Data are given as means (SDs). Before elective diverticulitis surgery, patients had 1.0 (0.9) inpatient claims, 1.5 (1.5) outpatient claims, and 0.5 (1.2) antibiotic prescription claims on average. If considering both inpatient and outpatient claims, patients had 2.4 (1.6) episodes before surgery. If considering all 3 claim types, they had 3.8 (2.0) episodes before surgery.
eAppendix. Diagnosis and Procedure Codes Used in the Study in the Order They Are Reported in the Article
Customize your JAMA Network experience by selecting one or more topics from the list below.
Simianu VV, Fichera A, Bastawrous AL, et al. Number of Diverticulitis Episodes Before Resection and Factors Associated With Earlier Interventions. JAMA Surg. 2016;151(7):604–610. doi:10.1001/jamasurg.2015.5478
Despite professional recommendations to delay elective colon resection for patients with uncomplicated diverticulitis, early surgery (after <3 preceding episodes) appears to be common. Several factors have been suggested to contribute to early surgery, including increasing numbers of younger patients, a lower threshold to operate laparoscopically, and growing recognition of “smoldering” (or nonrecovering) diverticulitis episodes. However, the relevance of these factors in early surgery has not been well tested, and most prior studies have focused on hospitalizations, missing outpatient events and making it difficult to assess guideline adherence in earlier interventions.
To describe patterns of episodes of diverticulitis before surgery and factors associated with earlier interventions using inpatient, outpatient, and antibiotic prescription claims.
Design, Setting, and Participants
This investigation was a nationwide retrospective cohort study from January 1, 2009, to December 31, 2012. The dates of the analysis were July 2014 to May 2015. Participants were immunocompetent adult patients (age range, 18-64 years) with incident, uncomplicated diverticulitis.
Elective colectomy for diverticulitis.
Main Outcomes and Measures
Inpatient, outpatient, and antibiotic prescription claims for diverticulitis captured in the MarketScan (Truven Health Analytics) databases.
Of 87 461 immunocompetent patients having at least 1 claim for diverticulitis, 6.4% (n = 5604) underwent a resection. The final study cohort comprised 3054 nonimmunocompromised patients who underwent elective resection for uncomplicated diverticulitis, of whom 55.6% (n = 1699) were male. Before elective surgery, they had a mean (SD) of 1.0 (0.9) inpatient claims, 1.5 (1.5) outpatient claims, and 0.5 (1.2) antibiotic prescription claims related to diverticulitis. Resection occurred after fewer than 3 episodes in 94.9% (2897 of 3054) of patients if counting inpatient claims only, in 80.5% (2459 of 3054) if counting inpatient and outpatient claims only, and in 56.3% (1720 of 3054) if counting all types of claims. Based on all types of claims, patients having surgery after fewer than 3 episodes were of similar mean age compared with patients having delayed surgery (both 47.7 years, P = .91), were less likely to undergo laparoscopy (65.1% [1120 of 1720] vs 70.8% [944 of 1334], P = .001), and had more time between the last 2 episodes preceding surgery (157 vs 96 days, P < .001). Patients with health maintenance organization or capitated insurance plans had lower rates of early surgery (50.1% [247 of 493] vs 57.4% [1429 of 2490], P = .01) than those with other insurance plan types.
Conclusions and Relevance
After considering all types of diverticulitis claims, 56.3% (1720 of 3054) of elective resections for uncomplicated diverticulitis occurred after fewer than 3 episodes. Earlier surgery was not explained by younger age, laparoscopy, time between the last 2 episodes preceding surgery, or financial risk-bearing for patients. In delivering value-added surgical care, factors driving early, elective resection for diverticulitis need to be determined.
Acute diverticulitis is one of the most common indications for hospitalization related to the gastrointestinal tract, resulting in an estimated 300 000 admissions and 1.5 million days per year of inpatient care.1,2 In recent years, diverticulitis has also become one of the leading reasons for outpatient visits,3 and outpatient management is believed to be 3 times more common than inpatient care.4 In caring for patients with diverticulitis, surgeons have conventionally offered elective colectomy for patients who recover from a few initial episodes of diverticulitis to prevent recurrence or progression of disease requiring an emergency operation.5,6 However, in the last decade, professional surgical organizations have shifted their recommendations to delaying elective surgery until 3 or more episodes7-9 or, as of 2014, to not using episode number to determine if resection should be offered.1 In seeming disconnect with these guidelines, the incidence of elective resection for diverticulitis more than doubled.10-12 In addition, almost 1 in 3 contemporary operations occurs after only 1 or 2 uncomplicated episodes,13 raising questions about guideline adherence in these patients.
However, an important limitation of almost all prior studies10-14 on practice patterns for elective surgery for diverticulitis has been the focus on only hospitalized episodes of care, most often because of available national and large-scale data. This approach misses commonly occurring outpatient visits for diverticulitis (emergency department or clinic care)3,4 or home care when patients refill prescriptions for oral antibiotics. Because of this limitation, it is not clear if recommendations about delaying surgery1,7 are being adhered to or in which groups earlier resection is more common.
Several factors have been suggested as drivers of early elective surgery for diverticulitis. Younger patients (often described as ≤50 years) have historically undergone resection for diverticulitis more often than older patients.10,15 While more recent guidelines1,7 no longer recommended earlier resection in younger patients, diverticulitis in the young has risen significantly,10,11,15,16 and continued reliance on older guidelines may be an explanation. In addition, early elective resection may be attributable to increasingly offered laparoscopic colectomy for diverticular disease,12 recommended in contemporary guidelines1 because of lower morbidity, shorter hospitalization, and higher patient satisfaction.17 As seen with the introduction of laparoscopic approaches to cholecystectomy, the availability of less invasive technology can change the threshold for recommending or accepting the recommendation to undergo an operation.18,19 Still other explanations for early surgery may relate to increasing recognition of patients with “smoldering” (or nonrecovering) diverticulitis, who have persistent or recurrent symptoms within short periods of stopping antibiotic therapy.20,21 Finally, financial risk to the patient’s health care delivery system (as with capitated insurance plans) or to the patients themselves because of high deductibles and copayments may influence health care decision making22 but has not been evaluated in the context of early surgery for diverticulitis.
Given the apparent discordance between recommendations and trends in elective surgery, the objective of this study was to describe nationwide practices related to earlier or later surgery for uncomplicated diverticulitis using the MarketScan (Truven Health Analytics) databases, which include inpatient, outpatient, and antibiotic prescription claims specific to the treatment of diverticulitis events. We hypothesized that patients commonly undergo elective resection after fewer than 3 preceding diverticulitis episodes and that this early surgery is more common in younger patients, those having laparoscopic procedures, those manifesting more frequent episodes, and those bearing more financial risk for their care.
This study was exempted from human participants review by the University of Washington Institutional Review Board. Participant informed consent was not required.
The MarketScan Commercial Claims and Encounters Databases capture person-specific clinical use and expenditures across inpatient and outpatient settings, including prescription data for antibiotics. These data originate from a selection of large employers, health plans, and government and public organizations and include data from 100 different payers, with more than 500 million claim records. Claims are linked longitudinally during an individual’s enrollment in a given plan. Current Procedural Terminology; International Classification of Diseases, Ninth Revision; and Healthcare Common Procedure Coding System codes found in these claim records are verified by Thomson Reuters (New York, New York) and were used in definitions, as with prior MarketScan studies.23,24 All codes are listed in the eAppendix in the Supplement in the order that they were applied.
A retrospective cohort of adult patients (age range, 18-64 years) with a confirmed diagnosis of diverticulitis using International Classification of Diseases, Ninth Revision diagnostic codes (562.11 and 562.13) and a prescription for antibiotics used to treat diverticulitis was created using MarketScan claims from January 1, 2009, to December 31, 2012 (n = 403 996). To identify only patients with an incident diagnosis of diverticulitis, patients who were not continuously enrolled for at least 2 years or who had a diverticulitis-related diagnosis or colorectal resection in the 2 years before the index claim for diverticulitis were excluded (n = 300 603). This exclusion eliminated patients with Medicare supplemental insurance, whose primary Medicare claims were not recorded. Because these data are administrative, important details are unavailable to inform why certain patients did or did not undergo operations after recurrent episodes. Quiz Ref IDTo avoid misinterpretation when comparing operative patients with nonoperative patients, the scope of this study was to evaluate only those patients who had operations and to describe factors associated with earlier interventions. Accordingly, patients were sequentially excluded for the following reasons: absence of colectomy performed during the years of the study (n = 81 857); presence of a diagnostic code associated with immunocompromise at the time of the index diagnosis of diverticulitis (n = 15 932); performance of emergency surgery, defined as operation at the first episode or during subsequent emergency admission (n = 1325); and operations performed for complicated diverticulitis, defined by fistula, gastrointestinal bleeding, peritonitis, stricture, or abscess requiring percutaneous drainage (n = 1225). The final cohort included only 3054 nonimmunocompromised patients who underwent elective surgery after incident, uncomplicated diverticulitis over the 4 years after the initial washout period (January 1, 2009, to December 31, 2012). The dates of the analysis were July 2014 to May 2015.
The primary outcome was early surgery, defined by fewer than 3 episodes of diverticulitis preceding surgical resection. Episodes of diverticulitis were characterized as inpatient, outpatient, or antibiotic prescription only, according to claims derived from inpatient admissions and services, facility fees, outpatient services, and outpatient pharmaceutical claims. To ensure the specificity for diverticulitis, episodes were defined by both a diagnostic code for diverticulitis and a claim for diverticulitis-specific antibiotics within 7 days (eAppendix in the Supplement) as follows: (1) inpatient episode (≥1 diagnostic code and ≥1 typical antibiotic prescription from claims related to inpatient services), or (2) outpatient episode (≥1 diagnostic code and ≥1 typical antibiotic prescription from claims related to outpatient services), or (3) antibiotic prescription only (claims for ciprofloxacin-metronidazole combination without another inpatient, outpatient, or emergency department claim within 7 days in patients who had already had ≥1 inpatient or outpatient index episode of diverticulitis). Multiple claims within a 6-week period were considered part of the same episode. Emergency department visits that resulted in admission were categorized as inpatient episodes or otherwise were considered outpatient care.
Demographic characteristics included age, sex, and geographic region of the policyholder (categorized as Northeast, North Central, South, West, or unknown). To explore the effect of financial risk on early surgery, available insurance plans were categorized into the following 3 groups with increasing financial burden placed on the patient: (1) none or the lowest patient burden, including health maintenance organization (HMO) or capitated point-of service (POS) plans; (2) some patient burden, including comprehensive plans, exclusive provider organizations, noncapitated POS plans, and preferred provider organizations; and (3) highest patient burden, including consumer-driven or high-deductible health plans. Specific definitions for plans are available through a secure site (http://marketscan.truvenhealth.com).25 To describe comorbidities, the Charlson Comorbidity Index based on 12 months of claims before the index episode was calculated for each patient according to the method by Quan et al.26
Patient demographics and clinical characteristics were compared between those having early surgery (<3 preceding episodes) and delayed surgery (≥3 preceding episodes). We anticipated that the sensitivity for capturing additional episodes of diverticulitis would increase with additional claims categories and made an a priori consideration that a counting technique using all available episode types would serve for the primary analysis of factors associated with earlier surgery. The number of days between the last 2 episodes was calculated for patients with 2 or more episodes preceding surgery. Age, laparoscopy, insurance plan types, and days between the last 2 episodes preceding surgery were compared between those patients having early and delayed surgery in a univariable fashion using inpatient, outpatient, and antibiotic prescription claims to define episodes.
Categorical variables were summarized using frequency distributions and were compared using the Pearson χ2 test. Continuous variables were summarized using means (SDs) and were compared using the Wilcoxon rank sum (Mann-Whitney) test. Because the expected outcome (early surgery) prevalence was high (>10%) and because odds ratios can overstate the effect in studies with nonrare outcomes,27 factors associated with early surgery on multivariable regression were reported as relative risk (RR). The RR estimates were derived with Poisson regression models using robust, sandwich-style variance estimators under the generalized linear model framework using logarithms as the natural link function. Sandwich-style variance estimators provide consistent estimates of the covariance matrix even when a parametric model fails to hold. These model estimates are believed to be more appropriate for common outcomes and less susceptible to the influence of outlier data.27,28 Factors included in the multivariable model were selected because they were a priori hypotheses for earlier surgery (younger age, comorbidity, laparoscopy, and financial burden) or because they were significant in univariable analysis (sex and geographic region). Age 51 years was the cutoff for the cohort, selected because 50 years or younger had been the prior definition of younger patients with diverticulitis.1P < .05 was considered statistically significant. All analyses were performed using statistical software (Stata, version 13; StataCorp LP).
Of 87 461 immunocompetent patients having incident diverticulitis between January 1, 2009, and December 31, 2012, a total of 5604 (6.4%) underwent a resection. The characteristics of 3054 nonimmunocompromised patients (mean [SD] age, 47.7 [8.3] years; 55.6% [n = 1699] male) who underwent elective resection for uncomplicated diverticulitis are listed in Table 1. All patients in the cohort had at least 2 years of continuous insurance plan enrollment, with a mean enrollment of 2.8 years, and 60.4% (n = 1845) of patients having 3 or more years of continuous enrollment. While rates of early surgery did not differ across all health insurance plans (P = .11), the 2 insurance plans with capitated services had the lowest frequencies of early surgery (50.0% [228 of 456] for HMO and 51.4% [19 of 37] for capitated POS). Rates of early surgery also differed by geographic region (P < .001), with policyholders in the South having early surgery 60.5% (694 of 1148) of the time.
Before elective surgery, patients had a mean (SD) of 1.0 (0.9) inpatient claims, 1.5 (1.5) outpatient claims, and 0.5 (1.2) antibiotic prescription claims (Figure). Using more claim types increased the apparent number of episodes before surgery, so that the mean (SD) number of any episode (inpatient, outpatient, or antibiotic prescription claim only) was 3.8 (2.0). Early resection occurred in 56.3% (1720 of 3054) of patients based on all types of claims (Table 2) but occurred in 94.9% (2897 of 3054) if inpatient claims only were considered and occurred in 80.5% (2459 of 3054) if inpatient and outpatient claims only were considered. Patients having early surgery were similar in age to those having delayed surgery (P = .73) and had laparoscopy less frequently (65.1% [1120 of 1720] vs 70.8% [944 of 1334] in delayed surgery, P = .001). The mean (SD) time between the last 2 episodes of diverticulitis preceding surgery for all patients was 119 (116) days. The time between the last 2 episodes was longer in patients having early surgery compared with patients having delayed surgery (mean [SD], 157  vs 96  days; P < .001). Patients with HMO or capitated insurance plans had lower rates of early surgery (50.1% [247 of 493] vs 57.4% [1429 of 2490] than those with other insurance plan types, P = .01).
The independent factors associated with early surgery are listed in Table 3. Quiz Ref IDMale sex (RR, 1.07; 95% CI, 1.02-1.13; P = .004) and Charlson Comorbidity Index of 1 (RR, 1.06; 95% CI, 1.02-1.11; P = .003) were independent predictors of early surgery. Geographic region remained a predictor of early surgery regardless of how episodes were counted (P < .001 for all). Age 51 years or older (P = .10), laparoscopy (P = .11), and HMO or capitated insurance plans (P = .07) were no longer significant after adjustment.
The incidence of elective colectomy more than doubled,10-12 despite professional guidelines recommending delaying or, more recently, avoiding elective surgery for uncomplicated diverticulitis.1,7 However, by focusing only on hospitalized episodes of diverticulitis, this prior research has been limited in answering fundamental questions about proportions of operations that are guideline adherent and factors that might motivate earlier resection. Quiz Ref IDIn this study, after considering inpatient, outpatient, and antibiotic prescription claims, 56.3% (1720 of 3054) of elective surgery for diverticulitis occurred in patients with fewer than 3 episodes. Although professional society guidelines for elective resection have been in flux over the last 20 years,1,5-7 the data included in this study reflect a steady-state period for guidelines related to elective resection and render the study of early surgery during this time window appropriate in assessing guideline adherence. Within this context, the suspected drivers of early elective surgery (younger age, laparoscopy, more frequent episodes, and personal financial risk) were not found to be associated with earlier operations for diverticulitis.
Despite contemporary guidelines no longer recommending earlier operation in patients 50 years or younger with diverticulitis,1,7 debate persists about whether younger patients have more severe diverticulitis,14-16,29,30 whether the effect of the disease on their lives and quality of life is greater,17 and whether physicians are making decisions about surgical treatments based on these patients’ comparatively good health or longer life expectancy. This debate is particularly relevant because most elderly patients with newly diagnosed diverticulitis do not have recurrences and do not undergo operations for their disease.31 The rate of operations in that cohort of Medicare patients was more than double the rate we report (14.0% in that study vs our rate of 6.4% [5604 of 87 461 before inclusion and exclusion criteria were fully applied]), likely because the recommendations during that time were for early elective surgery. Quiz Ref IDWhile our study excluded patients 65 years or older because Medicare claims were not captured using this data set, the age distributions of included patients (age range, 18-64 years) did not differ between those who had early vs delayed surgery. Although our findings cannot be extrapolated to the older population, the mean age of patients hospitalized with diverticulitis in national inpatient sample analyses is 59 to 61 years,10,11 implying that this study’s population is similar to other national samples and is particularly relevant to understanding the surgical treatment patterns in the growing young population with diverticulitis. In this context, our findings challenge the notion that younger age is the reason for earlier elective surgery.
In addition, we postulated that the threshold to recommend and undergo surgical resection for diverticulitis might be lowered by the availability of laparoscopy. This hypothesis originated from evidence that (in the United States32,33 and in other countries34) greater use of laparoscopic resection has been associated with higher rates of elective surgery for diverticulitis. However, this association was not borne out in our study when outpatient and antibiotic prescription–only episodes were added. These findings corroborate the results of a recent large cohort by our group35 suggesting that episode number should not drive the decision to approach diverticulitis resection laparoscopically. Alternatively, the findings in this study may reflect a steady state in laparoscopic colectomy for diverticulitis because the data are from years during which the incidence of elective surgery for diverticulitis seems to have leveled off.12 The years in which elective surgery had the greatest rise (the early 2000s) predated the availability in 2008 of laparoscopic colectomy procedure codes to investigate this question.12,33
Given the substantial discrepancies in how episodes are counted, a better approach to determining the actual clinical burden of diverticulitis on patients may be evaluation of the healthy time between episodes. A normal interval between episodes has not been described, to our knowledge. Similarly, a time-based interval definition of smoldering diverticulitis has not been proposed.21 Rather, this recognized entity is defined by rebound symptoms when the treatment is withdrawn or more rapid recurrence of diverticulitis than occurs in typical patients. We examined this phenomenon by studying the time between the last 2 episodes of diverticulitis preceding surgery but did not find shorter intervals in patients having earlier resection.
Finally, this study explored the burden of financial risk to patients and their health care delivery systems associated with the use of earlier surgery. It has previously been suggested that patients who have insurance plan types with high out-of-pocket expenses choose care differently from those who do not bear risk22,36,37 (in particular, choosing less appropriate and inappropriate care than those with less direct financial risk). We theorized that patients who had greater financial risk would be less likely to undergo early resection, gambling that the less expensive outpatient visits would be less risky than choosing a certain, larger cost for inpatient surgery. Quiz Ref IDWe found no association of high-deductible and high-copayment insurance plan types with early surgery for diverticulitis. Furthermore, we found that patients covered by plans with capitated features, such as HMOs, had the lowest rates of early surgery. It might follow that these patients were exposed to decision making that prevented unnecessary costs to the delivery system, such as earlier surgery.38 Independent of the interpretation, these findings highlight the persistent evidence gap that exists in identifying which aspects of disease burden are most influential in determining patients’ and surgeons’ willingness to proceed with early elective surgery. Recently, a diverticulitis-specific quality-of-life instrument was developed, but it has yet to be formally applied to surgical patients.39 This type of assessment, rather than counts of episodes, may better relate patient decision making about elective surgery.
Several limitations of this study should be considered. First, the MarketScan databases are drawn from a large convenience sample, predisposing the data to biases and limiting the generalizability. For instance, because the data mostly originate from insurance plans related to large employers, patients receiving health insurance through medium and small firms, as well as the self-insured, are underrepresented. Second, a 2-year washout period may still be insufficient to avoid misclassifying index and recurrent episodes. While the ideal washout period required to establish a true inception cohort is not clear, 2.8% (84 of 3054) of recurrences in this cohort occurred after more than 2 years, so this washout definition was selected. Third, while we used diagnostic codes to eliminate patients with complicated diverticulitis, these patients are not reliably coded in administrative databases. The misclassification of recurrent episodes of diverticulitis as incident cases or the misclassification of complicated episodes as uncomplicated ones may falsely elevate the number of patients having early surgery but would not be expected to differ across the variables of interest (eg, laparoscopic surgery). Fourth, broadening the definition of a diverticulitis episode beyond inpatient claims may increase the sensitivity of the counting method, but it is unclear how specific that approach may be. To address this limitation, we included antibiotic prescriptions to help define outpatient episodes. Furthermore, we included antibiotic prescription–only episodes—perhaps the least specific of all—but only after an index inpatient or outpatient episode because this factor is reported as a common scenario in clinical care. Overall, this group remains one of the largest cohorts of patients undergoing elective resection for diverticulitis, allowing longitudinal assessment of episode count, and we believe that these limitations are balanced by the strengths of the study.
After considering inpatient, outpatient, and antibiotic prescription–specific episodes, 56.3% (1720 of 3054) of all elective surgery for uncomplicated diverticulitis in privately insured patients still occurred after fewer than 3 episodes. The use of laparoscopy does not appear to be a driver of earlier surgery, nor does younger age, more frequent episodes, or personal financial risk. Patient factors that were not tested in this study, including the burden to the patient’s quality of life or anxiety about future episodes, may be important in decision making related to early surgery. Similarly, surgeon training, local practice and referral patterns, strongly held beliefs, and other clinical factors not assessable by this study may also drive these patterns and explain some of the regional variation we found. In our nation’s quest to deliver higher-value health care, understanding what constitutes appropriate care for a growing population of patients with diverticulitis and encouraging adherence to appropriateness criteria are critical. These data suggest that there is a strong need for fundamental research in this setting.
Accepted for Publication: November 3, 2015.
Corresponding Author: Vlad V. Simianu, MD, MPH, Department of Surgery, University of Washington, PO Box 354808, 1107 NE 45th St, Ste 502, Seattle, WA 98105 (firstname.lastname@example.org).
Published Online: February 10, 2016. doi:10.1001/jamasurg.2015.5478.
Author Contributions: Dr Flum had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Simianu, Fichera, Thirlby, Flum.
Acquisition, analysis, or interpretation of data: Fichera, Bastawrous, Davidson, Florence, Flum.
Drafting of the manuscript: Simianu.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Simianu, Davidson, Flum.
Obtained funding: Simianu, Flum.
Administrative, technical, or material support: Simianu, Flum.
Study supervision: Fichera, Davidson.
Conflict of Interest Disclosures: None reported.
Funding/Support: The research reported in this publication was supported by grants T32DK070555 and R01DK103915 from the National Institute of Diabetes and Digestive and Kidney Diseases. The Comparative Effectiveness Research Translation Network, a program of the University of Washington, provided research and analytic assistance for this publication and was supported by grant HS20025 from the Agency for Healthcare Research and Quality.
Role of the Funder/Sponsor: Funding from the National Institute of Diabetes and Digestive and Kidney Diseases was used to support data collection and management. The Comparative Effectiveness Research Translation Network approved the final manuscript. All other components of the article represent the authors’ own work, including the design and conduct of the study, analysis and interpretation of the data, and preparation and review of the manuscript.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, the Agency for Healthcare Research and Quality, or the University of Washington.
Previous Presentation: This study was presented at the American College of Surgeons Clinical Congress 2015; October 7, 2015; Chicago, Illinois.