Objective
The aim of this study was to determine the impact of gynecological surgeon volumes on patient outcomes.
Data Sources
Eligible studies were selected through an electronic literature search from database inception up until September 2015 and references in published studies. Search terms included surgical volume, surgeon volume, low-volume or high-volume, and gynecology or hysterectomy or sling or pelvic floor repair or continence procedure.
Study Eligibility
The literature search was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We defined a low-volume surgeon (LVS) as one performing the procedure once a month or less, and studies were excluded if their definition of LVS was > ±33% of our definition. Primary outcomes were total complications, intraoperative complications, and postoperative complications.
Study Appraisal and Synthesis Methods
All outcome data for individual studies were entered into systematic review software. When 2 or more studies evaluated a designated outcome, a meta-analysis of the entered data was undertaken as per the Cochrane database methodology. Data analysis was entered into a software product, which generated a summary of findings table that included structured and qualified grading (very low to high) of the quality for the evidence of the individual outcomes and provided a measure of effect.
Results
Fourteen peer-reviewed studies with 741,760 patients were included in the systematic review. For gynecology the LVS group had an increased rate of total complications (odds ratio [OR], 1.3, 95% confidence interval [CI], 1.2–1.5), intraoperative complications (OR, 1.6, 95% CI, 1.2–2.1), and postoperative complications (OR, 1.4 95% CI, 1.3–1.4). In gynecological oncology, the LVS group had higher mortality (OR, 1.9, 95% CI, 1.3–2.6). In the urogynecology group, a single study reported that the LVS group had a higher rate of any complication (risk ratio [RR], 1.4, 95% CI, –1.2-1.6). Another single study found that LVS had higher rates of reoperation for mesh complications after midurethral sling procedures (RR, 1.4, 95% CI, 1.2–1.5). The evidence is of moderate to very low quality.
Conclusion
Gynecologists performing procedures approximately once a month or less were found to have higher rates of adverse outcomes in gynecology, gynecological oncology, and urogynecology, with higher mortality in gynecological oncology.
Related editorial, page 1 .
In his popular book Outliers , Malcolm Gladwell famously argued that the mastering of any skill requires 10,000 hours, or 20 hours a week for 10 years, of deliberate practice. It seems a feasible theory that to obtain a skill with a repetitive technique, be it surgery or landing a plane, recurrent practice is required. In Australia, airline pilots, for example, are required to land at least 3 times every 90 days to maintain their proficiency certificates.
In many surgical fields, the relationship between surgical volume and outcome is well established. An American study of greater than 470,000 Medicare patients undergoing either cardiovascular procedures or cancer resections found that the operative mortality rate was strongly and inversely related to surgeon volume for each procedure.
Several papers have examined this relationship in the field of gynecology, specifically looking at midurethral slings, pelvic reconstructive procedures, hysterectomies for benign indications, myomectomies, and gynecological-oncological procedures and have reported conflicting results. A 2013 review article, without meta-analysis, of surgeon volumes and outcomes for benign hysterectomy concluded that morbidity was higher for low-volume surgeons and high-volume surgeons were more efficient.
The lifetime risk of undergoing major gynecological surgery in many developed countries is in the order of 15-40%. Estimating the association between adverse outcomes and risk factors that can potentially be addressed through practice or policy changes, such as surgeon volume, is an important public health concern.
We performed a systematic review and meta-analysis to determine whether gynecological surgeon volumes had an impact on patient outcomes. Our null hypothesis is that surgeon volume had no impact on surgical outcomes or surgical efficiency.
Materials and Methods
Eligibility
Eligible studies were selected through an electronic literature search from inception up until September 2015 using PubMed, the Cochrane Central Register of Controlled Trials, the Cochrane Database of Systematic Reviews, Medline, and clinicaltrials.gov . Search terms included the following key words: surgical volume, surgeon volume, low-volume or high-volume, gynecology or hysterectomy or sling or pelvic floor repair or continence procedure. There were no exclusion criteria for language or geographic location. Additional records were identified from references of articles identified through database searching.
Study selection
The literature search was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and included prospective and retrospective studies that compared surgical complications or markers of surgical efficiency between high-volume surgeons (HVS) and low-volume surgeons (LVS) for any major gynecological procedure. All the studies involved women older than 18 years of age undergoing major gynecological surgery.
We defined a low-volume surgeon as one performing the procedure once a month or less (12 or fewer procedures a year), and studies were excluded if their definition of an LVS was > ±33% of our definition (range, 8–16). Studies that divided surgeons into low-, medium-, and high-volume surgeons were included if 2 of the groups could be merged to fit our inclusion criteria.
Primary outcomes were total complications, intraoperative complications, and/or postoperative complications. Secondary outcomes were mortality, medical complications, cystotomy, ureteric injury, bowel injury, vascular injury, transfusion rates, operating time, length of stay, estimated blood loss, readmission rates, and reoperation rates.
For the primary outcome of total complications, studies were included if they provided data for an outcome of total complications, total morbidity, or any complication. Intraoperative complications included ureteric, bladder, bowel, vascular, and other intraoperative injuries.
Postoperative complications included wound complications (including vault hematoma), hemorrhage, ileus, bowel obstruction, and venous thromboembolism. Medical complications included cardiopulmonary arrest, stroke, respiratory failure, pneumonia, renal failure, gastrointestinal complications, sepsis, fever, and urinary tract infections. Papers that did not include any of these outcomes were excluded.
Gynecology has a wide range of surgical interventions and analysis was divided into gynecology, gynecological oncology, and urogynecology. Before data extraction the review was registered with PROSPERO International Prospective Register of Systematic Reviews (registration number CRD42015026154).
Data extraction and analysis
Data extraction was undertaken independently by 2 reviewers to ensure accuracy. If disagreement occurred, a decision was made by mutual agreement. Outcome data for individual studies were entered into Review Manager 5.3 systematic review software. When 2 or more studies evaluated, a designated outcome meta-analysis was performed as per the Cochrane methodology.
For the analysis of the categorical variables, we calculated the odds ratio (OR; odds of women with a certain outcome in relation to the odds of women without the outcome in the group). For continuous variables, we used means and SDs to derive a mean difference (MD). Unless otherwise stated, the outcomes in this meta-analysis were calculated from the raw data reported in the papers and presented without adjustment for confounders. Where possible, adjusted ORs and risk ratios were combined to give outcomes adjusted for possible confounders, including patient age, body mass index (BMI), and comorbidities. If there was significant heterogeneity in the outcomes recorded in different studies as defined by the I 2 calculation being greater than 75%, a random-effects model was used; otherwise, a fixed-effect model was used for the calculation of summary estimates and their 95% confidence intervals (CI).
Data analysis was entered into GRADEpro software, which generated a summary of findings (SOF) table that included structured and qualified grading (very low to high) of the quality for the evidence of the individual outcomes and provided a measure of effect.
Results
Study selection and characteristics
A total of 2151 abstracts met the initial search criteria. Of those, 2123 were excluded by reviewers because they did not meet the predefined criteria. Twenty-eight full articles were assessed for eligibility and 14 were excluded for not meeting the defined inclusion criteria as outlined in the PRISMA flow study ( Figure ).
Fourteen peer-reviewed studies from 3 countries (The Netherlands, The United States, and Canada) with a total of 741,760 patients were included in the systematic review. The 2 urogynecology studies were unable to be combined, so 12 studies were combined in the meta-analysis. Two studies (Wallenstein et al, Rogo-Gupta et al ) potentially included the same patients from the Premier (Perspective) database between 2004 and 2007 for the 3 outcomes intraoperative complications, postoperative complications, and medical complications. To minimize the risk of duplication, data from Rogo-Gupta et al was excluded from the analysis for these 3 outcomes.
In the gynecology group, 5 studies evaluated hysterectomy and 2 evaluated myomectomy. In the gynecological oncology group, 3 studies reported on endometrial cancer and 2 on ovarian cancer. In the urogynecology group, 1 study evaluated pelvic reconstructive surgeries and another evaluated reoperation rates after midurethral sling surgery.
Patient characteristics including age and comorbidities were reported in 9 of the 14 articles, and in 6 studies, the HVS group had older women and/or women with more comorbidities, in 1 study the LVS group had older women with more comorbidities, and in 2 studies the preintervention groups were similar ( Table 1 ).
Study | Type of study | Subgroup/procedure/indication | Procedures, n | Surgeons, n | Volume definitions (n/y) | Preintervention patient characteristics in the HVS and LVS groups | Outcomes |
---|---|---|---|---|---|---|---|
Vree et al, 2014, The Netherlands | Retrospective Single institution | Gynecology/Hysterectomy/ All benign excluding obstetric indications | 1914 | 83 | LVS < 11 MVS 11–50 HVS > 50 | HVS group had older patients with more comorbidities | Any in-hospital complications Postoperative in-hospital complications Operating time (min) LOS (mean number of days) EBL (mL) Readmission |
Wallenstein et al, 2012, United States | Retrospective National database (multiinstitutional, fee supported) | Gynecology/Laparoscopic hysterectomy/all benign indications a | 124,615 | 7925 | LVS < 5.88 MVS 5.88–14.1 HVS > 14.1 | LVS group had older patients with more comorbidities | Any in-hospital complications Intraoperative complications Postoperative in-hospital complications Medical complications Death Blood transfusion Reoperation rate LOS (> 2 d) Cystotomy Ureteric injury Intestinal injury Vascular injury |
Rogo-Gupta et al, 2010, United States | Retrospective National database (multiinstitutional, fee supported) | Gynecology/vaginal hysterectomy/all benign indications | 77,109 | 6195 | LVS < 5.4 MVS 5.4–13 HVS > 13 | HVS group had older patients with more comorbidities | Any in-hospital complications Intraoperative complications Postoperative in-hospital complications Medical complications Death Operating time (min) LOS (mean number of days) Blood transfusion Readmission cost |
Boyd et al, 2010, United States | Retrospective Statewide database (multiinstitutional) | Gynecology Hysterectomy | 146,494 | 4511 | LVS < 10 HVS ≥ 10 | No comment | Any in-hospital complications Intraoperative complications Postoperative in-hospital complications Medical complications Death Blood transfusion Reoperation rate LOS (> 2 d) Cystotomy Ureteric injury Intestinal injury |
Betjes et al, 2009, United States | Retrospective Single institution | Gynecology/abdominal myomectomy/leiomyoma-mass, abnormal bleeding, infertility | 415 | 36 | LVS < 15 HVS ≥ 15 | Groups had similar mean age | Operating time |
Hanstede et al, 2009, United States | Retrospective Single institution | Gynecology/hysterectomy/all benign indications | 7166 | 214 | LVS < 10 HVS ≥ 10 | HVS group had older patients with more comorbidities | Any in-hospital complications Intraoperative complications Postoperative in-hospital complications Medical complications Operating time (min) Blood transfusion Readmission LOS (mean number of days) Urinary tract injury Intestinal injury Vascular injury |
Hanstede et al, 2008, United States | Retrospective Single institution | Gynecology/abdominal myomectomy/leiomyoma-mass, abnormal bleeding, infertility | 527 | 43 | LVS ≤ 10 HVS > 10 | HVS group had older patients with more comorbidities | Operating time (min) EBL (mL) Blood transfusion |
Wright et al, 2012, United States | Retrospective National database (multiinstitutional, fee supported) | Gynecological oncology/laparoscopic hysterectomy/endometrial cancer | 4137 | Not stated | LVS ≤ 2.8 MVS 2.81–8.0 HVS > 8.0 | HVS group had older patients with more comorbidities | Mortality Any in-hospital complications Intraoperative complications Postoperative in-hospital complications Medical complications Operating time (min) Blood transfusion Reoperation LOS (> 2 d) Cost Cystotomy Ureteric injury Intestinal injury Vascular injury |
Wright et al, 2011, United States | Retrospective National database (multiinstitutional, fee supported) | Gynecologic oncology/open abdominal hysterectomy/endometrial cancer | 6015 | Not stated | LVS < 14.5 MVS 14.6–30 HVS > 30 | Groups had similar comorbidities | Mortality Any in-hospital complications Intraoperative complications Postoperative in-hospital complications Medical complications Blood transfusion Reoperation LOS (> 2 d) Cost Cystotomy Ureteric injury Intestinal injury Vascular injury |
Vernooij et al, 2009, The Netherlands | Retrospective Nationwide, data collected from random sample medical records | Gynecological oncology/laparotomy/ovarian cancer | 1077 | not stated | LVS ≤ 6 MVS 7–12 HVS > 12 | No comment | 5 y survival |
Bristow et al, 2009, United States | Retrospective Statewide database (multiinstitutional) | Gynecological oncology/laparotomy/ovarian cancer | 1894 | 352 | LVS <10 HVS ≥ 10 | No comment | Mortality LOS (mean number of days) Cost |
Diaz-Montes et al, 2006, United States | Retrospective Statewide database (multiinstitutional) | Gynecological oncology/hysterectomy/endometrial cancer | 6181 | 894 | LVS < 8.3 HVS ≥ 8.3 | No comment | Mortality |
Welk et al, 2015, Canada | Retrospective Three national databases (population based) | Urogynecology/reoperation for mesh complication after midurethral sling | 59,887 | 1068 | LVS < 16 HVS ≥ 16 | HVS had older patients and had higher risk patients (previous history of fistula, urethral diverticulum, urethral injury, or radiotherapy) | Reoperation for stress urinary incontinence mesh-related complications |
Sung et al, 2006, United States | Retrospective National database (multiinstitutional, fee supported) | Urogynecology/urogynecological procedures/pelvic organ prolapse or urinary incontinence | 310,759 | Not stated | LVS < 8 MVS 8–18 HVS > 18 | No comment | Any in-hospital complication Nonroutine discharge |
a Endometriosis, leiomyomas, menorrhagia/bleeding, ovarian/adnexal mass, pelvic organ prolapse.
Synthesis of results
Low-volume surgeon vs high-volume surgeon and outcomes in gynecology
Total in-hospital complications
Low-volume surgeons had a higher rate of total in-hospital complications than high-volume surgeons as reported in 4 studies (OR, 1.3, 95% CI, 1.2–1.5, Table 2 ). This means that if in-hospital complications occur in 97 per 1000 patients in the HVS group, between 114 and 137 per 1000 patients in the LVS group would develop in-hospital complications.
Outcomes | Risk in high-volume surgeon group | Risk in low-volume surgeon group | Relative effect (95% CI) | Participants (studies), n | Quality of the evidence (grade) |
---|---|---|---|---|---|
Total complications | 97 per 1000 | 125 per 1000 (114–137) | OR, 1.3 (1.2–1.5) | 283,119 (4 studies) | ⊕⊝⊝⊝ Very low 1 |
Total complications adjusted OR | OR, 1.4 (1.3–1.5) | 153,660 (2 studies) | ⊕⊕⊕⊝ Moderate 2 | ||
Total complications adjusted OR excluding gynecological oncology | 68 per 1000 | 167 per 1000 (133–208) | OR, 2.8 (2.1–3.6) | 3427 (1 study) | ⊕⊕⊕⊝ Moderate 3 |
Intraoperative complications | 22 per 1000 | 35 per 1000 (26–45) | OR, 1.6 (1.2–2.1) | 358,296 (3 studies) | ⊕⊝⊝⊝ Very low 4 |
Intraoperative complications adjusted OR | OR, 1.8 (1.1–3.2) | 84,275 (2 studies) | ⊕⊕⊕⊝ Moderate 2 | ||
Intraoperative complications adjusted OR excluding gynecological oncology | 15 per 1000 | 50 per 1000 (30–83) | OR, 3.4 (2.0–5.9) | 3427 (1 study) | ⊕⊕⊕⊝ Moderate 3 |
Postoperative complications | 39 per 1000 | 52 per 1000 (50–54) | OR, 1.4 (1.3–1.4) | 359,528 (4 studies) | ⊕⊕⊕⊝ Moderate 2 |
Postoperative complications adjusted OR | OR, 1.5 (1.2–1.7) | 84,275 (2 studies) | ⊕⊕⊕⊝ Moderate 2 | ||
Postoperative adjusted OR excluding gynecological oncology | 53 per 1000 | 117 per 1000 (89–153) | OR, 2.4 (1.8–3.2) | 3427 (1 study) | ⊕⊕⊕⊝ Moderate 3 |
1 Imprecision: significant heterogeneity with an I 2 = 85% and downgraded the quality rating by 1 level
2 Plausible confounder: inclusion of gynecological oncologists diminishes the volume and upgraded the quality rating by 1 level
3 Magnitude effect: the magnitude of effect was large (OR, ≥ 2) and upgraded the quality rating by 1 level
4 Imprecision: significant heterogeneity with an I 2 = 96% and downgraded the quality rating by 1 level.
Two studies provided data for total in-hospital complications adjusted for age and comorbidities, and the increased risk of any in-hospital complication in the LVS group was slightly greater (OR, 1.4 95% CI, 1.3–1.5, Table 2 ). On a number-needed-to-treat analysis, this translated to 1 in-hospital complication being avoided for every 30 operations that were performed by an HVS rather than an LVS. Hanstede et al reanalyzed their data excluding gynecological oncologists and reported a greater difference between the LVS and HVS groups (OR, of 2.8, 95% CI, 2.1–3.6, adjusted for age and comorbidities). On a number-needed-to-treat analysis, this translated to 1 in-hospital complication being avoided for every 10 operations that are performed by an HVS rather than an LVS.
Intraoperative complications
Three studies reported on this outcome, and the LVS group had a higher rate of intraoperative complications compared with the HVS group (OR, 1.6, 95% CI, 1.2–2.1, Table 2 ). This means that if intraoperative complications occur in 22 per 1000 patients in the HVS group, between 26 and 45 per 1000 patients in the LVS group would develop intraoperative complications. Two studies calculated an OR adjusted for age and comorbidities, and a further increase in the risk of intraoperative complications was seen in the LVS group compared with the HVS group; (OR, 1.8, 95% CI, 1.1–3.2, Table 2 ). On a number-needed-to-treat analysis, this translates to 1 intraoperative complication being avoided for every 38 operations that are performed by an HVS rather than an LVS surgeon.
Hanstede et al reanalyzed their data, without gynecological oncologists, and the greater risk of intraoperative complications in the LVS group was more evident (OR, 3.4, 95% CI, 2.0–5.9, Table 2 ). On a number-needed-to-treat analysis, this translates to 1 intraoperative complication being avoided for every 20 operations that are performed by an HVS rather than an LVS.
Postoperative complications
Four studies reported on this outcome, and the LVS group had a higher rate of postoperative complications compared with the HVS group (OR, 1.4, 95% CI, 1.3–1.4, Table 2 ). This means that if postoperative complications happen in 39 per 1000 patients in the HVS group, between 5 and -54 per 1000 patients in the LVS group would develop postoperative complications.
Two studies calculated an OR adjusted for age and comorbidities, and a further increase in the risk of postoperative complications was seen when comparing the LVS and HVS groups (OR, 1.5, 95% CI, 1.2–1.7, Table 2 ). On a number-needed-to-treat analysis, this translates to 1 postoperative complication being avoided for every 41 operations that are performed by an HVS rather than an LVS.
Hanstede et al reanalyzed, without gynecological oncologists, and the difference between the 2 groups was more pronounced (OR, 2.4, 95% CI, 1.8–3.2, Table 2 ). On a number-needed-to-treat analysis, this translates to 1 postoperative complication being avoided for every 15 operations that are performed by an HVS rather than an LVS.
Mortality
There was no difference in mortality between the 2 groups.
Medical complications
Three studie reported on medical complications and found that medical complications were more common in the LVS group compared with the HVS group (OR, 1.6, 95% CI, 1.5–1.6, Table 3 ). This means that if medical complications happen in 79 per 1000 patients in the HVS group, between 115 and 122 per 1000 women in the LVS group would develop medical complications.
Outcomes | Risk in high-volume surgeon group | Risk in low-volume surgeon group | Relative effect (95% CI) | Participants (studies), n | Quality of the evidence (GRADE) |
---|---|---|---|---|---|
Mortality | 0 per 1000 | 1 per 1000 (0–2) | OR, 1.3 (0.4–4.7) | 351,148 (3 studies) | ⊕⊝⊝⊝ Very low 1 |
Medical complications | 79 per 1000 | 118 per 1000 (115–122) | OR, 1.6 (1.5–1.6) | 358,296 (3 studies) | ⊕⊕⊝⊝ Low |
Operating time, min | Mean operating time (min) 17.7 higher (10.4–25.0) | 9335 (4 studies) | ⊕⊝⊝⊝ Very low 2 | ||
Transfusion | 55 per 1000 | 56 per 1000 (37–84) | OR, 1.0 (0.7–1.6) | 234,203 (4 studies) | ⊕⊝⊝⊝ Very low 3 |
Estimated blood loss, mL | Mean estimated blood loss (mL) 59.3 higher (32.0–86.6) | 1754 (2 studies) | ⊕⊕⊝⊝ Low | ||
Cystotomy | 8 per 1000 | 9 per 1000 (8–12) | OR, 1.1 (0.9–1.4) | 273,949 (2 studies) | ⊕⊝⊝⊝ Very low 4 |
Ureteric Injury | 1 per 1000 | 2 per 1000 (1–2) | OR, 1.7 (1.4–2.1) | 274,039 (2 studies) | ⊕⊕⊝⊝ Low |
Cystotomy or ureteric injury | 9 per 1000 | 13 per 1000 (10–17) | OR, 1.4 (1.1–1.9) | 281,205 (3 studies) | ⊕⊝⊝⊝ Very low 5 |
Bowel injury | 12 per 1000 | 14 per 1000 (13–15) | OR, 1.1 (1.1–1.2) | 274,039 (2 studies) | ⊕⊕⊝⊝ Low |
Vascular injury | 0 per 1000 | 1 per 1000 (0–5) | OR, 2.2 (0.4–11.6) | 131,781 (2 studies) | ⊕⊝⊝⊝ Very low 2 |
Readmission | 20 per 1000 | 16 per 1000 (14–18) | OR, 0.8 (0.7–0.9) | 85,489 (3 studies) | ⊕⊝⊝⊝ Very low 6 |
Reoperation | 3 per 1000 | 2 per 1000 (2–3) | OR, 0.9 (0.7–1.2) | 124,615 (1 study) | ⊕⊕⊝⊝ Low |
LOS > 2 d | 63 per 1000 | 84 per 1000 (80–88) | OR, 1.4 (1.3–1.4) | 124,615 (1 study) | ⊕⊕⊝⊝ Low |
1 Imprecision: significant heterogeneity with an I 2 = 85% and downgraded the quality rating by 1 level
2 Imprecision: significant heterogeneity with an I 2 = 83% and downgraded the quality rating by 1 level
3 Imprecision: significant heterogeneity with an I 2 = 98% and downgraded the quality rating by 1 level
4 Imprecision: significant heterogeneity with an I 2 = 85% and downgraded the quality rating by 1 level
5 Imprecision: significant heterogeneity with an I 2 = 88% and downgraded the quality rating by 1 level
6 Imprecision: significant heterogeneity with an I 2 = 74% and downgraded the quality rating by 1 level.
Visceral and vascular injuries
Ureteric injury was reported in two studies and was more likely in the LVS group (OR, 1.7, 95% CI, 1.4–2.1, Table 3 ). Bowel injury was reported in 2 studies and the risk was higher in the LVS group (OR, 1.1, 95% CI, 1.1–1.2, Table 3 ).
Urinary tract injury (ureteric and bladder injury combined) was reported in 3 studies and was more likely in the LVS group (OR, 1.4, 95% CI, 1.1–1.9, Table 3 ).
There was no difference in risk of cystotomy or vascular injury ( Table 3 ).
Operating time
Four studies reported on this outcome and found that operating time was longer in the LVS group (MD, 17.7 minutes, 95% CI, 10.4–25.0, Table 3 ).
Estimated blood loss and transfusion rates
Two studies reported that estimated blood loss was greater in the LVS group (MD, 59.3 mL, 95% CI, 32.0–86.6 mL, Table 3 ). There was no difference in transfusion rates.
Length of stay
There was no difference in mean length of stay between the HVS group and the LVS group. Wallenstein et al found that women in the LVS group were more likely to stay in the hospital for more than 2 days (OR, 1.4, 95% CI, 1.3–1.4, Table 3 ). Boyd et al reported that women in the HVS group had a shorter length of stay by 0.4 days (95% CI, 0.4–0.5, adjusted for mode of hysterectomy, comorbidities, and postoperative complications).
Readmission rates
Readmission rates were reported in 3 studies and were lower for the LVS group (OR, 0.8, 95% CI, 0.7–0.9, Table 3 ). This means that if 20 in 1000 patients are readmitted in the HVS group, 14–18 in 1000 patients would be readmitted in the LVS group.
Reoperation rates
There was no difference in reoperation rates.
Low-volume surgeon vs high-volume surgeon and outcomes in gynecological oncology
Mortality
Mortality was reported in four studies and was higher in the LVS group compared with the HVS group (OR, 1.9, 95% CI, 1.3–2.6, Table 4 ). This means that if the mortality rate is 7 per 1000 patients in the HVS group, the rate would be between 9 and 18 per 1000 in the LVS group. Three of these studies adjusted the outcomes for age and comorbidities and the difference between the 2 groups became more significant (OR, 2.5, 95% CI, 1.7–3.8, Table 4 ). On a number-needed-to-treat analysis, this translates to 1 perioperative death being avoided for every 97 operations that are performed by a HVS rather than a LVS.