Quality improvement in gastrointestinal surgical oncology with American College of Surgeons National Surgical Quality Improvement Program Donald J. Lucas, MD, MPH,a and Timothy M. Pawlik, MD, MPH, PhD,b Bethesda and Baltimore, MD

Objective. To assess the impact of American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) participation on outcomes in gastrointestinal surgical oncology. Study design. A total of 6,076 resections for esophageal, gastric, pancreatic, hepatobiliary, and colorectal cancers at 316 hospitals from the 2006 to 2011 ACS NSQIP were examined. Thirty-day complication rates were analyzed longitudinally over time with the use of multiple regression; we adjusted for operation type and preoperative risk factors. Results. The procedure mix was 3% esophagectomy, 5% gastrectomy, 16% pancreatectomy, 4% hepatectomy, 63% colectomy, and 9% proctectomy. Median age was 66 years, and 52% were male, 41% were American Society of Anesthesiologists class 2, and 52% were American Society of Anesthesiologists 3. Depending on anatomic surgical site, 21–45% of patients experienced a postoperative complication and 1.1–4.4% died. The incidence of patients with any complication decreased from 28 to 24% over the period (risk-adjusted odds ratio 0.95 per year, 95% confidence interval 0.94–0.96). In contrast, there was no substantial change in risk-adjusted mortality over the period (odds ratio 1.03, 95% confidence interval 0.99–1.07). Conclusion. There was a decrease in complications over time for ACS NSQIP participants in gastrointestinal surgical oncology, but mortality did not decrease. (Surgery 2014;155:593-601.) From the Department of Surgery,a Walter Reed National Military Medical Center, Bethesda; and Department of Surgery,b Johns Hopkins University School of Medicine, Baltimore, MD

COMPREHENSIVE QUALITY IMPROVEMENT in surgical oncology must address perioperative morbidity and mortality as well as long-term recurrence and survival.1 The literature on quality improvement in surgical oncology has focused on specialization and the relationships between hospital or surgeon volume and outcomes.2-4 It is clear that highvolume, specialized centers have better outcomes, but despite more than a decade of literature The views expressed in this paper are those of the authors and do not represent the official policy of the US Navy, Department of Defense, or US Government. The authors have no other sources of support or conflicts to report. Accepted for publication December 6, 2013. Reprint requests: Timothy M. Pawlik, MD, MPH, PhD, Professor of Surgery and Oncology, Chief, Division of Surgical Oncology, John L. Cameron Professor of Alimentary Surgery, Department of Surgery, Johns Hopkins Hospital, 600 N. Wolfe Street, Blalock 688, Baltimore, MD 21287. E-mail: [email protected] 0039-6060/$ - see front matter Ó 2014 Mosby, Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2013.12.001

espousing regionalization in surgical cancer care, up to one in three major oncologic resections are still performed at low-volume centers.5 A combination of patient desire for local care, market pressures for smaller hospitals to offer these services, and lack of market transparency will likely ensure that these operations continue to take place at low-volume centers for some time. Furthermore, data have suggested that most of the recent quality improvement in cancer care has been attributable to better outcomes at each volume level, rather than shifting patients from low- to high-volume centers.5 In addition, even at high-volume centers, the incidence of complications, especially after major oncologic operations, remains relatively high. Accordingly, quality improvement efforts in surgical oncology should focus not only on who is operating (eg, low- vs high-volume surgeon) but also how they are operating (eg, quality outcome metrics). One established way to assess perioperative outcomes in general surgery is through the American College of Surgeons National Surgical Quality SURGERY 593

594 Lucas and Pawlik

Improvement Program (ACS NSQIP), a voluntary hospital-level, risk-adjusted perioperative outcomes assessment. The ACS NSQIP program is expanding rapidly and distributed widely by geography and hospital type with more 438 current members in 42 states and three countries.6 Member hospitals receive semiannual reports detailing their risk-adjusted outcomes compared with all other de-identified member hospitals. A driving concept behind ACS NSQIP is that the information on outcomes will be used to direct improvement efforts relevant to the local context. ACS NSQIP was developed from a successful Veterans Affairs project,7 and a pilot study was performed from 2001 to 2004 in a few academic hospitals to test whether the model could be applied to nongovernment centers.8 Successful results in this study led to expansion of the program. In the most comprehensive study on improvement that used ACS NSQIP, investigators looked at all participating centers and all operation types from 2005 to 2007, but this was done early in the program, when there were far fewer member hospitals than today.9 A similar, more recent study was performed but was limited to member hospitals of a single-state quality collaborative.10 No study has specifically assessed the effect of ACS NSQIP participation in major gastrointestinal (GI) oncology operations. Furthermore, with rapid expansion of the NSQIP program, a contemporary investigation of the impact of NSQIP on quality metrics is important to ensure that improvement is replicated on a wide scale beyond the initial cohort. To determine whether there have been improvements in the perioperative outcomes of patients treated at institutions participating in ACS NSQIP, we investigated trends in the outcomes of patients undergoing major GI oncology procedures during a 6-year period. Specifically, we hypothesized that hospitals in ACS NSQIP would have decreased risk-adjusted morbidity and mortality over time for major GI oncology resections. METHODS ACS NSQIP collects data from member hospitals from a directed sample of surgical patients. Variables include demographics, indications, preoperative risk factors, procedural details, and 30-day postoperative morbidity and mortality. Data are collected in a standardized format from clinical records. These methods are described elsewhere.11,12 ACS NSQIP releases de-identified data for research in the Participant Use File (PUF). The level of analysis was the entire cohort of participating hospitals by calendar year. Tracking performance

Surgery April 2014

over time for individual hospitals would be ideal, but this is not possible with the PUF because hospital identifiers are specifically and intentionally not included. ACS NSQIP has been very careful to assure their anonymity such that these data are not used competitively across practices. This anonymity encourages participation. New hospitals join ACS NSQIP every year. The number of hospitals in the PUF increased from 37 in 2005 to 316 in 2011. Using information from ACS NSQIP and other sources,8,13 we created an estimate for the average number of years of ACS NSQIP participation per hospital for each year in the PUF (Table I). This approach did not account for hospitals that withdrew from ACS NSQIP or were excluded from the PUF in particular years because of failed data audits, but these effects were likely small. Estimated average years of participation decreased from 2005 to 2006, because a large number of new sites were added in 2006. Average years of participation increased every year afterwards through 2011. To have calendar year positively correlate with average years of ACS NSQIP experience, 2006–2011 were selected for analysis. Major GI cancer resections were identified using the international classification of diseases and current procedural terminology codes (Supplemental Table online version only). Widely metastatic cancer and emergent cases were excluded. Primary outcome measures were any complication and death. Secondary outcomes included all other ACS NSQIP complication types. Bleeding complications were excluded because of a substantial definition change in 2010. Statistical analysis was performed using Stata (StataCorp LP. College Station, TX). Changes in postoperative outcomes over time were assessed by the use of logistic and negative binomial multiple regression models as appropriate, adjusting for risk factors and operation type. ACS NSQIP reports expected morbidity and mortality probabilities for each patient with the use of a proprietary formula,13 but because these are recalibrated every 6 months, they are not accurate for risk-adjustment comparisons across calendar years and were not used in this analysis. Furthermore, as use of laparoscopy is not a preoperative patient characteristic but rather part of operative strategy---a mediator rather than a confounder of surgical outcomes14---it was not included in the risk-adjustment models. RESULTS Baseline patient characteristics. A total of 76,076 patients met inclusion criteria. Median age was 66 (interquartile range 56–76; Table II); 52.0% of

Lucas and Pawlik 595

Surgery Volume 155, Number 4

Table I. Estimated years of participation in ACS NSQIP Hospitals by years of participation, n Year

Hospitals, n



1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

3 3 18 18 18 28* 37 121 183 211 237 258 316

3 0 15 0 0 10 9 84 62 28 26 21 58

3 0 15 0 0 10 9 84 62 28 26 21


3 0 15 0 0 10 9 84 62 28 26


3 0 15 0 0 10 9 84 62 28


3 0 15 0 0 10 9 84 62


3 0 15 0 0 10 9 84


3 0 15 0 0 10 9


3 0 15 0 0 10


3 0 15 0 0


3 0 15 0


3 0 15


3 0


Average participation, y


0.0 1.0 0.3 1.3 2.3 2.1 2.4 1.0 1.3 2.0 2.7 3.4 3.6

*No information was available for 2004; an average of 2003 and 2005 was used. ACS NSQIP, American College of Surgeons National Surgical Quality Improvement Program.

patients were male. Racial composition was 72.3% non-Hispanic white, 9.7% black, 4.3% Hispanic, 3.7% Asian, and 10.1% other or unknown. Median American Society of Anesthesiologists (ASA) class was 3 (interquartile range 2–3). Common comorbidities included obesity (31.0%), diabetes (19.0%), current smoking (17.0%), heart disease (12.3%), and cerebrovascular disease (6.6%). Among all patients, 8.0% had substantial preoperative weight loss, 7.2% underwent neoadjuvant radiotherapy, and 2.9% underwent neoadjuvant chemotherapy. Because of the large sample size, several risk factors had statistically significant changes across years, but most showed no dramatic clinically relevant trends. There was an increase in ASA class, with ASA 2 decreasing from 44.6% to 38.7% and ASA 3 increasing from 48.3% to 53.9% from 2006 to 2011 (P

Quality improvement in gastrointestinal surgical oncology with American College of Surgeons National Surgical Quality Improvement Program.

To assess the impact of American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) participation on outcomes in gastrointe...
390KB Sizes 0 Downloads 0 Views

Recommend Documents