Volume-outcome correlation in adrenal surgery—an ESES consensus statement

This issue has been addressed predominantly in the USA. Linderman et al. analysed the New York SPARCS database of 6.054 adrenalectomies and found that urologists (UR) performed more adrenalectomies (47%) than general surgeons (GS) (35%) and endocrine surgeons (ES) (18%) [18]. ES who performed adrenalectomies were more likely to be high-volume surgeons compared with GS or UR [18]. In contrast, Park et al. analysed 3.144 adrenalectomies performed between 1999 and 2005 and found that UR performed only 37% of operations while non-UR operated on 2.253 patients (73%) [19]. Another study from the USA showed that the majority of procedures were performed by non-UR (without distinction ES/GS) for both benign (57% vs. 43%) and malignant disease (66% vs. 34%), with no observed differences between the two specialties regarding morbidity and mortality [20]. In 2016, Sood et al. also showed that the large majority of adrenalectomies within the USA was performed rather by GS/ES than by UR both for benign (4.368 vs. 187) and malignant disease (268 vs. 21) [21]. In the largest study to date (58,948 adrenalectomies), Monn et al. showed that UR performed more adrenalectomy than GS (60% vs. 40%, respectively) but the percentage of adrenalectomies performed by UR has slowly decreased over time [22].

There are three reports from Europe on this topic. In 2008, the national Spanish database registered 1.042 adrenalectomies but unfortunately no distinction by specialty was available [23]. In 2016, Palazzo et al. reviewed hospital admissions the UK for the year 2013–2014 of 795 adrenalectomies. Urologists represented 19% (7 out of 36) high-volume surgeons (threshold ≥ 6 procedures/year) and 43% (63/146) of low-volume surgeons [24].

Overall, the published data show that surgical sub-specialty has no significant impact on perioperative outcomes and rather focus the debate on the positive correlation between increased surgical expertise (as estimated by volume workload) and perioperative outcomes.

There is a plethora of papers reporting initial experience, extended experience, and multi-institutional, national, or international collaborations. Some results extrapolated from national surveys or registries may harbour some limitations and biases.

The national audit of the British Association of Thyroid and Endocrine Surgeons (BAETS) reported 2.073 adrenalectomies between 2012 and 2017 [25] but disclosure was on voluntary bases and the audit captured only a fraction of the cases as there are 7–800 adrenalectomies performed yearly in the UK [24]. The individual annual workload of 50 surgeons who reported their own data ranged between 1 and 40/year [25]. Data from Netherlands does not provide an individual workload but emphasises that those working in a hospital recognised by the Danish Adrenal Network have better outcomes [26].

In France, 1681 adrenalectomies were performed in 2017 and university hospitals performed the majority of them in 25 high-volume centres (all academic, almost all with a specialised unit in endocrine surgery) with a mean workload of 42 adrenalectomies/year. The rest of 229 public and private hospitals shared the remaining 627 adrenalectomies (mean value = 2.8 adrenalectomies/year). This trend between university and teaching hospitals and small peripheral hospitals had become evident since the early 2000 [27]. In a more recent analysis of 9820 patients operated in France between 2012 and 2017, 280 centres had performed five or fewer adrenalectomies during the 6-year period of the study and only 26 centres performed half of the national instances of adrenalectomy [28].

At the top range of number of adrenalectomies performed in the same institution, one can find reports of 560 retroperitoneoscopic adrenalectomies or over 300 robotic adrenalectomies [29, 30] or 420 adrenalectomies during 19 years and 520 patients over 20 years [31, 32]. A recent meta-analysis that reviewed 29 series from endocrine referral centres comparing different surgical approaches for adrenalectomy included reports of 10 up to 267 cases [33]. Due to the extreme rarity of the disease, large series adrenocortical carcinomas have reported only through multi-institutional cohorts studies (e.g. the report by Lee et al. on 201 patients with adrenocortical cancer [34]).

Low-volume series are also found in the literature predominantly from developing countries where centralisation might be scarce hence the authors reported 34 adrenalectomies in 10 years in Saudi Arabia (a mean of 3/year) [35], or 46 adrenalectomies in 6 years in Thailand (7.6/year) [36] or 35 adrenalectomies over 12 years in three different hospitals (1 adrenalectomy/year/hospital) [37].

The literature about the learning curve for minimally invasive adrenalectomy is scarce. Few small controlled trials are available and most studies are retrospective in nature, with significant heterogeneity among them and increased risk for publication bias as well as other confounding factors. In addition, reporting of outcomes and follow-up varies significantly, making it difficult to combine and compare such data. Finally, most of the studies do not report details on the previous expertise of their surgeons making the generalisation of their findings difficult [38].

Adrenalectomy has a low mortality rate. The BAETS 2017 national audit report identified 9 deaths during the index admission out of 1840 adrenalectomies registered (0.5% mortality rate) but no correlation was made with the workload of individual surgeons and no cross-checks made with other hospitals records for these self-declared events [25]. The French study analysed 162 cases of mortality out of 9820 adrenalectomies, with a mortality rate of 1.0 % in 254 low-volume centres and 0.4% in 12 high-volume centres. At 90 days, the mortality rate was 1.8% in low-volume centres and 0.9% in high-volume centres (P < 0.001) [28]. Similarly, in the study by Park et al. of 3144 adrenalectomies, high-volume surgeons had a significantly lower mortality rate of (0.56% vs. 1.25%, P = .004) [18]. It is important to note that this mortality rate is twice as high as after bariatric surgery hence this aspect cannot be ignored and surgeons with low volume and limited experience should use a risk stratification in order to choose patients who might benefit from referral to a regional centre with larger experience.

Postoperative in-hospital complication rate of 20% and variable hospital durations of stay of 2–9 days have been reported. A number of studies have assessed the impact of both surgeon and hospital volumes on postoperative outcomes for adrenalectomy. While all report an association between treatment by either a high-volume surgeon or at a high-volume hospital and improved outcomes, the method of defining a high- vs. low-volume surgeon varied across the studies. Volume thresholds are often selected arbitrarily or based on quartiles or quintiles, which makes comparison across studies difficult [52]. In the study by Park et al. of 3144 adrenalectomies performed between 1999 and 2005, adrenalectomy by low-volume surgeons (< 4 cases/year) was associated with more complications (18.3 vs. 11.3% for high-volume surgeons, respectively, P < 0.001) and longer durations of stay (5.5 vs. 3.9 days for high-volume, P < 0.001) [19]. In 2018, Lindeman et al. confirmed those data and showed that high-volume surgeons had a significantly lower complication rate (14.4% vs. 8.8% P < .001) and lower median LOS (high volume: 2 days [IQR 1–5] vs. low volume: 4 days [IQR 2–7], P < .001) [18].

Anderson et al. performed a retrospective cohort study of hospital discharges of 6712 patients operated by 3496 surgeons who performed adrenalectomies at 687 hospitals from the USA between 1998 and 2009 [53]. Overall 1378 (20.5%) patients experienced at least one complication after adrenalectomy. After adjusting for patient demographic, clinical, and hospital variables, increasing annual surgeon volume was associated with decreasing odds of experiencing a postoperative complication up to 6 cases per year. Based on the threshold ascertained from this analysis, high-volume surgeons were categorised as those performing ≥ 6 adrenalectomies per year. Patients who had their adrenalectomy completed by a low-volume surgeon were the majority (83%), were more likely to be older (60 vs. 56 years, respectively), of black race (11% vs. 7%), and covered by government-provided health insurance (Medicare and Medicaid; 48% vs. 37%) in a nonteaching (40% vs. 5%) hospital. They were also more likely to have a urologic complication (5% vs. 2%, respectively, P < .001), respiratory complication (7% vs. 5%, P = .008) or any complication (22% vs. 14%, P < .001) and they had higher rate of in-hospital mortality (2.4% vs. 0.6%, P < .001). Their hospital duration of stay was greater (median 6 vs. 3 days, respectively, P < .001) and the cost of their care was higher, with a difference of $1,659 per patient (median $11,543 vs. $9884, P < .001). Compared with the patients of high-volume surgeons, patients of low-volume surgeons had a 36% increase in the odds of having a complication if the surgeon performed 1 case per year, 24% for 2 cases per year, 15% for 3 cases per year, 8% for 4 cases per year, and 3% for 5 cases per year (all P < .05) [53].

Using the same threshold cut-off point of 6 adrenalectomies/year applied to data retrieved from Hospital Episode Statistics in England, Palazzo et al. found that hospital duration of stay was 60% greater and rates of 30-day readmissions were 47% greater for low- vs. high-volume adrenal surgeons [24].

Despite their limitations, all these studies suggest that a significant improvement in outcomes can be observed at a threshold of 6 cases/year. This model could be tested in the future using a prospective database recording details of surgery, comorbidities, and 30-day complication rates stratified on severity (e.g. Clavien scale).

It is impossible to show whether unusual or very severe complications occur in patients treated in low-volume centres. Such data does not exist in the literature. Reports about significant intraoperative vascular lesions address problem solving during minimal invasive procedures. For example, Morelli et al. reported two vascular lesions (one vena cava and one left renal vein damage) among intraoperative complications in RtA for large adrenal tumours managed by using sutures without the necessity to convert to laparoscopic or open surgery [54]. Bleeding has been the most common reported complication of adrenalectomy. However, a number of high-grade complications of adrenalectomy not typically reported in the literature have come to light in the minimally invasive era including injuries to the porta hepatis that have resulted in liver failure and have led to liver transplantation, injuries to the renal vessels or ureter requiring nephrectomy, and inappropriate organ resection, particularly mistaken resection of the tail of the pancreas instead of the adrenal gland [55].

The question of the minimal annual workload has been discussed for many years of many surgical specialities and a large amount of literature has been published [74]. A Pubmed search using the keywords ‘operative volume’ and ‘adrenalectomy’ led to 40 papers.

A multicentre study from the New York Statewide Planning and Research Cooperative System from 2000–2014 included 2839 adrenalectomies performed by 462 urologists (47%), 1098 by 23 endocrine surgeons (18%), and 2117 by 599 general surgeons (35%). Median annual surgeon volume was 1 case (IQR 1–2) with a mean of 2.1 cases, and a range of 1 to 29 cases. This emphasises that most of adrenalectomies in this area were performed by surgeons doing less than 3 cases per year. This trend is likely similar in Europe and this was confirmed in one study from the UK [24] reporting 795 adult adrenalectomies performed by 222 different surgeons with a range of between 1 and 34 adrenalectomies performed per surgeon. Only thirty-six (16%) adrenal surgeons performed 6 or more adrenalectomies. A total of 186 surgeons (84%) performed a median of one adrenalectomy a year. Overall, this shows that the number of adrenalectomy per year and per surgeon is very low.

This low average number of cases per year and per surgeon also infers that it has been difficult to define a meaningful/relevant cut-off number associated with improved perioperative outcomes. Overall, there is no validated consensus to define minimum annual workload in relation with improved perioperative outcomes. However, data from two studies support a minimum annual workload of at least 6 adrenalectomies per surgeon and per year.

There is a significant relationship between minimal annual workload and costs. As funding principles vary widely in the health care systems of different countries, direct comparison is impossible and extrapolation is not valid. It remains likely that financial models of reimbursement have an impact on the feasibility of a specific new technique and could influence the decision to adopt new techniques.

In a recent study, patients undergoing resection by low-volume surgeons (< 6 cases/year) had an increased cost (+ 26.2%, 95% confidence interval, 12.6–39.9, P = .02) [53]. Using data on 7045 patients from the Nationwide Inpatient Sample (2003–2009), some calculated that if all operations performed by low-volume surgeons (1 case/year) were selectively referred to intermediate-volume surgeons (2–5 cases/year), a 7.7% cost savings would have been incurred. Potential savings were even higher (8.1%) if the operations had been performed by the high-volume surgeons (≥ 7 cases/year). With the conservative assumption that there are 5000 adrenalectomies per year in the USA, the high-volume surgeons would produce savings of $8.8 million over a span of 14 years [75].

In a similar analysis of 6416 patients who underwent adrenalectomy in the Healthcare Cost and Utilization Project Nationwide Inpatient Sample (HCUP-NIS) from 2003 to 2008, patients 61–70 years (22% of the cohort) and > 70 years (13% of the cohort) had more complications (14% vs. 20% vs. 23%), longer mean length of stay (3.3 vs. 4.0 vs. 4.9 days; P < 0.001) and higher mean costs ($12,307 vs. $13,226 and $14,649; P < 0.001). As older age seems to be independently associated with adverse short-term clinical and economic outcomes after adrenalectomy, the authors suggested that enhanced access to high-volume surgeons is a potentially modifiable factor of particular importance in these patients [76].

The financial concerns related to delivery of adrenal surgery are more significant when considering robotic surgery. Widespread adoption of robotic technology has positioned robotic adrenalectomy as an option in some medical centres. Speculative advantages associated with the use of the robotic system have rarely been evaluated in clinical settings and cost increase remains an important drawback associated with robotic surgery. The higher cost of robotic adrenalectomy is a documented drawback of the procedure. It was calculated that robotic adrenalectomy was 2.3 times more expensive than laparoscopic adrenalectomy [47]. Increasing the number of robotic cases performed per year and depreciation of robotic system were more effective in cost reduction compared with decreasing operative time. Winter et al. calculated in the USA that if a centre performs over 500 robotic operations per year, then capital and maintenance costs for the robot would be $380 per procedure [77]. From this standpoint, it has been considered that robotic adrenalectomy could become more affordable in high-volume robotic surgery centres in the future [78].

In 2018, Feng Z et al. reviewed their experience and strategies to reduce the cost of robotic adrenalectomy (RtA) comparing it with laparoscopic adrenalectomy (LA). The calculated relative costs were similar: $3527 for RtA and $3430 for LA. The average consumables fees were $1106 for RA vs. $1009 for LA (P = 0.62). The average postoperative hospital stay was similar (1.7 days vs.1.9 days). It appears that limiting the number of robotic instruments and energy devices while utilising an experienced surgical team can keep the costs of RtA comparable with those of LA [79].

Overcosts due to the use of the robotic system could also be balanced by hospital stay decrease, patients' referral increase, improved postoperative outcomes in more difficult patients, and ergonomics for the surgeon. However, the current surgical intuitive business model is counterproductive, because there are no available strong clinical data that could balance overcosts associated with the use of the robotic system [80].

There is no available data published on the reimbursement of different adrenalectomy techniques (i.e. retroperitoneal, transperitoneal, robotic) from different health care providers. A comparison between different countries is not available. In France, robotic adrenalectomy has the same reimbursement than conventional laparoscopic adrenalectomy (flat reimbursement). The posterior approach and transperitoneal approach have also the same amount for cost reimbursement. Consequently, this flat reimbursement rate does not favour the use of new techniques (ex robotic platform) because they are often more expensive. Currently South Korea being the only country having a higher reimbursement rate for robotic surgery.

All available techniques for adrenalectomy can be divided into two groups. The first one is open adrenalectomy and in 2018, this approach is (or should be) used only in patients with large tumours and/or with local extension. The decision of using laparotomy is made by necessity when resection of large adrenal tumours and/or with local extension is technically too difficult using a laparoscopic approach or when the laparoscopic approach leads potentially to decreased postoperative patients survival (e.g. capsular effraction). This decision is very subjective and depends upon personal experience and expertise. The threshold in terms of tumour diameter to define a large tumour is also not consensual and would be ranging from 6 to 12 cm (depending on experience, surgeon and first aid skill, collaborating endocrinologists, anesthesiologists, use of an updated laparoscopic or robotic platform). There is no risks/benefits balance to be performed when choosing an open approach because this choice is imposed by clinical factors specific to each patient.

The second group represents all minimally invasive approaches with two main options: laparoscopic transabdominal and laparoscopic posterior adrenalectomy. Perioperative results are similar between these two techniques with the exception that tumours larger than 6 cm are more difficult to be resected using the posterior approach. Risks vs. benefits evaluation for these two approaches are likely similar in patients with tumours < 6 cm. For larger tumours, the feasibility of transabdominal laparoscopic approach is likely superior.

The robotics system allows to perform both transperitoneal and posterior approaches. There is the same feasibility issue for posterior robotic approach (than conventional posterior laparoscopic approach) in patients with tumours > 7 cm. Conventional laparoscopy and robotic approaches have similar results in patients with usual tumours. The robotic system may be superior in patients with more complex and difficult cases (tumour > 8 cm, right side, need to control inferior vena cava, large phaeochromocytoma or paraganglioma). This remains to be validated in future studies

The role of the European database EUROCRINE is crucial (www.eurocrine.eu). Eurocrine has been designed to evaluate activity and quality in endocrine surgery in each participating European country. The minimum dataset has been discussed extensively and is already running within the database. Audits have also been anticipated for this database. In the coming years, it is likely that this large dataset will facilitate significant studies in this field.

There are significant uncertainties that could be addressed in multicentre cohort studies, such as the following:

The ESES membership considered the information outlined in this manuscript and debated the issues raised during the ESES2019 meeting. This overall consensus provides guidance for the development of adrenal surgery in Europe in the next decade.

1.a. ESES recommends that laparoscopic adrenal surgery should be performed only in units with a predicted annual workload of 6 or more cases/year (level of evidence III/grade of recommendation B).

Though this number might appear surprisingly low, achieving this aim would bring significant restructuring of the current service provision for adrenal surgery. It is deemed to be a realistic and feasible goal and each national society should explore ways of addressing this issue.

1.b. Open adrenalectomy for high-risk patients (i.e. suspected or diagnosed adrenocortical cancer or phaeochromocytomas) should be offered only in units with a track record of involvement in the care of such patients and with a workload of minimum of 12 cases/year (level of evidence III/grade of recommendation B).

2.a. For the ‘novice’ adrenal surgeon, laparoscopic transabdominal surgery should be the first choice as it can be offered to the widest spectrum of adrenal conditions and it can be ‘mastered’ within 20–40 cases (level of evidence III/grade of recommendation C).

2.b. Retroperitoneoscopic surgery should be adopted by surgeons with previous experience in transabdominal adrenalectomy and with a high-volume practice so that sufficient eligible patients can be encountered within a reasonable time during their personal learning curve (level of evidence III/grade of recommendation C).

2.c. Robotic adrenal surgery should be implemented only in units with previous experience in robotic surgery, familiar with laparoscopic adrenalectomy and with large-volume practice. Robotic adrenalectomy is not a technique for the occasional adrenal surgeon with minimal previous personal experience (level of evidence V/grade of recommendation C).

Adrenal surgery should be offered only in centres with capacity for a valid multidisciplinary team consisting of oncologist, surgeon, endocrinologist, radiologist (level of evidence III/grade of recommendation B).

The complexity of medical care of many patients with adrenal pathology, the potential need for multi-specialty input in some adrenal operations (e.g. IVC involvement, simultaneous resections of small-volume metastatic disease), the need for accurate pathology reporting, and the benefits of oncological opinion regarding adjuvant or palliative chemotherapy should restrict adrenal surgery to medical institutions where such multidisciplinary input can be provided.

The governing principles for defining a referral centre should be based on (i). sufficient annual workload (> 12 cases/year), (ii) presence of an experienced multidisciplinary team and (iii) involvement of certified surgeons. Good clinical practice demands the need for clinical governance, the requirements of obtaining informed consent based on quoting personal incidence of untoward events (e.g. morbidity, conversion to open operation, intraoperative bleeding).

The increasing scrutiny of medical profession by governing bodies and patients should trigger a change in the current delivery of adrenal surgery. Each EU country will have to define the mechanism of establishing referral centres based on advice from national surgical and medical professional societies in collaboration with ESES and ENSAT. Because the variation in health care systems and mechanisms for reimbursement vary between countries, the models and process to reach such a change in practice is expected to differ between countries.

Clinical information regarding adrenalectomies should be recorded prospectively and contribution to the established EUROCRINE and ENSAT databases is strongly encouraged.

The ESES recommends that surgeons who want to develop an interest in adrenal surgery should seek training in reputable units, should aim to secure mentorship from an establish surgeon, should liaise with other specialties (endocrinology, radiology, pathology) to secure the development of an integrated multidisciplinary clinical service, and should monitor their personal learning curve through accurate data collection and audit.

Training of junior doctors with an interest in endocrine surgery and adrenal surgery is discussed in a separate document [83]