participants could also report ‘other’ complications. These are listed in Appendix 1, Table 70. Complications that could not be coded in specific categories in Table 13 were classified as ‘other’ complications.

The ‘other surgery complications’ were assessed by the chief investigator and a TMG clinician. Three were recategorised and are included in Table 13 in the relevant categories. Nine of these were not deemed to be complications and are not included in Table 13, leaving seven other complications which are listed in Appendix 1, Box 1. One complication is related to adhesions, and so is included in Table 13. Total unique participant numbers vary owing to incomplete data collection. Table 13 also summarises the total reports of adhesions and perforations that occurred by randomised arm. One adhesion and two perforations were reported as AEs, but were also listed in the clinical exam electronic case report form (eCRF).

Introduction

The question addressed by the within-trial analysis was as follows: for adults experiencing nasal obstruction associated with a deviated septum, what is the cost-effectiveness of surgical management compared with medical management at 12 months?

The within-trial analysis included a cost-effectiveness analysis (CEA) and a cost–utility analysis (CUA). The CEA estimated the incremental cost per improvement in SNOT-22 score (defined as a change of ≥ 9 points, as outlined in Chapter 2, Sample size calculation) and the incremental cost per AE avoided. The CUA estimated the incremental cost per QALY gained. For the CUA, responses to the SF-36 were converted into SF-6D utility scores using standard algorithms; these were then used to estimate QALYs82 using the area under the curve approach. 83 As costs and effects were reported for 12 months, they were not discounted in the within-trial analysis.

For the within-trial analysis, the following outcomes are reported:

• NHS/PSS costs of managing individuals with nasal airway obstruction associated with a deviated septum

• direct and indirect costs to participants from nasal airway obstruction associated with a deviated septum

• changes in the SNOT-22 score at 6 and 12 months post randomisation l total number of AEs at 12 months post randomisation

• QALYs based on responses to the SF-36, administered at baseline and at 6 and 12 months post randomisation

• incremental cost-effectiveness ratios (ICERs) –

  • incremental cost per improvement (of ≥ 9 points) in SNOT-22 score

  • incremental cost per AE avoided

  • incremental cost per QALY gained.

Methods

This analysis was designed and conducted according to best practice, conforming to the Consolidated Health Economic Evaluation Reporting Standards. 97 All analyses were undertaken using Stata version 16. All economic analyses were based on an ITT principle.

Estimation of effects

Three effectiveness measures were used in this economic evaluation: nasal function measured by SNOT-22 score, number of AEs, and QALYs based on responses to the SF-36.

Estimation of health outcomes for the cost-effectiveness analysis

Two different outcome measures were used in the CEA:

1. improvement in SNOT-22 score at 6 and 12 months post randomisation

2. number of AEs.

First, we used nasal function, measured by the SNOT-22 administered at baseline and at 6 and 12 months post randomisation, as a measure of effectiveness. A clinically significant change in SNOT-22 scores was defined as a difference of ≥ 9 points (see Chapter 2, Statistical considerations). The individual SNOT-22 scores were derived by the statistical team as previously outlined (see Chapter 2, Analysis of the primary outcome). The change in SNOT-22 score between baseline and 12 months was estimated for every participant and presented as the average total proportion of participants who had an improvement in SNOT-22 scores (i.e. a difference of ≥ 9 points) per arm. This analysis was replicated to identify the difference in SNOT-22 scores between baseline and 6 months for every participant.

The second measure of effectiveness was the number of AEs reported in each arm. 110,111 The total number of AEs was aggregated for each participant and presented as the average total number of AEs per arm.

Estimation of quality-adjusted life-years for the cost–utility analysis

The SF-36 was administered at baseline and at 6 and 12 months post randomisation. The SF-36 is a practical, reliable and valid measure of physical and mental health. 112 The responses to the SF-36 were converted into the SF-6D, a preference-based utility index, using a standard algorithm to produce a health-state utility score. 82 The SF-6D comprises six multilevel dimensions: physical functioning, role limitations, social functioning, bodily pain, mental health and vitality. The area under the curve approach was used to assign time weighting to the utility scores. The time-weighted average of the scores based on the responses to the SF-36 throughout the follow-up period allowed us to generate QALY values for each participant. Equation 1 is an illustrative example of how QALYs were estimated:

⁽¹⁾ Q A L Y = ( u t i l i t y b a s e l i n e + u t i l i t y 6 m o n t h s ) ÷ [ 2 × 6 / 12 ] + ( u t i l i t y 6 m o n t h s + u t i l i t y 12 m o n t h s ) / [ 2 × 6 / 12 ] .

Comparative incremental analyses of costs and outcomes between trial arms

Unadjusted and adjusted (regression) analyses were performed to estimate the cost-effectiveness of septoplasty compared with medical management. All results were presented as point estimates of the mean incremental costs, effects and cost-effectiveness.

The cost-effectiveness plane (Figure 18) illustrates how a decision is made based on the economic results. The difference in QALYs is shown on the x-axis and the difference in costs is shown on the y-axis. At the origin there is no difference in costs and effects between the two management strategies. If septoplasty was found to be both less costly and more effective (south-east quadrant), it would be dominant and would be considered cost-effective. If septoplasty was found to be both more costly and less effective (north-west quadrant), medical management would be dominant and septoplasty would not be considered cost-effective. If septoplasty was found to be more effective, but more costly, or less effective and less costly (north-east and south-west quadrants, respectively), then we would have to consider which management strategy is more likely to be cost-effective. In this situation, decisions were based on the ICER, which estimates the cost per additional unit of effect (i.e. the difference in costs divided by the difference in effects between the two arms). In this situation the ICER is compared with the threshold value society places on an additional unit of effect, if there is one available (the additional unit of effect is illustrated by ‘a’, the dotted line in Figure 18).

FIGURE 18.

Cost-effectiveness plane. a, Society’s threshold to pay for one more unit of effect.

Results

Economic evaluation

Incremental cost-effectiveness analysis

Table 16 presents the unadjusted average total costs and average total proportion of participants who reported an improvement in nasal function measured using the SNOT-22 per randomised arm at 12 months. On average, septoplasty was more costly and more effective than medical management. Adjusted analyses were used to estimate incremental costs and effects; these estimates were used to estimate the ICER. The incremental cost for an improvement of ≥ 9 points in SNOT-22 score per participant was £4855. As the threshold value placed on an improvement in SNOT-22 score increases, so does the probability of septoplasty being considered cost-effective.

TABLE 16 - Incremental cost per improvement in SNOT-22 score at 12 months

Point estimates are based on the unadjusted analysis (costs, n = 307, effects, n = 244).

Incremental estimates are based on the adjusted analysis (n = 243).

Investigation strategy	Cost (95% CI) (£)a	Incremental cost (95% CI) (£)b	Effect (95% CI)a	Incremental effect (95% CI)b	ICER (£)	Probability that septoplasty is cost-effective for different threshold values for society’s willingness to pay for an improvement in SNOT-22 scores
						£0	£500	£1000	£3000	£5000
Outcome: SNOT-22 score – results
Medical management (costs, n = 155; outcomes, n = 125)	973 (810 to 1137)	1306 (1124 to 1489)	0.504 (0.42 to 0.59)	0.269 (0.16 to 0.38)	4855	1.00	1.00	1.00	1.00	0.45
Septoplasty (costs, n = 152; outcomes, n = 119)	2162 (2102 to 2222)		0.756 (0.68 to 0.83)			0.00	0.00	0.00	0.00	0.55

The results of the stochastic sensitivity analysis are presented in Figures 19 and 20. Figure 19 is the cost-effectiveness plane; for all of the bootstrapped iterations, septoplasty was more costly and more effective than medical management. Figure 20 is the CEAC and shows that, as the value placed on an additional participant’s SNOT-22 score improving by ≥ 9 points in SNOT-22 score increases, so does the probability of septoplasty being considered cost-effective.

FIGURE 19.

Cost-effectiveness plane for septoplasty vs. medical management using the adjusted bootstrapped CEA SNOT-22 score at 12 months (multiple imputation results).

FIGURE 20.

The CEAC for septoplasty vs. medical management using the adjusted bootstrapped CEA SNOT-22 score at 12 months (multiple imputation results).

Introduction

The within-trial results are useful to inform decisions about the cost-effectiveness of septoplasty in the short term (12 months). However, they provide limited information on the costs and benefits associated with septoplasty over the longer term. This is an important limitation of the within-trial analysis. We anticipated that the surgical arm could be more costly and potentially more effective, but the time horizon of the trial would not be sufficient for the additional benefit to offset the additional costs. This is illustrated in the analyses presented above when we compare the incremental cost per QALY for septoplasty versus medical management at 6 months’ follow-up (>£100,000) with that at 12 months (£27,114). To address this, we conducted a further model-based analysis. The question being addressed by the model-based analysis was as follows:

For adults suffering with nasal obstruction associated with deviated septum, if costs and effects were extrapolated beyond the 12-month follow-up period, at what point, if any, does surgical management have the higher probability of being considered cost-effective compared with medical management?

For the model-based analysis, the following outcomes are reported:

• NHS/PSS costs of managing individuals with nasal airway obstruction associated with a deviated septum (including septoplasty) over a longer time horizon

• QALYs over a longer time horizon

• incremental cost per QALY gained.

Methods

An economic model was used to extrapolate the trial findings and estimate the cost-effectiveness of septoplasty, compared with medical management, at 24 and 36 months post randomisation. The model design and parameters used are described in this section.

Economic model

Model structure

A simple decision tree model was developed using TreeAge (TreeAge Software, Inc., Williamstown, MA, USA) to extrapolate costs and QALYs beyond the 12-month trial follow-up period. The model was designed to replicate the pathway of the trial participants. Figure 21 is an illustration of the model pathway. Data were obtained from the trial to estimate the relative costs and utilities of septoplasty, compared with medical management (see Incremental cost–utility analysis). Costs, QALYs and cost-effectiveness were estimated at 24 and 36 months.

FIGURE 21.

Model structure.

Model assumptions

The model was designed to incorporate the relative differences between septoplasty and medical management identified from the within-trial analysis.

The following assumptions were made for the economic model:

• The model took the perspectives of the UK NHS and PSS.

• Costs and utility data used in the model were based on the multiple imputation data.

• Both arms of the model (medical management and septoplasty) were equivalent at the start of the model (i.e. it was assumed that costs and utilities reported by the medical management arm throughout the 12-month trial follow-up were the base for both arms).

• The utility and healthcare resource costs for those who went on to have surgery were assumed to be equivalent to the medical management arm plus an additional adjustment was made based on the trial data (see Costs and Utilities for further information).

• For extrapolation, the cost and utility values reported after 12 months were assumed to be the same as those reported in the last 6 months of the trial (i.e. between 6 and 12 months).

• Costs and utilities at 24 and 36 months were discounted at 3.5%. 113

• Assumptions concerning further treatments (septoplasty and nasal sprays) post 12 months were based on clinical guidance and explored in sensitivity analyses.

• Given the short time horizon of the model, the average age of trial participants (40 years) and the trial data, it was assumed that no participants died over the 36-month time horizon.

• It was assumed, given the limited data available, that those who had surgery would not need a revision surgery.

• All model parameters were defined as statistical distributions in the model, and distributional assumptions were based on trial data.

• Alternative assumptions were explored in sensitivity analysis (see Sensitivity analysis below).

Model parameters

The model parameters were informed using trial data and the distribution of each parameter using the mean, SD and shape of the distribution. Table 18 summarises all of the parameters used in the model.

TABLE 18 - Model parameters

Model parameters	Mean (SD)	Distribution
Costs
Nasal spray	91 (–)	No assumptions made on distribution as this was a fixed cost
Surgery	1956 (–)	No assumptions made on distribution as this was a fixed cost
Healthcare resource use costs at 6 months (multiply imputed)	138 (235)	Gamma
Healthcare resource use costs at 12 months (multiplyimputed)	148 (156)	Gamma
Healthcare resource use costs adjustment at 6 months for those in the septoplasty arm who had surgery	21 (23)	Gamma
Healthcare resource use costs adjustment at 12 months for those in the either arm who had surgery	−61 (18)	Gamma
Utilities
Baseline	0.720 (0.14)	Beta
Utility at 6 months (multiply imputed)	0.728 (0.14)	Beta
Utility at 12 months (multiply imputed)	0.735 (0.15)	Beta
Utility adjustment at 6 months for those who had surgery in the septoplasty arm	0.063 (0.012)	Beta
Utility adjustment at 12 months for those in either arm who had surgery	0.068 (0.14)	Beta
Transition probabilities
Probability of having surgery (medical management) at 12 months	0.30	No assumptions made on distribution as based on trial data
Probability of having surgery (medical management arm only) at 24 months	0.15	No assumptions made on distribution as based on clinical advice, but explored in sensitivity analyses
Probability of having surgery (medical management arm only) at 36 months	0.075	No assumptions made on distribution as based on clinical advice, but explored in sensitivity analyses
Probability of having medical management at 24 and 36 months (medical management arm only)	0.50	No assumptions made on distribution as based on clinical advice, but explored in sensitivity analyses
Probability of having surgery (septoplasty arm only) at 12 months	0.97	No assumptions made on distribution as based on trial data

Costs

The costs and cost adjustments for surgery used in the model are reported in Table 18. Intervention costs (surgery and/or nasal sprays) were assumed to be as reported in the trial and are assigned in the model depending on the pathway (i.e. nasal spray costs were assigned to the medical management arm only, the NHS tariff for septoplasty was assigned to those who went on to have surgery regardless of their randomised allocation). After 12 months it was assumed that there would be a demand for further treatments by those randomised to medical management. Those who went on to have surgery between 12 and 36 months were assigned the NHS tariff for septoplasty and those who continued with management received nasal spray costs for the remainder of the model.

We assumed that healthcare resource use costs would be equivalent at the start of the model and adjustments were made depending on whether or not surgery had been performed. The adjustment in costs at 6 and 12 months was based on an ordinary least squares regression of the multiply imputed healthcare resource use costs at 6 and 12 months. The same covariates used in the SUR (randomised arm, age, gender, ethnicity and baseline utility scores) were used. At 6 months, those randomised to septoplasty reported slightly higher healthcare resource use costs, but at 12 months they reported lower costs than those in the medical management arm (see Table 18). We assumed that healthcare resource use costs incurred after 12 months, assigned at 6 monthly intervals, were equivalent to those reported during the last 6 months of the trial (i.e. 12-month costs). Adjustments to healthcare resource use costs associated with surgery were assigned depending on when during the 36 months surgery was performed.

All costs (treatment and healthcare resource use) incurred after 12 months were discounted at 3.5%.

Utilities

The utility values and utility adjustments used to estimate QALYs in the model are reported in Table 19. Similar to costs, the multiply imputed utility data reported by the medical management arm throughout the 12-month trial follow-up were assumed to be the base and adjustments were made if surgery was performed. Similar to the cost adjustments, utility adjustments were estimated using an ordinary least squares regression controlling for the same covariates as the SUR (randomised arm, age, gender, ethnicity and baseline utility scores). At both 6 and 12 months, those randomised to septoplasty reported higher utility values; this adjustment was applied to the QALY equation depending on when septoplasty was performed. Similar to costs, utilities were assigned at 6-monthly intervals (18, 24, 30 and 36 months) and those incurred post 12 months were discounted at 3.5%.

TABLE 19 - Incremental cost per QALY gained at 24 and 36 months (model results)

Investigation strategy	Cost (£)	Incremental cost (£)	Effect	Incremental effect	ICER (£)	Probability that septoplasty is cost- effective for different threshold values for society’s willingness to pay for an additional QALY
						£0	£10,000	£20,000	£30,000	£50,000
Outcome: QALYs at 24 months – results
Medical management	1483	833	1.46	0.06	13,221	1.0	0.97	0.01	0.00	0.00
Septoplasty	2316		1.53			0.0	0.03	0.99	1.00	1.00
Outcome: QALYs at 36 months – results
Medical management	1785	703	2.20	0.10	7368	1.0	0.05	0.00	0.00	0.00
Septoplasty	2488		2.29			0.00	0.95	1.00	1.00	1.00

Transition probabilities

Transition probabilities in the model were based on the probability of those randomised to each arm undergoing septoplasty; these probabilities are detailed in Table 18. As expected, a higher proportion of those randomised to septoplasty than to medical management underwent surgery (97% vs. 30%, respectively). We assumed in the base-case analysis that no additional surgeries were undertaken at 24 or 36 months for those randomised to septoplasty. Based on clinical advice, we also assumed that there would be a steady decline in the uptake rate of septoplasty for those in the medical management arm after 12 months, and that a proportion of those randomised to medical management would recommence using nasal sprays to manage their deviated septum. These assumptions were explored in sensitivity analyses that varied the transition probabilities by ± 50%.

Model validation

We internally validated the model by checking the model structure, calculations and data parameters. 121 We undertook further validation of the model by running it for 12 months, to replicate the results of the within-trial analysis. Comparing these results allowed any potential discrepancies in the model parameters to be identified.

Results

The model results at 24 and 36 months are presented in this section.

Methods in the pre-recruitment phase: pre-emptive support

Anticipated sources of recruitment difficulty were addressed by:

• critically reviewing patient-facing materials

• designing and delivering training to raise awareness of common recruitment issues and solutions

• consolidating key messages from training in a written ‘tips’ document.

Methods in recruitment phase: phase 1 and phase 2 of the QuinteT Recruitment Intervention

Methods to investigate actual (rather than anticipated) recruitment processes in the NAIROS are outlined below (phase 1), followed by a summary of how these informed responsive actions to optimise recruitment (phase 2).

Pre-emptive strategies to optimise recruitment

Given that the NAIROS team anticipated that patient preferences/expectations for surgery would be an obstacle to recruitment, key issues considered at the outset of the RCT centred around conveying equipoise and exploring treatment preferences, to ensure that patients were fully informed. These topics were covered in training materials and considered in patient-facing refinements to recruitment materials (e.g. PISs and videos), as summarised below.

Phase 1: QuinteT Recruitment Intervention findings – understanding recruitment in practice

The first site opened to recruitment 5 months later than planned, in January 2018, owing to governance and regulatory approval delays. Ten sites were projected to be open by this point. With recruitment already behind schedule, rapid understanding of barriers to patient accrual was critical in the remaining 7 months of the funded QRI.

All phase 1 QRI data collection (apart from two preliminary interviews with TMG members which were conducted before began) was focused on actual recruitment interactions and experiences. QRI data collection in phase 1 comprised 19 interviews, 108 audio-recorded consultations and regular scrutiny of screening logs. Findings from these sources were triangulated to build an in-depth understanding of recruitment processes across sites. Findings from the interviews and audio-recorded consultation analysis are presented separately, with cross-references to other data sources when relevant.

Interview findings

Sample of interview informants and presentation of findings

A total of 19 recruiters, from 11 of the 15 sites open to recruitment at the time, participated in interviews between May 2017 and August 2018. Of these 19 informants, three were research nurses and 16 were consultant ENT surgeons. Most informants (17/19) had experience of recruiting to the NAIROS as their sites had been open for 2–5 months at the time of interview. Two interviews were conducted before recruitment began with individuals who held dual roles as recruiters and TMG members. Thirteen of the 19 interviews were semistructured and directed by the QRI topic guide. Six were structured discussions about sites’ recruitment pathways.

Findings from the interviews have been presented according to three topics: (1) perceptions of equipoise and the need for the NAIROS, (2) organisation of recruitment at the site level and (3) experiences of discussing the trial with patients. Illustrative anonymised quotations have been presented throughout, with careful attention to present negative cases when relevant. Identifiers for quotations show the following:

• the interview participant’s number (‘R’ = ‘recruiter’)

• site number, specific to the QRI study (note that a QRI-specific site numbering system was used that was different from the wider trial’s site numbering; this was partly because only a subset of sites took part in the QRI study. The QRI-specific numbering system also preserves sites’ anonymity)

• role of the individual [research nurse or surgeon, with an indication of which surgeons were also the principal investigator (PI) for the site].

Perceptions of equipoise and the need for the Nasal AIRway Obstruction Study

Recruiters’ perceptions of equipoise were explored in the early stages of the QRI to provide context for understanding recruitment issues, but these were also important topics covered in the trial’s qualitative process evaluation (see Chapter 6). To avoid replication, detailed insights around recruiters’ perceptions of the trial question and eligibility criteria are reported in the context of the process evaluation, which builds on some of the data collected for the QRI. In terms of optimising recruitment, the below findings (relating to this part of the QRI topic guide) were key:

• Surgeons and research nurses indicated support and commitment to recruiting to the NAIROS, acknowledging a need for evidence to inform policy and practice around management of nasal obstruction. Several informants (all surgeons) alluded to the NAIROS’s significance to commissioners and policy-makers, as shown through comments that there was an increasing expectation or ‘demand’ (R9, site 5) to produce evidence to justify continuation of existing practices:

I think it’s a really important subject. Politically, it’s important […], where there are increasing pressures to give good evidence that what we’re doing, optimally, for patients is the right thing and is going to be a proven benefit for them.

R15 (surgeon PI, site 7)

• Most surgeons anticipated that the NAIROS would address clinical uncertainties around which patients would benefit the most from septoplasty. In alignment with this, most surgeons anticipated that the NAIROS would show a relationship between degree of septal deviation and benefit:

I think it will show that, in patients who have significant symptoms, and a significantly deviated nasal septum, that there are benefits to be had from surgery. So I think it will be very much based on our baseline stratification.

R11 (surgeon, site 6)

• Most informants expressed that they were comfortable with and willing to approach the full spectrum of eligible patients for the NAIROS, despite some surgeons acknowledging that this required them to over-ride instincts about which patients would benefit from septoplasty. Two surgeons raised the prospect of not approaching patients whom they felt would benefit from a surgical procedure outside the NAIROS protocol. Based on interviews/screening log data, it was not possible to gauge the extent to which this quantitatively affected recruitment.

These findings provided insight into clinical professionals’ vantage points as they embarked on recruiting to the NAIROS (see Chapter 6 for more insights about eligibility assessment).

Organisation of recruitment at the site level

Recruiters’ interview accounts revealed variable ways in which the NAIROS had been integrated within routine practice, which had potential to support or hinder recruitment, specifically by having implications for the identification of potentially eligible patients (referred to as ‘NAIROS candidates’ hereafter).

Identifying Nasal AIRway Obstruction Study candidates and ascertaining eligibility Initially, most (and eventually all) sites relied on dedicated research clinics, to conduct eligibility assessments, consent discussions, randomisation and baseline assessments, given the difficulties of incorporating these tasks into time-limited routine clinics:

The first time we did it, it took us 2 and a half hours, and that is purely because it was the first time we were doing it. You need to cater for the patients needing to go through all that. So, it’s not really feasible in a clinic setting.

R20 (surgeon PI, site 10)

Most sites were able to accommodate research clinics into clinical schedules, but their frequency varied, ranging from once a week to once a month. Therefore, although NAIROS clinics were necessary, their frequency and the limited numbers of patients they could accommodate constrained recruitment:

… our clinics are full every month. We don’t have scope at the moment to add more patients in […], room space is a problem. We have a room and the consultant has a room, so if we’re both recruiting different patients, even if we stagger patients, we would still need an extra room.

RN7 (research nurse, site 3)

Although most sites had managed to set up research clinics, identifying patients to invite to these clinics was a commonly reported challenge across sites. Research nurses and surgeons described two approaches to identifying NAIROS candidates, as shown in Figure 22.

FIGURE 22.

Approaches to identifying patients potentially eligible for the NAIROS from routine NHS care.

The first method involved screening referral letters accessed through ‘choose and book’ (described below) and written referral letters. Site PIs (all of whom were surgeon consultants) tended to triage patients referred to their own clinics, with further involvement from research nurses and other consultants depending on the site. Irrespective of who and how many individuals were involved in screening, identifying patients based on referral letters alone had its limitations. According to several recruiters, letters did not contain the level of detail needed to gauge NAIROS suitability, resulting in those screening erring towards a more inclusive approach. As a result, the limited capacity research clinics could be filled with patients who were not even suitable for eligibility testing:

We’ve found it quite difficult, I think, to assess patients adequately enough from the GP letters, so it’s difficult to know whether they’re even a suitable patient until you bring them into the NAIROS [clinic]. For example, the first patient that I saw in the clinic had other problems with their nose that meant that they were ineligible to meet the eligibility criteria.

R15 (surgeon PI, site 7)

Some sites were able to address these issues with support from research nurses, who were able to conduct additional investigations of patient records to better gauge suitability, but this support was not universally available.

Another limitation of identification through letters was the transition to electronic ‘choose and book’ processes in some sites, whereby patients select their own appointment in routine clinics. Informants from one site explained that as a result of these ‘choose and book’ proceses, patients had often already arranged their routine care appointment when they were invited to attend a NAIROS clinic:

Patients, when they get their information pack, are usually halfway along the journey.

R1 (surgeon PI, site 1)

Considering the above difficulties, most sites either moved towards relying on consultants to identify patients who had been referred to their routine clinics (see the pathway on the right in Figure 22), or reportedly adopted this approach from the outset. According to several informants, consultant-led identification in clinics had the benefit of greater precision in referring suitable candidates to NAIROS clinics:

… if we draw our patients from a pool that have already been already seen in the main clinics and are already deemed as potentially suitable for septal surgery or a trial of medical treatment, I think that’s probably the way forward. Otherwise, I think you end up spending a lot of time seeing patients that, probably, will never have been eligible to be included in the trial in the first place.

R15 (surgeon PI, site 7)

‘In-clinic’ identification was operationalised in different ways across sites, with each model involving consultants who did not hold formal recruitment roles. In some sites, consultants examined patients in routine clinics and referred potentially eligible candidates to research clinics. In others, consultants triaged letters and re-routed patients with a septal deviation to a research nurse or the routine clinic of the NAIROS PI, who would then invite suitable candidates to research clinics:

I’m the main recruiting clinician. […] We triage our inpatients every week. So different consultants would triage patients and then they will identify potential patients who might fit the NAIROS criteria and let me know about them.

R9 (surgeon PI, site 5)

The success of in-person identification of NAIROS candidates hinged on engagement from clinical staff who were not directly involved in the NAIROS. Informants varied in their impressions of how engaged their colleagues had been, but several emphasised the importance of regular promotion of the NAIROS, through presentations and departmental meetings:

Because I think the thing about recruitment, you’ve got to remind people all the time, yes? Putting just a poster is not enough. […] Yes, so I’m seeing them – I think it’s the [date] is our [meeting name] – and I will remind everybody. That’s the whole department there, which is good. So all the consultants, all the registrars, all the doctors.

R18 (surgeon PI, site 9)

Supporting non-Nasal AIRway Obstruction Study consultants to introduce the trial There was some indication from interviews that recruitment to the NAIROS was being constrained by patients declining invitations to attend research clinics:

So, for every one that considers it, I think there are at least two who would say, ‘No, no’. So, round about 40–50%. It is high.

R20 (surgeon PI, site 10)

The original NAIROS screening log did not capture events preceding eligibility assessments, and thus the proportion of patients lost in the pathway before hearing the ‘full explanation’ of the trial was not formally documented. Interviews did, however, highlight the ENT consultants’ pivotal role in introducing the trial, in addition to their role in identifying candidates:

… well, usually it’s not easy to find someone that is a proper candidate for the study, for different reasons. But even when you do that, sometimes it’s not easy seeing the patient in 15 minutes of an appointment to easily introduce them to the study or to have enough time to do that. I believe this is not only for me, it’s for everyone, because we try to ask the help of other clinics so we can get someone that is a candidate. I mean, it doesn’t need to be too long or too much to introduce the patient at the start, but sometimes it’s not easy to get there.

R19 (surgeon, site 9)

Taken together, these accounts reinforced the idea that NAIROS recruitment relied on events and processes that preceded the formal ‘recruitment setting’ outlined in the protocol. Engagement from non-recruiting clinicians in identifying patients and providing high-level introductions to the trial appeared to be key to recruitment success.

Recruiters’ experiences of discussing the trial with patients

Explaining the NAIROS to patients was a key element of the recruitment process. The sections that follow are based on recruiters’ interview accounts, and draw on their experiences of explaining the trial.

Informants from sites that had not yet opened to recruitment anticipated that patient expectations and preferences for surgery would be a key barrier to recruitment. Those with recruitment experience (at the time of interview) also reflected back on their belief that patients would expect or prefer septoplasty. This belief stemmed from their assumption that many patients would have already received steroid sprays in primary care:

I think the majority will probably come, or the majority who have a preference will have a preference towards surgery. I think that’s possibly culturally, and traditionally, because in the UK, unlike a lot of other healthcare systems, you have a general practice where you have a sort of gatekeeping role. So, you have quite a lot of treatment done in general practice, and they will have had treatment in general practice, and so I think quite a few of them will, when you are referred to hospital, sometimes it’s almost with an expectation you’re going to have an operation.

R11 (surgeon, site 6)

Recruiters’ descriptions of their actual patient encounters suggested that the above concerns had not materialised. Surgeons and research nurses from different sites commented that recruitment had proceeded better than anticipated, as evident through unprompted remarks about how patients’ reactions had surprised them:

Yes, I think we are meeting our targets so far. I’m certainly surprised. I thought some patients would say no, but I think just about every patient we’ve approached has said yes so far.

R4 (surgeon PI, site 2)

In contrast to concerns about patients holding strong preferences/expectations for surgery, several recruiters discussed how the uncertainty around surgery versus medical management had resonated with patients:

I feel confident and there are some patients I’ve come across who genuinely don’t know which way they would like to go – either medical therapy or surgical therapy – and they’ve been delighted that that’s been taken out of their hand with the trial.

R9 (surgeon PI, site 5)

Certainly the feedback and the impression I’m getting from the patients we talk to, unless they’ve already definitely made up their minds which they want to try, which we’ve had a couple of … otherwise they seem quite open to it, the question itself, you know? A lot of them are maybe asking themselves, ‘What is best? Should I take medication or should I have surgery?’.

R6 (research nurse, site 3)

Preferences were, nonetheless, still raised by several recruiters when asked about the main barriers to recruitment. When discussed, references to preferences were either tempered with the positive accounts of recruitment (as per the previous quotation from R6), or framed as a reason why some patients had declined participation:

Interviewer: From your perspective, what were the facilitators and the barriers to getting people engaged in the study?

R14 (Research nurse, site 6): I think it is people who are determined that they want surgery. That’s the biggest barrier. I think there has been three or four who quite clearly, from the outset, they maybe have reasonable reasons and rationale for wanting surgery. Again, that is something that, just through the way it’s gone, has changed my mind. I thought NAIROS would struggle to recruit […]. I thought that people would, once they got as far as this, would be expecting surgery and they would reject the ideas of the sprays because they would have tried that. So I thought it would be quite difficult to recruit to, but just from our first few clinics and first few patients, I can see that the questions for patients are much wider and they’re not as straightforward.

We may not have regarded preferences as a significant obstacle to recruitment as most recruiters expressed confidence in engaging with patients’ preferences. There were some exceptions to this (n = 3, all research nurses), whereby informants framed preferences as a clear-cut signal to stop discussing the trial, as further discussion would be futile or inappropriate. Notably, two of these individuals had not been involved in the NAIROS from its outset and had not attended the pre-recruitment training. By contrast, all surgeons (including site PIs) indicated that they were comfortable exploring preferences, particularly if these appeared to be based on expectations for surgery:

R1 (surgeon, site 1): Most of them come with an expectation to have surgery.

Interviewer: How do you manage that?

R1 (surgeon, site 1): I tell them what the evidence is and why we conduct trials – to help decide what the best way to manage patients with this is. I help them to understand that if it’s an anatomical obstruction, where the nasal septum is slightly off to one side and causing an obstruction that, actually, will have been there for most of their lives, but often their symptoms are only recent and so it’s something else that’s changed. That’s where equipoise comes in, in that it’s a good research question.

The one informant who framed preferences as a dominant issue discussed this in the context of inviting patients to attend research clinics. This informant estimated that around half of patients had declined the invitation [see the quotation from R20 (surgeon PI, site 10) in Supporting non-Nasal AIRway Obstruction Study consultants to introduce the trial]; however, none of the other informants discussed these challenges.

In summary, interviews with the NAIROS staff provided an opportunity for staff to reflect on their interactions with patients. This was, for most, a positive experience. Nonetheless, these positive accounts must be considered alongside the fact that discussions in the research clinic are unlikely to reflect the views of those patients less willing to consider participating in the NAIROS. There is also a limit as to what can be confidently gauged from people’s accounts of their interactions with others, especially when reflecting on past events. However, it was clear that many recruiters were surprised about how patients had engaged with the trial, despite the existence of preferences and expectations for surgery. It was also clear that most recruiters felt equipped to engage with preferences and expectations. The final section presents findings from the analysis of audio-recorded recruitment conversations, to provide insight into how the trial was conveyed by recruiters and received by patients.

Phase 2: actions to optimise recruitment

The findings from the interviews and audio-recorded consultations were initially fed back to the chief investigator via a report in August 2018 (month 8 of recruitment) and circulated to the TMG in September 2018. The report informed a series of actions which are summarised in the following sections.

Limitations of the QuinteT Recruitment Intervention

Although it was possible to discern key recruitment issues and deploy strategies to address these through the QRI, there were limitations around the reach and depth of the methods. The sample of recruiters who agreed to audio-record consultations and participate in an interview were self-selected, and not all recruiters/centres took part. Therefore, a full picture of recruitment across all sites was not possible, potentially hindering opportunities for further bespoke recruitment support. The QRI findings are also limited by snapshots of recruitment practices captured through select audio-recorded consultations; as a result, it was not possible to capture what patients had been told before or after the recorded encounter, which, in turn, limited potential to fully understand how the recruitment pathway (and interactions throughout) may have shaped decisions about trial participation. Although resources did not permit this in the NAIROS, case studies of a sample of eligible NAIROS patients’ experiences as they move through clinics and interact with different personnel would have provided more in-depth and comprehensive insights into how recruitment plays out in practice. The process evaluation, presented in the next chapter, does, however, provide insight into patients’ experiences of the trial, and provided opportunities for triangulation through regular meetings and sharing of emerging findings between the QRI and process evaluation researchers.

Design

The process evaluation was designed as a qualitative study comprising in-depth interviews with patients and staff (surgeons and research nurses) involved in the NAIROS trial and a thematic analysis informed by normalisation process theory (NPT). 136,137 NPT has been widely applied in understanding the implementation of complex interventions in health care. 137 In the context of the NAIROS, the trial itself (and all associated processes, including those under evaluation) was identified as the complex intervention. NPT incorporates four constructs: coherence (how people make sense of the trial), cognitive participation (whether or not people are willing and able to buy in to implementing the trial), collective action (people’s ability to take on the work needed to implement the trial) and reflexive monitoring (people’s reflection on the benefits and costs of the trial). 136

Recruitment and sampling

Data collection

Analysis of the interviews

Anticipated impacts of the Nasal AIRway Obstruction Study trial

Views of the NAIROS research question were relevant to both patient and staff motivation to participate in the trial. For staff, this related to how valuable they felt the evidence that a trial would generate might be, whereas patients were more interested in how the NAIROS research question resonated with their own understandings and expectations of the causes of, and treatment for, their nasal obstruction.

A lack of high-quality evidence supporting the use of septoplasty and the uncertainty of its future appeared to motivate some surgeons to take part in the trial:

Yes, if we are honest, you know, the evidence base for the septoplasty was almost non-existent … understanding what could benefit them and which patients – having some objective measures to determine suitability was high time.

Surgeon 1

Staff members felt that the research question was appropriate and long overdue in the ENT specialty. They felt it was an important question that needed to be investigated by a robust RCT:

In terms of the question it’s asking, my feeling was, ‘Thank [expletive] somebody is finally doing this properly’. There have been a couple of small studies that have shown some quite interesting results, pointing at patients with post-decongestion unilateral nasal … You know, objective nasal airflow reduction being the patients that benefit. They’re small studies, they weren’t done in a very rigorous way, so it was hard to be clear that that was a thing that needs to happen.

Surgeon 2

I think it’s a very good question, and we don’t have a formal answer for it without having done a randomised controlled trial. We do an awful lot of operations on the NHS, and that’s just from historic, and having the data behind it to say, ‘Actually, this is very effective’ or, ‘This isn’t effective’ or, ‘This treatment is better’ is obviously a great way forward. ENT, like a lot of specialties, we’re starting to get pressurised by the NHS to justify why we do something, what our action is.

Surgeon 3

There was variation as to whether some investigators felt that the aim of the NAIROS was to provide high-quality evidence for the effectiveness of septoplasty, or whether this was a trial to compare which treatment for nasal obstruction was more effective. One research nurse stated that when they originally took part in the NAIROS they felt that the aim of the trial was to confirm the effectiveness of septoplasty. However, after observing that patients returning for their NAIROS review appointment had positive and negative experiences of both treatment arms, they changed their view and instead saw it as a trial comparing two equal treatments:

Interviewer: Did you have any reservations about NAIROS before you started recruiting to it?

Research nurse 1: Yes, because with thinking that it was just to confirm what people thought I, kind of, had the reservation thinking that people who were randomised to the medical arm were not getting the optimum treatment.

Interviewer: Right. Did that change, then, in terms of …?

Research nurse 1: Yes, when people have come back and people have been quite happy with the sprays.

Several surgeons described wanting the results of the NAIROS to clearly demonstrate which patients are going to benefit from surgery and which will benefit from medical management. In turn, these results could lead to a reduction in the number of unnecessary procedures performed:

Interviewer: What would you hope NAIROS to show?

Surgeon 1: Show which patients, in these groups, do best for the surgery and which patients, perhaps, can just be managed medically.

[W]hat I’m hoping comes out of the trial is probably that septoplasty is useful, but it’s useful for certain people and not for everybody, and medical treatment is better for some others.

Surgeon 4

Surgeons also described wanting a more standardised approach to assessing which patients will benefit from septoplasty:

If we find that some of the symptoms’ scores have a good correlation with people that benefit, or the rhinospirometry measurements and stuff correlate well, then that might be quite a useful way of judging who would be a good candidate for the operation and who might not.

Surgeon 4

However, one surgeon was cautious of the idea of using objective measures for septoplasty referral:

I’m not convinced it’s [assessment tools] the only thing that you should rely on.

Surgeon 5

The results from the NAIROS trial also had the potential to demonstrate that septoplasty is no more effective than medical management. One surgeon noted that their colleagues were reluctant to be involved in the trial as they felt that the NAIROS trial could threaten the future of septoplasty:

I think some people felt threatened by the study because they felt, ‘By putting patients in the study, this is all adding to the evidence that’s going to take away septoplasty. We’re not going to be allowed to do it anymore.’

I was trying to explain to them, ‘Well, actually, this is going to produce evidence that might support septoplasty. As far as taking it away, if they’re going to take it away, they’ll take it away anyway. This is our one opportunity to contribute to potentially preserving the operation, if we can show a benefit.’.

Surgeon 6

Overall, the impacts that surgeons anticipated or desired from the NAIROS trial appeared to influence their motivation to participate.

Patients’ motivation to participate in the NAIROS trial was also influenced by their expectations of the impacts of the NAIROS treatment options. Patients’ understandings of the cause of their symptoms (e.g. that they had a clear nasal septum deviation associated with a specific injury) and their perceptions of the likelihood of a successful treatment outcome (drawing on prior experience with medical management and on wider experiences of friends and family) affected their views of the NAIROS research question:

My first reaction, to be honest, I told her that I would like to have an operation done because, in my personal opinion, if the bone is then tilted towards the side, a medicine cannot cure that.

Patient 16

Because I was put in this situation with a traumatic blow to the face, I think, probably, going under the knife was the way to fix it.

Patient 18

I was hoping I wouldn’t be [randomised to surgery], because a friend of mine, he’s just had, basically, the same surgical procedure done, and it’s made no difference to him whatsoever.

Patient 19

However, the majority of patients were uncertain as to what the best treatment for their nose symptoms was, and so regarded the NAIROS research question in a positive light:

I think people think of an operation as a straightforward thing, but nothing is straightforward, really. It’s under a general anaesthetic, it’s the dangers that go with it. And with the success rates, as well, they weren’t 100% that it would work. So, I thought well, rather than going down the route of the operation, why not try the study, using the spray, and see what sort of success I have from that first. And then, because I had the option to go back for an operation, that was always there in the background, if the steroids failed.

Patient 20

Making a decision about surgery

There was considerable complexity and heterogeneity in terms of both patients’ symptoms of nasal obstruction and how clinicians used these symptoms, together with other factors, to make a decision about whether or not to offer surgery.

Staff described a decision-making process that took into account the nature and severity of the symptoms; the length of time these had been experienced; the nature of symptom onset (whether or not it was associated with a particular injury or other event); whether or not symptoms varied seasonally; and the structure of the nose, including the degree of deviation and where the deviation occurred.

Patients presented with a range of different symptoms in clinic, including nasal blockage, trouble sleeping, snoring and frequent nosebleeds (Table 22). Several patients could recall an injury that had damaged or broken their nose, whereas others did not remember such an incident.

Staff members also reported a complex pattern and history of symptoms presented by patients in clinic:

I suppose there are the people that are not quite congenital, but something has happened very early in life and they have grown up with it to some extent. Quite often they don’t complain that much, because they have never known anything different. Sometimes it’s because they have developed a bit of an allergy or something, and maybe the septum is not the whole story.

Then I suppose quite a few of them are traumatic, and that can be people who play rugby or football or things like that, or the younger core who go out on a Friday night and have a fight. There are a few of them around.

Then you probably move on to older people, who I think have probably always had a bit of a septal deviation, but the cartilages have been quite firm. Then they get to 50, 60 or a bit more and everything starts to droop a little bit, and then they have a problem.

Surgeon 4

The heterogeneity across patients seen in clinic and the lack of standardised measurement tools for assessing patients’ septa and symptoms has led to a subjective approach among clinicians with regard to septoplasty referrals:

We don’t really have a good way of objectively assessing the degree of deviation of the septum. So it did come down to, pre-NAIROS, essentially just looking in the nose and seeing how twisted it was and it is a challenge there because nobody has a completely straight septum and some patients will have a septum that’s twisted into the opposite nostril rather than the affected nostril. So the decision to operate was not really based on just the degree of deflection of the septum, it was the additional symptoms as well that the patient’s complaining of.

Surgeon 7

Examining the nose, I will be looking to see if they actually do have a septal deviation and if, in my mind, I think it’s significant enough to cause the degree of symptoms that they have. If they have such a mild deviation and such a significant degree of symptoms, I wouldn’t necessarily correlate the two. I’d be very wary about offering them surgery because their expectations are that all those symptoms that they’re complaining of will improve, whereas I find that they don’t tend to. Structurally, it has to be significant enough for me to think, ‘I can make a difference’.

Surgeon 6

Clinicians also displayed subjective decision-making in relation to managing patients’ symptoms. Most surgeons reported that they often recommended medical management to patients before offering them surgery. Surgeons whose current practice more closely followed the protocol of the NAIROS may have also been more inclined to take part in the trial:

Personally, I don’t think it’s changed what I do. I’ve always maintained giving them proper medical treatment first. That’s, I think, partly why I was quite happy to buy into the study, because it followed my normal practice more or less.

Surgeon 6

However, some clinicians stated that, in routine care, they would often refer patients straight for surgery if their septum was perceived to be highly deviated, or if they had a particular incidence of a traumatic injury:

Interviewer: And in terms of the reasons why you would put people straight through to surgery?

Surgeon 8: Well might be the GP letter, it might be the patient’s history, you know, ‘I could breathe alright up my nose, then I fell on it’ … so that chronological context would be helpful. And just the severity of some people have a very twisted septum that you can see they’ve got a vertical fracture line that’s effectively presenting– the bottom end of the septum is a door across one nostril so those ones you’d be more inclined just to go for surgery.

In the trial, surgeons who felt that particular patients were more in need of having surgery straight away, rather than potentially being prescribed medical management, may not have recruited these patients to the NAIROS trial for this reason. Therefore, the trial population may not have included patients with the most severely deviated septa:

I’m aware that some colleagues have not necessarily recruited the most, well when I say– anecdotally, I’m aware that some colleagues have not necessarily recruited the most severe nasal septal deflections to the study. So I think we may not necessarily be looking at the entire population of deviated nasal septums.

Surgeon 7

Generally, despite the high heterogeneity in staff criteria for referring patients for septoplasty, surgeons felt that the eligibility criteria in the NAIROS protocol was appropriate for the trial:

Interviewer: [A]nd you are happy with the patients who were supposed to be put into NAIROS, the eligibility criteria?

Surgeon 4: Yes. There was nobody we recruited that I wasn’t happy to operate on. I think that’s the main thing.

There was also variation in other aspects of ENT routine practice, including whether or not clinicians used any outcome measures (e.g. SNOT-22, NOSE), and whether or not they followed up with patients after their operation outside the trial setting:

In general we wouldn’t have employed any objective testing, I wouldn’t have done any patient-reported outcome measures, I would have just proceeded with the operation.

Experiences of treatment for nasal obstruction

This theme relates to the experiences of delivering (staff) and receiving (patients) treatment for nasal obstruction.

Reflections on the trial

Overall, the trial recruited the target number of participants. A number of potential facilitators of the NAIROS trial that contributed to its successful recruitment were identified by participants (Table 23). Challenges in relation to the set-up and running of the trial are presented in Table 24. Participants also identified a number of ways in which they had reflected on or made changes in practice as a result of their participation in the NAIROS.

Coherence: how people make sense of the Nasal AIRway Obstruction Study trial

Healthcare professionals considered the research question of the NAIROS to be of great importance to the field of septoplasty, which had previously suffered from a lack of high-quality evidence. In usual practice there were variations in decision-making for septoplasty, including how septum deviation was assessed and which patient symptoms were considered important. This variation in current practice appears to be due to the range of symptoms patients present with in clinic, and the lack of valid and standardised methods for septoplasty referral. However, most surgeons felt that medical management could be a valid alternative to septoplasty for at least some patients. Overall, clinicians described the NAIROS protocol positively, owing to its consistency with current practice, its suggested criteria for assessing eligibility of patients and the flexibility with regard to the surgical technique it allowed them to employ (e.g. optional turbinate reduction).

Health professionals had subtle but important differences in terms of what they understood the aim of the NAIROS to be. Some surgeons believed that the aim of the trial was to provide high-quality evidence for the effectiveness of septoplasty to ensure its future funding. In contrast, several surgeons believed more strongly in the clinical equipoise of the trial; these surgeons hoped that the findings of the NAIROS could help to identify which patients benefit from septoplasty and which benefit from medical management. Generally, patients demonstrated a good level of understanding of the trial, and used the NAIROS information resources (e.g. leaflet, website, digital versatile disc) well to understand and make sense of the trial. However, there were instances when patients’ sense-making of the trial did not align with the trial itself. For example, as in previous studies, a small number of patients demonstrated a lack of understanding of the computer randomisation process. 144

Cognitive participation: whether or not people were willing and able to buy in to implementing the trial

Many surgeons were motivated to take part in the NAIROS because they felt that the results could provide the high-quality evidence needed to ensure the future of septoplasty, whereas other surgeons felt that the results could reduce the number of unnecessary septoplasties conducted. However, there were also difficulties with healthcare professionals buying in to the trial, with some surgeons struggling to convince their colleagues of the value of the NAIROS.

In addition, there were instances of healthcare professionals not adhering to the trial inclusion and exclusion criteria, which may have led to patients with a highly deviated septum not being approached to take part in the NAIROS.

Patients were positive about the fact that the trial was testing established medication and surgery, rather than new and experimental treatments. Previous studies have identified patients’ preference for a particular treatment arm as a barrier to trial recruitment. 131,145,146 By offering patients enrolled in the medical management arm the option of being put on the septoplasty waiting list after 6 months, this may have minimised the effect of patients’ preference on recruitment rates.

Collective action: people’s ability to take on the work needed to implement the Nasal AIRway Obstruction Study trial

Most staff members were positive about trial implementation and felt that the research team collaborated effectively together to execute the trial at site. The involvement of the QRI in the NAIROS was also cited by staff as having a positive influence on recruitment rates. As in previous studies,148,149 research nurses in particular were highlighted by surgeons as being essential to the implementation of the trial at their sites. However, participants also described several challenges in relation to implementation. For example, there were issues with retention of patients at follow-up and logistical difficulties organising NAIROS recruitment. Furthermore, colleagues’ initial enthusiasm about the NAIROS trial did not always translate into action during the trial, such as aiding with screening and recruitment of patients.

Reflexive monitoring: people’s reflection on the benefits and costs of the trial

One staff participant described how their views of the treatments changed during the trial. Originally, they believed that surgery was much more effective than medical management; however, after witnessing a number of patients experience symptom relief from the steroid sprays, they came to view the trial as comparing two more equal treatments. Although the NAIROS quantitative data show that, as a group, those randomised to septoplasty experience more improvement than those randomised to medical management, the qualitative data demonstrate nuance and variation at the level of individuals, with some participants not observing any improvement following septoplasty and others experiencing benefit with medical management. For one patient, visible changes in turbinate size were felt to be attributable to steroid use and to account for improved nasal air flow. Steroid treatment is an established treatment for enlarged turbinates,150 but further research is needed to clarify the role of medical management in the presence of a deviated septum. Patients with septal deviation could clearly understand how straightening the deviation might improve airflow, but sometimes struggled to understand a comparable mechanism for medical management. Providing a mechanism that includes observable changes in nasal structures (‘shrink’) may be more consistent with patients’ understandings of nasal obstruction in the presence of septal deviation than broader terms like ‘decongestant’.

Previous quantitative research has shown mixed findings in regard to patients’ rates of satisfaction and quality of life following septoplasty. 43,62,134,135 In the NAIROS interviews, many patients suggested that information on recovery post septoplasty was lacking, which led to them feeling unsupported following surgery. Inadequate information on aspects such as recovery has been linked to dissatisfaction with surgery and increased anxiety levels among those undergoing day surgery. 151 More comprehensive information on recovery post septoplasty might increase patients’ feelings of support and satisfaction with surgery. Medical management was generally well tolerated in the trial; however, a number of patients did experience nosebleeds and/or discomfort when using the spray. Patients incorporated the sprays into their daily routines to help them remember to use them, but some expressed concerns about the potential cost and burden of long-term medication use. In interviews no-one mentioned potential side effects of steroids, or specific concerns about the use of steroid medications long-term, but these have been a concern in other populations152 and might emerge as a greater concern in the medium to long term.

Overall, many healthcare professionals did not plan to change their current practice until the results of the NAIROS were released. Even then, many felt that the NAIROS matched their current practice quite closely and were considering implementing only a small number of changes, such as the use of particular outcome measures. One surgeon also described how taking part in the NAIROS had altered their recruitment technique for future trials.

Strengths and limitations of the qualitative study

A strength of the qualitative study is that we recruited patients from multiple trial sites, from different age groups and at different time points in the trial pathway. However, the sample was less diverse in terms of ethnicity. In addition, we recruited and interviewed multiple surgeons and research nurses from a range of trial sites. An important limitation of the study is that we recruited only one patient who declined the trial. Therefore, we were unable to explore the factors that influenced patients’ decision not to participate in the NAIROS.

Clinical effectiveness

The NAIROS demonstrates that septoplasty, with or without turbinate reduction, is a highly effective procedure with evidence that it may be considered cost-effective at 24 months.

The primary analysis was ITT based on a sample of 378 participants, comparing SNOT-22 over 6 months from randomisation to initial follow-up, adjusting for baseline severity (NOSE score), baseline SNOT-22 score and gender. A 6-month interval was chosen as a suitable compromise, allowing sufficient time for surgery to be performed and wound-healing to take place, yet ensuring a timely primary outcome assessment to minimise default from follow-up. As stipulated in the commissioning brief, clinical effectiveness and cost-effectiveness were also assessed at 12 months.

The baseline characteristics of the NAIROS recruits were balanced between the treatment arms (see Table 3). Males comprised 67%: septoplasty is known to be more commonly undertaken among males of middle age as a result of septal deviation caused by assault to the face and sporting injuries. 17 A peak age in mid-adulthood among participants is to be expected for similar reasons. There was a significant preponderance of the white ethnic group (88%), as expected from existing UK census data. 157

Overall, the improvement in SNOT-22 score in the septoplasty arm occurred notably earlier and was greater than expected; this was sustained over the trial period. A more modest improvement from 6 to 12 months was noted in the medical management arm, even after removing those participants who withdrew and had non-trial septoplasty. Those who withdrew and who underwent non-trial septoplasty had an appreciably greater reduction in SNOT-22 score at 12 months.

The primary analysis indicated a greater reduction of an average of 20 points (p < 0.0001) in SNOT-22 symptom severity score in the septoplasty arm than in the medical management arm at 6 months (see Table 6). In addition, the lower limit of the 95% CI was −16.4 points, substantially in excess of our a priori MCID of a 9-point reduction in SNOT-22 scores post treatment. Additional treatment efficacy analyses adjusted for covariates all showed similarly positive impacts of surgery over medical management (see Figure 8).

The impact of baseline severity as measured by the NOSE score was evident, demonstrating greater SNOT-22 scores at the primary outcome point with increasing baseline severity. This was accounted for by the inclusion of the baseline SNOT-22 scores as a covariate. The coefficient for this variable was highly significant and close to 0.5, meaning that, at 6 months, SNOT-22 scores were approximately half what they were at baseline (holding all other factors constant). The effect was mirrored in both genders by the subgroup analysis of paired primary outcome data to assess individual changes in SNOT-22 scores by a STEPP analysis (see Figure 16). Several retrospective studies158,159 have noted a similar impact of baseline severity on post-operative subjective nasal obstruction, but without robust quantification. This novel and important NAIROS analysis enables clinicians to quantify expected outcome improvements of patients contemplating septoplasty, with or without turbinate reduction, predicated on the baseline NOSE score. The STEPP analysis, in parallel with an understanding of the potential risks associated with both medical treatment and surgery, substantially improves the quality of information available to the clinician and patient in the clinical decision-making process around septoplasty.

Van Egmond et al.’s58 2019 study reached similar conclusions to those of the NAIROS in a trial of 203 participants randomised to receive either septoplasty, with or without turbinate surgery, or non-surgical management. Neither study found a relationship between outcomes and age or gender. The NAIROS reported a 14-point SNOT-22 surgical superiority score at 12 months, compared with 9.7 points in van Egmond’s58 study, although the latter’s data transformation and variable medical arm treatments makes direct comparison difficult. Van Egmond et al. ’s58 subjective, categorical, clinician-rated assessment of baseline severity failed to correlate with outcome. The mild category comprised 30% of participants, but there is no evidence for the comparability of the subjective assessment with quantitative NOSE scores. In contrast, the NAIROS substantially enhances understanding of the role of quantitative patient-reported data in the selection of patients for surgery by demonstrating that the degree of improvement in symptoms is closely related to baseline severity stratification.

Discontinuation of allocated treatment

The NAIROS trial design did not allow for formal crossovers from one treatment arm to the other. The ITT analysis therefore risks accepting a null hypothesis if the non-surgical arm ultimately contains too many individuals who have undergone surgery, resulting in an underestimation of the treatment effect size. 160 Approximately 20% of participants in the trial did not receive the treatment to which they were allocated. Of 190 participants allocated to medical management, 46 (24%) subsequently had surgery. At 6 months’ follow-up, only five patients had discontinued medical treatment and had undergone surgery, but, by the 12-month follow-up, a further 37 (30%) of the 125 retained participants had undergone surgery and completed SNOT-22 scores (see Figure 7). All but 12% of the septoplasty arm underwent surgery. Despite these groups discontinuing allocated treatment, the NAIROS ITT analysis found superior benefit from surgery, confirmed by per-treatment and per-protocol analyses (see Figure 9).

Secondary outcome measures

The benefits of septoplasty are also seen in the secondary outcome measures, including clinical airflow assessment, PROMs and objective measurements of PNIF and rhinospirometry.

What is the relationship of subjective and objective measures?

Aziz et al.’s173 systematic review found objective assessment tools for diagnosing nasal septal deviation to be of limited use if used in isolation, lacking sensitivity and specificity compared with clinician nasal examination. However, Boyce et al. ,56 using rhinospirometry as a gold standard, reported that the patient-reported DOASS had a sensitivity of 100% and specificity of 60% when assessing 46 patients with a deviated nasal septum. Unlike some objective assessments of the nasal airway, the DOASS can provide discriminatory airflow information on both nostrils, even in the presence of complete unilateral nasal obstruction.

The NAIROS assessed the relationship between the DOASS and the NPR in all 307 participants who attended for baseline assessments. A correlation of p = 0.77 was demonstrated between the DOASS and MIV (see Appendix 1, Figure 29) and a further correlation of p = 0.78 between the DOASS and tidal breathing (see Appendix 1, Figure 30), indicating a strong correlation between subjective and objective airflow measurements. This suggests that the DOASS can provide useful airflow assessment in lieu of objective assessments such as rhinospirometry, which is expensive, cumbersome and operator dependent.

To test the relationship between the primary outcome and airflow measurements, the NAIROS analysed SNOT-22 outcomes in relation to baseline severity stratification with post-decongestant DOASS (see Appendix 1, Figure 32). The NAIROS did not find a clinically useful relationship to facilitate the use of the DOASS to predict postoperative symptom (PROM) improvement following surgery, although this was a post-decongestion measurement, thus not necessarily reflecting normal nasal airflow.

In summary, objective measures of nasal airflow show greater nasal airway improvements across the range of baseline severity in the septoplasty arm than in the medical management arm. Rhinospirometry and the DOASS appear to demonstrate greater utility than PNIF as objective assessments. However, the use of the NOSE in combination with clinical examination offers a robust, reliable and inexpensive assessment for surgical decision-making.

Reduction of the inferior turbinate

The NAIROS protocol permitted turbinate reduction, reflecting UK surgical practice. Of the 155 septoplasties for which information on turbinate reduction was available, 88 (57%) included a turbinate reduction.

There was considerable variation across trial sites on the frequency of turbinate reduction performance. The intention had been to perform a stratified analysis by turbinate reduction, but, given that there was considerable variance between the intention to reduce the turbinate at baseline and actual turbinate reduction at the time of surgery, this was not particularly informative (see Appendix 1, Table 57). At baseline, those in the surgical arm who were recommended for turbinate reduction had slightly higher SNOT-22 scores.

The outputs of the NAIROS infer that turbinate surgery added no additional improvement to septoplasty alone (see Table 10). However, the decision to perform turbinate reduction was at the discretion of the investigator and, had this additional procedure not taken place, the improvements noted may not have been evident. Van Egmond et al. 58 similarly reported no additional benefit of turbinate surgery. The evidence for turbinate surgery in the literature is conflicting. Van Egmond et al. ’s61 2018 systematic review concluded that ‘the limited number of studies comparing septoplasty with concurrent turbinate surgery to septoplasty alone generally showed postoperative improvement, but their results should be interpreted with caution due to methodological “flaws” ’. The trial does not address the question of whether or not turbinate surgery is more beneficial in improving the nasal airway than septoplasty alone.

Treatment success

The NAIROS specifically asked participants who received septoplasty about success of treatment (see Table 9). At 6 months, 134 out of 166 participants in the septoplasty ITT arm provided information on their perception of nasal airway improvement (i.e. better/same/worse): 108 out of 134 (80.5%) felt that their nasal breathing had improved, 19 (14.1%) considered their nasal breathing to be the same and seven (5.2%) reported that it had deteriorated. At 12 months, fewer data were available, with only 89 out of 166 participants providing information. Of those, 66 out of 89 (74.2%) participants felt that their nasal breathing was better and eight (8.9%) believed it was worse. Van Egmond et al. 58 reported that subjective and objective benefits of septoplasty persisted over the full 24 months of follow-up, although they did not expressly report patient satisfaction rates. In contrast, Pedersen et al. ,158 published an assessment of patient-reported satisfaction at 12 months using a subjective, categorical, patient-rated assessment of severity. They reported no improvement in 69% of patients with mild nasal obstruction and in 43% of patients with moderate nasal obstruction before surgery. In addition, 15% of all patients in the study by Pedersen et al. reported severe nasal obstruction 12 months after surgery. Neither study accurately quantified mild symptoms. Overall, the NAIROS participant perception of improvement in their nasal airway appears favourable, compared with other studies, potentially because patients with mild baseline scores were excluded. Our qualitative work reported that septoplasty had varying levels of success among the participants sampled. However, those who found the operation effective typically experienced quite high levels of success, with participants feeling that they had experienced a major reduction in their symptoms following the operation. Studies reporting success of septoplasty demonstrate considerable heterogeneity21,174 in the manner of assessment, making direct comparisons difficult.

Only 6 out of 166 surgical arm participants (3.6%) were recommended to consider revision septoplasty, although the reasons underpinning these decisions were not recorded. For four participants, the decision to offer revision surgery was reported for either ‘same’ or ‘worse’ symptoms. For the remaining two participants, symptoms were reported as ‘better’. In the only other RCT of septoplasty with or without turbinate surgery,58 1% of patients required revision septoplasty. Reported rates in the literature vary from 1% to 12%. 175–177

In all NAIROS cases, surgery was performed by experienced surgeons in both teaching and district hospital otolaryngology departments. The results of both the NAIROS and van Egmond et al. 58 provide a benchmark for colleagues and trainees in the wider surgical community to assess nasal airway surgery outcomes in their own practices.

Safety

Economic evaluation

On average, septoplasty was more costly and more effective in terms of improvements in SNOT-22 scores and QALYs gained. When AEs were used as the outcome measure, septoplasty, a surgical intervention, incurred a greater number, as expected.

The incremental cost per QALY gained at 12 months was £27,114. The probability of septoplasty, with or without turbinate surgery, being cost-effective, compared with medical management, was 15% at a £20,000 threshold for an additional QALY, and 68% at a £30,000 threshold. Using strict NICE guidance, septoplasty, with or without turbinate surgery, has a low probability of being considered cost-effective at 12 months. A sensitivity analysis estimating the cost of surgery using microcosting, instead of the NHS tariff, suggested that septoplasty had a 79% probability of being considered cost-effective at 12 months at a £20,000 threshold for an additional QALY. This result provides further support that it is the cost of surgery determining cost-effectiveness at 12 months. However, these results need to be interpreted with caution.

It was anticipated that septoplasty was likely to be more costly and potentially more effective, but the 12-month duration of the trial was insufficient for the additional benefit to offset the additional costs of surgery. This is illustrated in the economic analyses in which we compare the incremental cost per QALY for septoplasty versus medical management at 6 months’ follow-up (> £100,000) with that at 12 months (£27,114). For this reason, an economic model was undertaken to extrapolate the trial results to 24 and 36 months post randomisation. The model was a simple decision-tree populated using the trial data and clinical inputs. At 24 months, septoplasty remained, on average, more costly and more effective in terms of QALYs gained; however, the ICER reduced to around £13,000, which is well within the £20,000 NICE threshold for an additional QALY. By 36 months, the septoplasty ICER had reduced to around £7000 per QALY. In these analyses, septoplasty had the highest probability of being considered cost-effective, compared with medical management, at 24 months (99%) and 36 months (100%), at a £20,000 threshold for an additional QALY.

This is the first RCT to evaluate the cost-effectiveness of septoplasty, compared with a standardised medical management regime, in the management of a deviated nasal septum.

Study strengths

• To our knowledge, the NAIROS is the first RCT of septoplasty, with or without turbinate surgery, incorporating two well-defined, reproducible treatment arms.

• The trial was designed to have 90% power to detect the MCID of 9 units on the SNOT-22 scale. The trial recruited to target and the observed attrition level was no higher than allowed for in the sample size calculations.

• The primary outcome is patient-reported, and a major strength of the trial is that we amended the protocol to collect this remotely by e-mail, post and an online platform. This helped to maximise data accrual, especially during the later period of the trial when face-to-face visits were halted owing to COVID-19.

• We recruited participants from a wide range of sites across Great Britain, which included both large teaching hospitals and smaller district general hospitals, to represent the whole population.

• A broad range of well-recognised, validated outcome measures, both subjective and objective, were used to determine the success of treatment among the trial participants. The trial recruited to target with no funding extension.

• An electronic version of the primary outcome PROM was successfully used to achieve the target number of participants reporting and returning the associated data.

• We conducted the primary statistical analysis on the complete-case data. A sensitivity analysis accounted for the small number of withdrawals from the trial and missing data using multiple imputation. The additional analyses undertaken supported the primary ITT analysis conclusions. The per-protocol and per-treatment analyses corroborate the ITT results, confirming the greater improvement in patient-reported outcomes of those participants receiving surgery than those treated medically.

• The objective measures of nasal airflow, specified by the funder, confirmed the impact of surgery over medical management, although these did not contribute over and above PROMs in determining the best management strategy for patients with a deviated nasal septum.

• A major strength of this trial was the embedded economic evaluation. The response rate to the data collection tools used to inform the economic analysis was relatively high, with nearly two-thirds of participants having complete data. Costs and effects were estimated at 12 months and were extrapolated so that the longer-term costs and benefits associated with septoplasty could be considered. Finally, a number of sensitivity analyses were undertaken to test the robustness of the economic conclusions. Reassuringly, these analyses did not change our conclusions, in that septoplasty was, on average, more costly but more effective than medical management. An adjusted analysis using SUR was used to estimate the difference in costs and QALYs. This is a robust method for estimating the ICER as it facilitates the simultaneous estimation of costs and effects and takes into consideration unobserved individual characteristics that may affect costs and effects. 114

• The inclusion of a qualitative process evaluation enabled a comprehensive understanding of the benefits and burdens of both treatments and highlighted ways in which treatments could be optimised. It also ensured that learning from the trial is captured to inform future surgical trials.

• Patients were influential voices throughout the evolution of the NAIROS RCT, from PPI at the design outset to ongoing contributions from the lay participant on the TSC.

Study limitations

• All patients who were sent a PIS were entered on the screening log. Pre-screening patient telephone contact and dedicated NAIROS clinics were not permitted by the NHS REC. As a result, it is likely that only those motivated patients identified in GP referral letters/busy ENT clinics, and who were prepared to return for NAIROS assessments, were recruited and took part in the randomisation process.

• The NAIROS consent process was complex and time-consuming, incorporating three stages. The first two stages of consent (eligibility and audio-recordings) could be completed by a research nurse, thereby saving time in a busy clinical environment. However, given that the trial was a CTIMP, stage 3, the consent to the main trial, required a clinician. The consent process and the trial assessments took at least 1–2 hours, limiting the number of patients recruited in each clinic.

• Although it was recommended that recruitment take place in research clinics, many sites were unable to do so, which led to challenges in finding both time and physical space for recruitment at some sites.

• At baseline > 80% of patients had NOSE scores in the severe/extreme category. It is likely that some patients with less severe nasal obstruction either responded to treatment provided in primary care or, once referred, did not wish to participate in the trial.

• The qualitative substudy interviewed only one person who declined to participate in the trial. We therefore know relatively little about the preferences and experiences of those who did not wish to participate in the trial.

• Nasal obstruction is a non-specific symptom with many underlying possibilities (e.g. allergic rhinitis, non-allergic rhinitis, nasal valve dysfunction). In addition, making a diagnosis of septal deviation is challenging in primary care because of the limitations of nasal examination without specialist equipment.

• The NAIROS surgical interventions were performed by experienced surgeons. In NHS practice, septoplasty with or without turbinate reduction is frequently performed by junior trainee surgeons, albeit often supervised by more senior colleagues.

• COVID-19 had a significant impact on the NAIROS trial. All forms of airway clinical assessment and objective testing of nasal airway function were suspended from March 2020 onwards. As a result of the smaller numbers of participants assessed at 12 months, this may have had an impact on the precision of the statistical outputs.

• Twelve months is too short a time horizon for the costs associated with a surgical intervention such as septoplasty to be offset by the benefits; however, surgery becomes cost-effective within 24 months.

• One of the main challenges of the economic evaluation was the microcosting exercise. First, the NHS tariff includes additional costs such as overheads that were not included in the microcosting exercise. Second, costs were sourced from only one site, but are presented for the reader to judge generalisability. Finally, septoplasty data used to inform the microcosting exercise were collected only for participants randomised to septoplasty, but 30% (n = 47) of participants randomised to medical management underwent septoplasty. This has meant that variations in surgical costs were not captured in the bootstrapping analysis as every participant who underwent surgery was assigned the same cost.

• The economic model was populated using data from a single study. However, validity checks and sensitivity analyses were undertaken to ensure the robustness of these results.

Implications for practitioners and health services

There are currently no evidence-based guidelines for the definitive treatment of nasal obstruction in the presence of a deviated nasal septum. In the UK many existing Clinical Commissioning Groups require up to 6 months of nasal steroid treatment in primary care before sanctioning referral to secondary care ENT services. 153,154 A 2015 US-based clinical consensus statement187 agreed a ‘trial of medical therapy of more than 4 weeks’ duration is unnecessary to assess surgical candidacy for septoplasty. However, the panel did not reach consensus on the statement that a four-week trial of nasal steroid prior to septoplasty is sufficient to assess surgical candidacy’.

Given that approximately 16,000 septoplasties are undertaken annually in the NHS, the NAIROS results justify the research investment made, despite its hitherto limited evidence base.

The NAIROS is the first RCT that definitively demonstrates that septoplasty, with or without turbinate reduction, in comparison with a medically treated cohort, is a highly effective procedure with evidence that it may be considered cost-effective at 24 months. The NAIROS-generated evidence confirms that the use of a brief assessment tool (the NOSE) in combination with endoscopic assessment/confirmation of a deviated nasal septum as the principal pathology defines those patients who will have benefit from septoplasty, with or without turbinate reduction. Although objective measurements have corroborated the superiority of surgery over medical management, the NAIROS has not demonstrated that such measurements provide useful information beyond that of clinical and baseline PROM assessment.

Most patients will have sought treatment in community and primary care settings and will have had a trial of nasal steroid spray. We recommend that patients who do not see improvement with this first-line treatment are referred to secondary care to undergo a clinical nasal airway assessment and complete the NOSE severity scale. When weighing up the risks and benefits of different treatment strategies, those with a NOSE score of > 30 can reliably be advised that septoplasty, with or without turbinate surgery, is a highly effective, evidence-based treatment option and that the level of predicted benefit will relate to the severity of their symptom burden at presentation.

Context of trial/informing NHS guidance

The NAIROS has shown that baseline assessment using the NOSE scale can predict the degree of postoperative symptom improvement. In this regard, we have operationalised the NOSE score and would recommend its use in standard clinical practice for preoperative decision-making for nasal septal surgery.

Modest overall improvements in SNOT-22 scores in the medical management arm were evident at both follow-up intervals. The reasons for this improvement, beyond the initial 6-month trial period of steroids/saline spray, are likely to be related to several factors. First, as previously described, around 25% of the medical management arm underwent septoplasty during the trial. Second, it is recognised that rhinitis has a fluctuating impact on nasal symptoms and this may have potentially contributed to the improvement effect noted. Nasal steroids/saline improve congestion and nasal airflow among patients with allergic and non-allergic rhinitis. 75,188,189 It is expected that a proportion of participants should obtain some benefit from this treatment given the prevalence of rhinitis in the general population. 190 Existing guidelines187 indicate that there is no consensus among clinicians as to the benefit of medical management (in comparison with surgery) as a treatment for patients with a deviated nasal septum. The most common reasons cited for ceasing medical treatment in the NAIROS were ‘not happy with sprays/side effects’ (78%) and worsening symptoms (15%). Pragmatically, nasal spray delivery may be impeded or impossible in cases of a particularly deviated septum. In practice, treatment discussions with patients wishing to consider medical management should also include reference to the potential risks of side effects including minor nosebleeds, nasal dryness and irritation, and the need for long-term treatment and associated costs.

The majority of patients presenting to primary care with nasal obstruction (and without red-flag symptoms) who do not have a deviated nasal septum may be treated medically. Those whose symptoms persist despite treatment should be referred to secondary care for a thorough and careful clinical assessment, including nasal endoscopy and NOSE score assessment. The NAIROS has demonstrated from both clinical and health-effectiveness perspectives that, for those patients with a NOSE score of > 30 and a deviated nasal septum, septoplasty, with or without turbinate reduction, is superior to long-term medical therapy.

Implications for patients and the public

Ultimately, it is for the patient to decide based on the information provided whether, after weighing up risks and benefits, they wish to consider medical therapy or to proceed with surgery.

The NAIROS has developed evidence showing that septoplasty, with or without turbinate reduction, outperforms a combination of nasal steroid/saline sprays in the treatment of nasal obstruction associated with a deviated nasal septum. Given the heterogeneity of patients who present with nasal blockage, an initial trial of medical management remains appropriate for most patients. The duration of this treatment is not yet defined.

After a trial in the primary care setting of nasal steroid/saline medication, those patients referred to secondary care who have endoscopic evidence of septal deflection in the presence of moderate, severe or extreme NOSE scores, when weighing up the risks and benefits of different treatment strategies, can reliably be advised that septoplasty, with or without turbinate surgery, is a highly effective, evidence-based treatment option, and that the level of predicted benefit will relate to the severity of their symptom burden at presentation.