Customize your JAMA Network experience by selecting one or more topics from the list below.
Worni M, Schudel IM, Østbye T, et al. Worse Outcomes in Patients Undergoing Urgent Surgery for Left-Sided Diverticulitis Admitted on Weekends vs Weekdays: A Population-Based Study of 31 832 Patients. Arch Surg. 2012;147(7):649–655. doi:10.1001/archsurg.2012.825
Hypothesis Among patients undergoing urgent surgery for left-sided diverticulitis, those admitted on weekends vs weekdays have higher rates of Hartmann procedure and adverse outcomes.
Design Analysis of data from the Nationwide Inpatient Sample between January 2002 and December 2008. Unadjusted and risk-adjusted generalized linear regression models were used.
Setting Academic research.
Patients Data on patients undergoing urgent surgery for acute diverticulitis.
Main Outcome Measures Rates of Hartmann procedure vs primary anastomosis, complications, length of hospital stay, and total hospital charges.
Results In total, 31 832 patients were included; 7066 (22.2%) were admitted on weekends, and 24 766 (77.8%) were admitted on weekdays. The mean (SD) age of patients was 60.8 (15.3) years, and 16 830 (52.9%) were female. A Hartmann procedure was performed in 4580 patients (64.8%) admitted on weekends compared with 13 351 patients (53.9%) admitted on weekdays (risk-adjusted odds ratio [OR], 1.57; P < .001). In risk-adjusted analyses, patients admitted on weekends had significantly higher risk for any postoperative complication (OR, 1.10; P = .005) and nonroutine hospital discharge (OR, 1.33; P < .001) compared with patients admitted on weekdays, as well as a median length of hospital stay that was 0.5 days longer and median total hospital charges that were $3734 higher (P < .001 for both).
Conclusions Patients undergoing urgent surgery for left-sided diverticulitis who are admitted on a weekend have a higher risk for undergoing a Hartmann procedure and worse short-term outcomes compared with patients who are admitted on a weekday. Further research is warranted to investigate possible underlying mechanisms and to develop strategies for reducing this substantial weekend effect.
Diverticulosis is a common disease, with a prevalence among the general population exceeding 60% of individuals 80 years or older.1 Although diverticulosis is often asymptomatic, acute diverticulitis is a frequent cause of emergency admission. In the United States, more than 300 000 patients were admitted in 2006 because of diverticular disease,2 and every year about 20 000 patients undergo urgent surgery for the problem.3-5
Worse outcomes for patients admitted on weekends have been described for gastrointestinal hemorrhage,6,7 acute kidney injury,8 myocardial infarction,9 pulmonary embolism,10 and intracerebral hemorrhage.11 In 2001, Bell and Redelmeier12 reported that 23 of 100 leading causes of death were significantly more prevalent on weekends than on weekdays. This finding is supported by others. Ricciardi et al13 reported higher mortality among patients who were nonelectively admitted to hospitals on weekends compared with weekdays, while Clark and Normile14 reported higher overall mortality for patients who were admitted to the intensive care unit on weekends than on weekdays. A major reason for worse outcomes on weekends (the so-called weekend effect) is thought to be related to treatment delays from the onset of symptoms to hospital admission and finally to treatment.6,7,9,12,15-18 In times of limited resources, hospital routines are becoming more and more streamlined. To maintain high-quality patient care, it is important to explore potential outcome differences related to the day of admission, particularly for diseases, such as left-sided diverticulitis, that are common and associated with high morbidity.
The objective of this population-based study was to assess short-term outcomes among patients undergoing left-sided colon resection for diverticulitis who were admitted on weekends vs weekdays. Our hypothesis was 2-fold: (1) patients admitted on weekends have higher rates of undgergoing a Hartmann procedure compared with those admitted on weekdays, and (2) rates of adverse outcomes are higher among patients admitted on weekends compared with those admitted on weekdays.
The institutional review board at Duke University Medical Center approved this study. We used deidentified data from the Nationwide Inpatient Sample (NIS) between January 2002 and December 2008. The NIS is part of the Healthcare Cost and Utilization Project, sponsored by the Agency for Healthcare Research and Quality. It is the largest all-payer database in the United States on patients receiving inpatient care. Information about hospital charges is available for all patients, including those covered by Medicare, Medicaid, private insurance, and other. Annually, up to 8 million hospital stays are captured, representing approximately a 20% stratified sample of US community hospitals.19 The American Hospital Association defined the sample as “all non-Federal, short-term, general, and other specialty hospitals, excluding hospital units of institutions.”20 To achieve optimal representation of US hospitals, the registry is stratified for the following 5 hospital characteristics: ownership and control, teaching status, bed size, US region, and urban vs rural location. Up to 1100 hospitals are included each year.
Patients 18 years or older were included in our study. We identified patients undergoing urgent surgery for diverticulitis by a multistep process using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. First, we identified patients with diverticulitis of the colon (ICD-9-CM diagnosis code 562.11). Second, we identified patients undergoing the following colonic operations: laparoscopic left hemicolectomy (code 17.35), laparoscopic sigmoidectomy (code 17.36), open and other left hemicolectomy (code 45.75), open sigmoid resection (code 45.76), anterior resection of the rectum with synchronous colostomy (code 48.62), and other anterior resection of rectum without synchronous colostomy (code 48.63). Third, we excluded from the analysis those patients with an additional ICD-9-CM diagnosis code of malignant neoplasm of the colon (code 153.x), malignant neoplasm of the rectum, rectosigmoid junction, or anus (code 154.x), or inflammatory bowel disease (codes 555.0-556.9). Fourth, we maintained in the analysis only those patients with an emergency or urgent admission.
Patients were categorized as having been admitted on weekdays (Monday through Friday) vs weekends (Saturday or Sunday), a distinct variable in the NIS database that is calculated from the original admission date. Furthermore, patients were categorized as having undergone primary anastomosis vs a Hartmann procedure. Patients undergoing primary anastomosis were identified by an ICD-9-CM procedure code for anterior resection of the rectum (code 48.63) or left-sided colectomy (laparoscopic left hemicolectomy [code 17.35], laparoscopic sigmoidectomy [code 17.36], open left hemicolectomy [code 45.75], or open sigmoidectomy [code 45.76]); we excluded patients who received a terminal colostomy (colostomy [codes 46.1 and 46.10], temporary colostomy [code 46.11], or permanent colostomy [code 46.13]). Patients undergoing a Hartmann procedure were identified by an ICD-9-CM procedure code for anterior resection of the rectum with synchronous colostomy (code 48.62) or left-sided colon resection (codes 17.35, 17.36, 45.75, and 45.76), together with an ICD-9-CM code for colostomy formation (codes 46.1, 46.10, 46.11, and 46.13). For patients undergoing primary anastomosis, we also extracted information about ileostomy creation using the ICD-9-CM procedure codes for loop ileostomy (code 46.01) and temporary ileostomy (code 46.21). In addition, the time from admission to surgery was extracted for all patients; this was considered zero days if the operation occurred on the day of admission.
Adverse intraoperative and postoperative outcomes were assessed using ICD-9-CM codes. As described by Guller et al,21 we grouped postoperative complications into the following 7 categories: wound, infection, urinary or renal, pulmonary, gastrointestinal, cardiovascular, and systemic (Table 1). In addition, we extracted data on in-hospital mortality, length of hospital stay, total hospital charges (edited for accuracy by the NIS and inflation adjusted), and discharge status (routine vs nonroutine [skilled nursing facility, another short-term hospital, another type of facility, home health care, intermediate care facility, against medical advice, or other]).19 We also assessed the rate of reoperations or reinterventions using ICD-9-CM procedure codes 46.94, 54.12, 54.61, and 54.92 (Table 1).
We extracted demographic data from the NIS, including age, sex, year of operation, race/ethnicity (white, black, Hispanic, other, or unknown), annual household income related to individual zip code ($1-$34 999, $35 000-$44 999, or ≥$45 000), and primary insurance (Medicare, Medicaid, private [including health maintenance organization], self-pay, or other). Comorbidity was assessed using the Deyo score, a modified score related to the Charlson Comorbidity Index. We categorized patients by Deyo score (0, 1, 2, or >2). Hospital characteristics included volume (in quartiles), US region (Northeast, Midwest, South, or West), and teaching status and location (rural, urban nonteaching, urban teaching, or unknown).
Demographic data about patients are given as means (SDs) for continuous data and as counts and percentages for categorical data. Statistical comparisons between groups were performed using the t test for continuous data and the χ2 test for categorical data.
Group comparisons for time from admission to surgery were performed using unadjusted and risk-adjusted negative binomial regression models. Risk-adjusted analyses included the following covariates: sex, age, year of operation, race/ethnicity, zip code household income, primary insurance, Deyo score, hospital teaching status and location, hospital US region, and hospital volume. Comparisons of intraoperative and postoperative outcomes and discharge status were performed with univariate and multivariate logistic regression models using the same covariates. Because of the right-skewness of these data, length of hospital stay and total hospital charges were compared after natural log transformation. Unadjusted and risk-adjusted medians (95% CIs) are given after reexponentiation.22 Adjustment was performed using the same covariates as in the logistic regression models. Total hospital charges were standardized using 2008 US dollars as the reference.23 Because multiple years of the NIS were combined in these analyses, year-specific sampling weights were not applied to the data set. Because race/ethnicity was unknown for 25.3% of the patients herein, we performed sensitivity analyses of the multivariate models without race/ethnicity as a covariate.
All analyses were performed using commercially available software (STATA/SE, version 11.2; StataCorp LP). Significance was set at α = .05, and all P values were 2-sided.
In total, 31 832 patients who were admitted with diverticulitis and underwent urgent left-sided colon resection were included in this analysis. Between January 2002 and December 2008, a total of 24 766 patients (77.8%) underwent urgent colectomy because of left-sided diverticulitis after weekday admission and 7066 patients (22.2%) after weekend admission. The mean age of patients was 60.8 (15.3) years, 16 830 (52.9%) were female, and most patients (63.1%) were of white race/ethnicity (Table 2).
Among patients admitted on weekdays, 11 415 (46.1%) underwent primary anastomosis, and 13 351 (53.9%) underwent a Hartmann procedure (Table 3). Among patients admitted on weekends, 2486 (35.2%) underwent primary anastomosis, and 4580 (64.8%) underwent a Hartmann procedure (P < .001). This difference between patients admitted on weekdays vs weekends persisted after adjusting for multiple covariates (age, sex, year of operation, race/ethnicity, zip code household income, primary insurance, Deyo score, hospital teaching status and location, hospital US region, and hospital volume) in multivariate analyses, with an odds ratio of 1.57 (95% CI, 1.48-1.68; P < .001). Among patients who underwent primary anastomosis, ileostomy creation was performed in 148 of 2486 patients (6.0%) admitted on weekends and in 508 of 11 415 of patients (4.5%) admitted on weekdays (P = .001). However, this difference was not significant in risk-adjusted analyses (odds ratio [OR], 1.25; 95% CI, 0.99-1.56; P = .06). The time from admission to surgery did not differ between patients admitted on weekdays (median, 1 day; interquartile range, 0-4 days) and patients admitted on weekends (median, 1 day; interquartile range, 0-4 days) (unadjusted P = .66 and risk-adjusted P = .31).
Intraoperative complication rates did not significantly differ between patients admitted on weekdays (948 [3.8%]) vs weekends (266 [3.8%]) (P = .81) (Table 4). Overall postoperative complications were significantly more frequent among patients admitted on weekends (2070 [29.3%]) than among patients admitted on weekdays (6667 [26.9%]) (P < .001), even after adjustment for multiple covariates (OR, 1.10; 95% CI, 1.03-1.18; P = .005). Postoperative complications with higher relative odds in patients admitted on weekends than on weekdays included mechanical wound complications (risk-adjusted OR, 1.41; 95% CI, 1.19-1.67; P < .001) and cardiovascular complications (1.21; 1.03-1.43; P = .02). Rates of reoperations were significantly higher in patients admitted on weekends (146 [2.1%]) than on weekdays (374 [1.5%]), even after adjustment for multiple covariates (OR, 1.50; 95% CI, 1.20-1.88; P < .001). Nonroutine hospital discharge was significantly more frequent among patients admitted on weekends (3843 [54.4%]) than among patients admitted on weekdays (11 950 [48.3%]) (P < .001). This difference remained after risk adjustment (OR, 1.33; 95% CI, 1.24-1.43; P < .001).
Risk-adjusted median lengths of hospital stay were significantly longer among patients admitted on weekends (10.88 days; 95% CI, 10.71-11.04) than on weekdays (10.37; 10.29-10.46) (P < .001) (Table 5). Risk-adjusted median total hospital charges were significantly higher in patients admitted on weekends ($53 126; 95% CI, $52 148-$54 123) than on weekdays ($49 392; $48 906-$49 883) (P < .001).
The results of sensitivity analyses that excluded race/ethnicity as a covariate did not substantially differ from the primary results. No significant outcome became nonsignificant after removing race/ethnicity as a covariate, and estimates that included 95% CIs did not change appreciably.
This study investigated differences between patients undergoing urgent colectomy for left-sided diverticulitis who were admitted on weekends vs weekdays. Using the population-based NIS, our investigation provides strong evidence that patients with urgent admission for diverticulitis on weekends undergo significantly more Hartmann procedures and have fewer primary anastomoses compared with those admitted on weekdays. Moreover, patients admitted on a weekend had longer length of hospital stay, greater total hospital charges, more overall short-term complications, and higher rates of reoperations, nonroutine hospital discharge, and mechanical wound and cardiovascular complications.
Although a Hartmann procedure has been the gold standard for urgent operation in patients with diverticulitis, growing evidence suggests that primary anastomosis can be performed safely in a large number of patients.24-29 The rationale for favoring primary anastomosis in selected patients is related to the substantial number of long-term problems associated with undergoing a Hartmann procedure, including a low rate of reversals (31%-69%),30-34 as well as a high rate of stoma-related complications (eg, parastomal hernia or prolapse) and a complication rate of up to 49% after undergoing a Hartmann reversal.26,35-37 It is striking that in the present study, only 35.2% of patients admitted on weekends underwent primary anastomosis compared with 46.1% of patients admitted on weekdays. Earlier studies38-40 have shown that routine hospital care is provided only from Monday to Friday and that staffing is reduced and colorectal surgical specialists may be less available on weekends. The fact that there is less expertise available on weekends might be a substantial contributor to our findings. This is supported by Biondo et al41 and Zorcolo et al,42 who showed that experienced and specialized colorectal surgeons perform more primary anastomoses compared with trainees or general surgeons.
The results of this population-based study provide strong evidence that postoperative adverse outcomes of urgent operations for left-sided diverticulitis are more frequent in patients admitted on weekends than in patients admitted on weekdays. The overall complication rates among our patients (26.9% for weekday admission and 29.3% for weekend admission) are comparable to those reported in other studies of urgent colon surgery, with overall complication rates ranging from 23% to 58%.4,18,41 A weekend effect was also observed in our study for rates of reoperations and for hospital discharge status, while no difference was found for mortality. Reported mortality rates among patients undergoing urgent colorectal surgery range from 3% to 25%.4,41,43 Our mortality of approximately 4.5% in the present study falls in the lower range. It has been suggested that understaffing on weekends is the main cause of the weekend effect.44-46 In addition, physicians working on weekends are thought to be less experienced than teams working during the week.39,47 Biondo et al41 compared the outcomes of emergency colorectal resections performed by specialized colorectal surgeons vs general surgeons. After adjustment for multiple prognostic factors, general surgeons had a significantly higher postoperative complication rate than specialized surgeons (60.5% vs 52.2%). Zorcolo et al42 posited 2 reasons for this: (1) trainees have higher complication rates than consultants, and (2) slightly higher 30-day mortality occurs in patients operated on by upper gastrointestinal tract surgery teams compared with specialized colorectal surgeons.
Reflecting higher rates of postoperative complications and reoperations or lack of progression on standard pathways because of reduced personnel, patients herein who were admitted on weekends stayed in the hospital about 0.5 days longer than patients who were admitted on weekdays. In addition, the median total hospital charges were $3734 higher for patients who were admitted on weekends. The median length of hospital stay in our study is shorter than that in other series, estimated at 12 to 18 days in patients with nonelective colon resection for diverticulitis.4,29 Others have reported the lengths of hospital stay for nonelective colon resection as ranging between 16 and 21 days but did not limit their patient samples to those with acute diverticulitis.18,41 Because the data for length of hospital stay are often right skewed, the median can be significantly smaller than the mean. Given the substantial number of urgent left-sided colectomies performed each year, the median difference in length of hospital stay of 0.5 days, together with the median $3734 higher total hospital charges, observed herein for weekend admission is an important finding that should motivate improvements in the quality of weekend care.
Reasons for the observed weekend effect among patients admitted with acute diverticulitis and undergoing urgent left-sided colectomy could not be established. Based on available data from the NIS, it is impossible to draw strong conclusions about underlying causes of the weekend effect. However, several mechanisms are plausible. In teaching and nonteaching hospitals, in-house staffing is typically reduced on weekends, especially among specialists (eg, colorectal surgeons) and ancillary staff. Therefore, the large differences in outcomes of patients admitted on weekends vs weekdays observed herein may be owing to the presence or absence of trained colorectal staff surgeons. Furthermore, some patients, especially those with milder symptoms, may prefer weekend or weekday admission and may time their admission accordingly. The relative importance of these proposed mechanisms is unknown, and further studies that include qualitative research methods are needed to better understand the underlying reasons for the observed large weekend effect among patients undergoing urgent surgery for left-sided diverticulitis.
Our study has some limitations. First, because the NIS captures only inpatient data, we were unable to detect any complications among patients after hospital discharge. Second, adjustment in multivariate regression analyses was only possible with the covariates available in the data set. Although we were able to adjust for the Deyo score, we were unable to classify disease severity (eg, according to Hinchey classification) and cannot rule out that patients admitted on weekends were experiencing more severe diverticulitis than patients admitted on weekdays. Nevertheless, because our results are statistically significant and of great clinical relevance, it is unlikely that these findings could be fully explained by unknown confounding. Despite the limitations, we believe that this population-based study, with excellent external validity and high power, provides important insight into the treatment and outcomes of patients undergoing urgent left-sided colectomy for diverticulitis.
In conclusion, our study provides strong evidence that patients admitted for acute diverticulitis on weekends more frequently undergo a Hartmann procedure than patients admitted on weekdays. Moreover, worse outcomes, including postoperative complications, rates of reoperations, longer length of hospital stays, and higher total hospital charges, reflect an important weekend effect for patients undergoing urgent surgery for acute diverticulitis. Further investigations are warranted to examine possible underlying mechanisms involved in the weekend effect and to apply goal-directed strategies to improve patient outcomes.
Correspondence: Mathias Worni, MD, Research on Research Group, Department of Surgery, Duke University Medical Center, 2301 Erwin Rd, Durham, NC 27710 (firstname.lastname@example.org).
Accepted for Publication: February 20, 2012.
Author Contributions: Drs Worni, Schudel, Østbye, Shah, Pietrobon, Marosky Thacker, and Guller and Mr Khare had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Worni, Schudel, Østbye, Marosky Thacker, and Guller. Acquisition of data: Khare and Pietrobon. Analysis and interpretation of data: Worni, Schudel, Østbye, Shah, Khare, Pietrobon, Marosky Thacker, and Guller. Drafting of the manuscript: Worni, Schudel, Østbye, Marosky Thacker, and Guller. Critical revision of the manuscript for important intellectual content: Shah, Khare, and Pietrobon.
Financial Disclosure: None reported.
Funding/Support: This work was supported in part by grants PBBEP3-131567 (Dr Worni) and PBBEP3-134913 (Dr Schudel) from the Swiss National Science Foundation.
Additional Contributions: Danny O. Jacobs, MD, MPH, carefully reviewed the manuscript.
Create a personal account or sign in to: