Keeping a Safe Distance From Surgical Volume Standards

The association between surgical volume and postoperative outcomes has been amply described in the surgical literature, with the assertion that higher-volume centers have better perioperative outcomes.¹ The premise for this stems from the notion that technical expertise (or surgery) is achieved and maintained only after a certain threshold of experience (or volume) is met, which in theory trickles down to other aspects of perioperative care (prehabilitation, postoperative management, etc).² As a result of this, several organizations and regulatory bodies have advocated for centralizing complex surgical procedures to centers and surgeons that meet certain volume minimums.³ For example, in 2015, several organizations committed to the so-called Take the Volume Pledge, restricting various surgical procedures to centers that meet subjective volume standards (eg, for lung cancer, ≥ 40 cases per year for hospitals and ≥ 20 cases per year for individual surgeons). However, a growing body of evidence has emerged that challenges the utility of the volume-outcome relationship as a reliable measure of surgical quality.⁴

THE TAKEAWAY

• In the article that accompanies this editorial, Baum et al⁵ found risk-standardized mortality rates to be a superior metric of surgical quality compared with volume-based metrics. These findings further challenge the policy of volume-based case thresholds for complex cancer surgeries.

In the article that accompanies this editorial, Baum et al⁵ compared volume-based standards versus risk-standardized mortality rates (RSMR; ie, the ratio of predicted to expected mortality on the basis of hierarchical logistic regression models) to assess hospital performance for complex cancer operations (for esophageal, lung, gastric, pancreatic, and colorectal cancer). Their analysis used a unique data set consisting of more than 480,000 patients treated in nearly 1,000 hospitals across Germany. Their primary end point was the number of patients who would need to be moved from a low-volume center to high-volume center or, conversely, a high-RSMR center to low-RSMR center to save one life. Importantly, the authors found significant variability between the RSMR- and volume-based rankings of hospitals. In other words, high-volume hospitals often had high mortality rates, and vice versa. Through impressive statistical techniques, the authors concluded that centralizing care to low-RSMR hospitals outperformed volume-based thresholds when analyzing the primary end point. Strikingly, the RSMR-based centralization model was also associated with lower median travel times for patients—even lower than observed travel times—which is an often-cited barrier to centralized surgical care. These findings question the notion of volume-based case minimums as a metric of surgical quality for complex cancer operations.

This article adds important knowledge to the large body of literature assessing hospital performance. Although the group's findings and conclusions are notable, several unanswered questions warrant further attention. First, it is unclear how to direct patients from low-performing centers to high-performing centers. The reality is that the medical referral system is often complicated and highly reliant on nonsurgeon providers who have little knowledge of surgical quality. Especially in the United States, designing policies and systems that use quality-based referrals will be difficult. Such systems will also have to consider access-to-care disparities, especially among patients who live far away from high-performing centers. Although travel times were found to be shorter on the basis of the RSMR model in this study, it is important to remember that Germany is approximately 30 times smaller than the United States, sparking concern as to whether such a system is possible in countries with a larger land mass. Second, it is notable from this study that the rates of postoperative mortality were high, especially compared with similar data sets in the United States. For example, the rates of in-hospital mortality were 2.9% and 7.2% for lung cancer and esophageal cancer resection, respectively. By contrast, the Society of Thoracic Surgery (STS) Database demonstrates operative mortality rates of 1.3% for lung cancer and 3.4% for esophageal cancer.^6,7 What should the expected risk-adjusted mortality rate for each of these procedures be? Finally, it is unclear whether structural differences between the German and US health care systems may differentially affect the findings observed in this study.

Although several studies have reported better outcomes among high-volume centers, alarm bells have been rung as to the statistical quality of these analyses.^2,8 Errors such as modeling volume as a categorical (as opposed to a continuous) variable; failing to report appropriate test statistics; ignoring reliability estimates; and disregarding hierarchical modeling techniques have led some to question the relationship between high-volume and better outcomes.^2,8 Indeed, more recent studies with appropriate statistical techniques have produced more muted conclusions regarding the volume-outcome relationship, with heterogeneous findings between procedures.⁹ For example, LaPar et al¹⁰ found no association between hospital procedure volume and mortality for pancreatectomy or esophagectomy. A recent study from the STS Database similarly found no association between the Volume Pledge standards and outcomes following lung cancer resection.⁴ Another study among patients undergoing esophagectomy found a weak correlation between the STS star rating system (a composite measure of hospital quality) and volume, with several low-volume hospitals achieving the highest star rating.⁷ Baum et al⁵ note that their findings do not necessarily “invalidate the notion of restricting care … to specific hospitals or surgeons”; however, it is reasonable to conclude from their analysis that volume standards alone are inadequate metrics for centralizing care.

Surgical quality is also important to understand in the context of the chosen outcome (ie, surgical v oncologic outcome). The study by Baum et al⁵ as well as most of the previously cited literature has assessed short-term mortality as the proxy metric for surgical quality (ie, surgical outcomes). Although undoubtedly important, it raises the question of why this metric is used. For example, operative mortality is an increasingly rare event for most cancer surgeries (eg, < 2% for patients with lung cancer receiving surgery in the United States die¹¹); does that mean that all patients who escape this fate (eg, > 98% of patients with lung cancer) have received a high-quality surgery? Similarly, how does one control for the reliability of hospital performance metrics in low-volume centers, where infrequent outcomes (like mortality) may not occur for years⁷? Perhaps a more prudent metric of surgical quality would be to assess long-term outcomes (ie, oncologic outcomes), since these are the outcomes that affect the majority of patients who survive the early postoperative period. For example, although operative mortality affects < 2% of patients with clinical stage I lung cancer, local recurrence rates can approach 20% despite curative-intent surgery.¹² Assessing a theoretically modifiable outcome such as local recurrence (via wider margins, meticulous intraoperative staging, etc) that affects far more patients may be a more appropriate metric of high-quality cancer care.

Lung cancer presents an interesting example of the paradox between short-term quality and long-term quality. It is well established that certain operations for lung cancer, such as sublobar resections, carry a lower risk of postoperative complications or mortality.⁶ It is similarly well established that sublobar resections carry a higher risk of recurrence and diminished overall survival.¹³ How does one value quality measures (such as receiving a sublobar resection) that affect short-term outcomes and long-term outcomes in a perfectly antithetical manner?¹⁴ Similarly, intraoperative lymph node sampling—a technically challenging and time-consuming process—has no effect on short-term outcomes but is associated with significantly worse disease-free survival when performed inadequately.¹⁵ Therefore, a surgeon who rapidly performs sublobar resections with inadequate lymph node sampling will certainly be high-volume and high-performing in terms of 30-day outcomes. However, if only short-term surgical outcomes are considered in quality assessments, then one could envision a worst-case scenario where patients are preferentially shunted to high-quality surgical centers with dismal long-term oncologic outcomes. It is important to avoid such myopic views of surgical quality.

Several groups have proposed more comprehensive metrics of surgical quality including preoperative and intraoperative metrics. For example, Samson et al¹⁶ assessed several quality metrics among patients with stage I non–small-cell lung cancer receiving surgery, including timely treatment, adequate lymph node sampling, negative margins, and anatomic resection. They found striking differences in overall survival depending on the number of quality metrics that were met. Composite metrics like this of hospital quality and surgeon quality that optimize both short-term outcomes and long-term outcomes are critical.

Evaluating surgeon and hospital quality is important. However, the reality is that a vast number of variables spanning the preoperative, perioperative, and postoperative periods affect surgical quality. Measuring each of these variables and how it relates to several short-term outcomes and long-term outcomes is far more complicated than a simple volume-outcome curve, which the article by Baum et al⁵ excellently highlights. The bottom line is this: patients should receive treatment from hospitals that perform high-quality surgery, not necessarily from hospitals that perform a lot of them.

AUTHOR CONTRIBUTIONS

Conception and design: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Keeping a Safe Distance From Surgical Volume Standards

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

No potential conflicts of interest were reported.