Key findings

Endoscopic pouch stapling is a safe but relatively uncommon procedure: over a period of one decade at our six centres, the number of primary pharyngeal pouch stapling procedures ranged between 1.2 and 9.2 cases per year. Many of the eight audited domains were satisfactory. Furthermore, all centres achieved 90 per cent resumption of oral intake within 2 days. The pooled data across all centres achieved the less than 2-day hospital stay standard. However, initial success, symptom resolution and recurrence rates all failed to meet the standard across pooled secondary centres, but did so at the tertiary centre. Although these differences are arguably of clinical significance, statistical significance was not reached (p = 0.06, p = 0.11 and p = 0.26 respectively). Standardised post-operative symptom assessment does not occur at any centre and may be a focus for future studies.

Our pooled mortality and perforation rates were superior to the recommended standard. Our tertiary centre outcomes were superior to audit standards on all eight measured domains (Leong et al.).^{Reference Leong, Wilkie and Webb5}

We excluded 21 revision procedures from the pooled secondary care case analysis. This is because we sought to compare primary cases against the benchmark standard, and revision surgery patients represent a discrete patient cohort. We are aware that this may have introduced further bias into our results and conclusions, but primary versus secondary subgroup analysis was not a pre-determined aim of this study.

For reference, if the revision cases are included in a pooled secondary centre cohort, the symptom improvement rate is 81.6 per cent (vs 82.7 per cent for primary case outcomes alone), so we feel the take-home message of our study is not significantly altered by this decision.

It is also worth noting that there was a higher success rate for revision surgery at the tertiary centre for patients who had undergone abandoned procedures at the secondary centres, even with repeat endoscopic stapling. Perforation rates were similar regardless of centre; however, at the head and neck centre, when a perforation did occur (3.2 per cent), all patients returned to the operating theatre for external repair (within 24 hours) and had no further complications.

Our study suggests that endoscopic pharyngeal pouch stapling is performed less than twice a year in some centres. It is noteworthy that there was still no significant discrepancy between numbers of perforations in our series, despite in-patient stay being prolonged in those managed conservatively; our numbers are too small to draw meaningful conclusions about this.

Although lower volume centres have low mortality rates and brief in-patient stays, the symptom resolution and procedure abandonment rates fell below our standard. We suggest that this may represent conservative surgical decision-making related to the full spectrum of treatment options being less readily available.

Where problems performing the procedure occurred, this was usually because of patient issues such as dentition or limited mouth opening. In these instances, the multidisciplinary teams based at the tertiary centre allowed for alternative and successful management strategies (e.g. dental extractions and subsequent dental implants, or open surgery). These findings raise the question of whether such a service should be centralised, with pouch procedures becoming the preserve of specialist high-volume centres or specialist practitioners.

The observed (but not statistically significant) discrepancy between procedure abandonment rates at the tertiary centre and the majority of other hospitals may represent an intentional, conservative approach adopted by surgeons out of consideration for the population in question (regarding age, frailty, co-morbidities, anaesthetic risk, etc.). These demographic and peri-operative issues are part of the reason endoscopic techniques first gained traction, when compared to open surgery. However, these same patients may be at risk from undergoing repeat or recurrent general anaesthesia if one or more surgical procedures are abandoned or symptoms recur. A conservative ‘mindset’ or risk-averse approach may also contribute to incomplete or partial resolution of the pouch and higher symptom recurrence rates. Whilst this is speculative, some operative notes in our series do record narrative comments to this effect. Furthermore, a large proportion (90.5 per cent) of the failed or recurrent symptom cases underwent successful endoscopic stapling at first attempt at the tertiary centre, which may also support our inference regarding differing approaches to risk. Although not a truly testable hypothesis, we surmise that this only lends further credence to the notion of centralisation of service.

Over recent years, many surgical specialties within the UK have embraced the concept of Getting It Right First Time (‘GIRFT’; the National Health Service quality improvement project addressing variation of care). Recommendations from the most recent Getting It Right First Time report¹³ suggest the development of surgical outcome metrics that are not yet covered by existing audits. Pharyngeal pouch surgery is an area within our specialty that could benefit from such an approach, and it would be considerably worthwhile for other regions to present their figures in accordance with the standards used here.

Study strengths and limitations

Our adopted audit standard was taken from the benchmarks suggested in a previous UK pooled analysis;^{Reference Leong, Wilkie and Webb5} therefore, this standard may well be subject to publication bias, with truly representative outcome figures potentially being less favourable.

There are also limitations inherent to retrospective data analysis. This study was non-randomised, and there was no comparison or control group. In addition, across the six centres, operative technique and experience of the individual operative surgeons were not assessed or recorded, and this will have potentially affected the reproducibility of our conclusions. Furthermore, eight cases had incomplete datasets and were excluded from the initial analysis. In such a small cohort, these outcomes could have significantly skewed the overall performance against the audit standard, and therefore influenced the fidelity of our findings and external validity of some of our conclusions.

An alternative option would have been to include all cases in our analysis, regardless of data completeness, but this could potentially underestimate the incidence of significant events (e.g. perforations) and would introduce the potential for confounding bias in our analysis. Although patients who have undergone unsuccessful procedures are routinely referred to the tertiary centre, our recurrence rate is also affected by patients who may be lost to follow up. For instance, if symptoms recurred, some patients may have been referred to another tertiary centre.

There may also have been cases lost to coding, which could lead to follow-up and selection biases in our series. Abandoned procedures, for example, may simply have been coded as pharyngoscopy. We sought to minimise this, however, by including International Classification of Diseases 10th Revision diagnostic codes in our data capture criteria, rather than simply searching on the basis of Office of Population Censuses and Surveys’ Classification of Surgical Operations and Procedures codes.

Unfortunately, in our region, there was no uniform recording of patient-reported outcomes using, for example, symptom-specific tools like the Eating Assessment Tool-10 (‘EAT-10’),^{Reference Belafsky, Mouadeb, Rees, Pryor, Postma and Allen14} the M D Anderson Dysphagia Inventory^{Reference Chen, Frankowski, Bishop-Leone, Hebert, Leyk and Lewin15} (originally validated in head and neck cancer) or a more generalisable score like the Glasgow Benefit Inventory.^{Reference Robinson, Gatehouse and Browning16} This could be facilitated through dedicated speech and language team involvement, which could be streamlined as part of a specialist service. It is important to consider multifactorial contributions to dysphagia in an ageing and complex patient population, and there is no guarantee that successful endoscopic or open surgery will lead to complete resolution of swallowing difficulty.^{Reference Martino and McCulloch2} Nonetheless, we conducted a crude assessment of patient-reported symptom improvements; this was not the primary purpose of this study, however, as the efficacy of endoscopic stapling from a patient-reported outcome measure perspective has been previously assessed elsewhere.^{Reference Leong, Wilkie and Webb5,Reference Belafsky, Mouadeb, Rees, Pryor, Postma and Allen14}

Despite these critiques of our study design, the authors were unable to identify any other prospective or retrospective multi-centre audits of outcomes against pre-determined standards for endoscopic pharyngeal pouch surgery in the literature. To our knowledge, there are no other comparable studies of pharyngeal pouch cohorts with a larger sample size.

Comparisons with other studies

In 1997, Koay et al. published the findings of a postal survey study which demonstrated that more than 75 per cent of surgeons (308 surgeons responded) who operated on pharyngeal pouches performed three or fewer pouch operations per year.^{Reference Koay, Sharp and Bates17} In 2004, Siddiq and Sood reported that, in their postal survey (n = 227), only 1 per cent of surgeons performed more than 20 procedures per year, and 65 per cent performed 5 or fewer procedures per year.^{Reference Siddiq and Sood10}

The findings of a 10-year UK retrospective series, published in 1995, revealed a similar incidence to our study. In that study of 103 patients, 35 had conservative treatment and 68 underwent external surgery (but with a fistula rate of 8.9 per cent and median hospital stay of 7 days).^{Reference Laing, Murthy, Ah-See and Cockburn18} In the intervening 25 years, these numbers seem largely unchanged, with relatively low mean numbers of annual endoscopic pouch surgical procedures undertaken by our peripheral units compared to our tertiary unit (four and nine procedures respectively). These numbers are comparable to annual pharyngeal pouch procedure numbers reported in recent case series from other parts of the world, including Las Vegas (n = 8),^{Reference Ching, Kahane, Reeve and Wang19} Montreal (n = less than 3),^{Reference Jones, Aloraini, Gowing, Cools-Lartigue, Leimanis and Tabah20} Tel Aviv (n = 8),^{Reference Oestreicher-Kedem, Wasserzug, Sagi, Carmel and Zikk21} Sydney (n = 11),^{Reference Veivers22} Victoria (Australia) (n = 11),^{Reference Alwan, Phyland, Pudel, Booth and Vallance23} Hull (UK) (n = 8)^{Reference Agalato, Jose and England24} and Oxford (n = 13).^{Reference Bola, Ashman and Winter25}

We can compare the trends demonstrated in these publications with current trends in Hospital Episodes statistics. The latter actually suggest that numbers of procedures performed on pharyngeal pouches (Office of Population Censuses and Surveys’ Classification of Surgical Operations and Procedures codes E23.3 and E24.3) in England since 2000 has risen from approximately 325 to approximately 750 in 2019.^{Reference Leong, Wilkie and Webb5,26}

Presuming these numbers are similar across the rest of the UK, the NICE and Confidential Enquiry into Perioperative Deaths recommendation for the centralisation of pharyngeal pouch surgery, limited to a few specialist high-volume surgeons or centres, could be an appropriate option for surgeons and patients, similar to the hub-and-spoke system used in head and neck cancer surgery. This would concentrate expertise and offer targeted training for registrars, whilst hopefully improving patient outcomes.

Dutch, Swiss and German surgeons have minimal operative numbers for certification and licensing, and North American surgeons have recommendations. There is a precedent in ENT for expected operative numbers from the British Thyroid Association, which suggests that surgeons who operate on patients with thyroid cancer should perform a minimum of 20 thyroidectomies per year.²⁷ Adam et al. found that the risk of permanent vocal fold paralysis after thyroid surgery decreases with the number of cases performed at the treating hospital. The best results (study of 17 000 patients operated on by 5000 surgeons) were achieved at an annual operating frequency of over 25 operations per surgeon.^{Reference Adam, Thomas, Youngwirth, Hyslop, Reed and Scheri28}

Bauer and Honselmann conducted a review of 35 systematic reviews of minimum operative numbers, and found a volume–outcome relationship, with two-thirds of studies relating this to hospital and physician volumes. The higher the number of cases, the more likely that good treatment results were achieved, across a variety of different surgical procedures.^{Reference Bauer and Honselmann29}

Markar et al. found that once a consultant surgeon performed 15 operations, they achieved a 50 per cent reduction in mortality.^{Reference Markar, Mackenzie, Lagergren, Hanna and Lagergren30} They found that a plateau of optimised long-term results in oesophageal surgery was achieved only after performing 35–59 operations. The patient threshold for improved mortality was 5 annual cases for oesophageal perforation, but 11 annual cases for paraesophageal hernia.^{Reference Markar, Mackenzie, Wiggins, Askari, Karthikesalingam and Faiz31} Similarly, another study by Markar and colleagues found that high-volume endoscopists have a significantly reduced 30-day mortality rate after endoscopic mucosal resection, and that endoscopist volume is of greater importance than hospital volume. The initial stage of endoscopist competency gain was found to be the most critical, and there was significant association with mortality in the first five cases.^{Reference Markar, Mackenzie, Ni, Huddy, Askari and Faiz32}

In the UK, there is also precedent for reorganisation of ENT surgical services delivery from both the head and neck cancer hub-and-spoke model and the cleft palate service. Cleft lip and/or palates have a greater annual incidence than pharyngeal pouch (approximately 1.6 per 1000 in UK). Akin to the head and neck ‘hub-and-spoke’ model, all cleft surgery takes place at the ‘hub’ centres, allowing those specialist centres to perform between 80 and 100 procedures per year.³³

Some authors have assessed pouch size,^{Reference Bola, Ashman and Winter25,Reference Buchanan, Riffat, Mahrous, Fish and Jani34} although there does not appear to be consensus for pouch size reporting between Morton and Bartley^{Reference Morton and Bartley35} (small = less than 2 cm, medium = 2–4 cm, and large = more than 4 cm) and Overbeek and Groote^{Reference van Overbeek and Groote36} (vertebral bodies – small = less than one vertebrae, and large = more than three vertebrae) grading. Although radiological data were available for determining pouch size, this was not part of our study. An analysis including pouch size, position and patient-reported long-term functional outcome assessment (e.g. Eating Assessment Tool-10) should be the focus of future studies of pharyngeal pouch outcomes if ethical approval is granted.