Hospital and Payer Costs Associated With Surgical Complications

Importance  Increased costs of surgical complications have been borne mostly by third-party payers. However, numerous policy changes aimed at incentivizing high-quality care shift more of this burden to hospitals. The potential effect of these policies on hospitals and payers is poorly understood.

Objective  To evaluate costs associated with surgical quality and the relative financial burden on hospitals and payers.

Design, Setting, and Participants  We performed an observational study merging complication data from the Michigan Surgical Quality Collaborative and internal cost accounting data from the University of Michigan Health System from January 2, 2008, through April 16, 2015; the merged files from these data were created between June 5, 2015, and July 22, 2015. A total of 5120 episodes of surgical care for 24 surgical procedure groups (17 general surgical, 6 vascular, and 1 gynecologic) were examined. We report unadjusted and log-transformed risk-adjusted costs.

Main Outcomes and Measures  We compared hospital costs, third-party reimbursement (ie, payer costs), and hospital profit margin for cases with and without complications.

Results  The mean (SD) age of the 5120 patients was 56.0 (16.4) years, and 2883 (56.3) were female. The overall complication rate was 14.5% (744 of 5120) for all procedures, 14.7% (580 of 3956) for general surgery, 15.5% (128 of 828) for vascular surgery, and 10.7% (36 of 336) for gynecologic surgery. For all studied procedures, mean hospital costs were $19 626 (119%) higher for patients with complications ($36 060) compared with those without complications ($16 434). Mean third-party reimbursement was $18 497 (106%) higher for patients with complications ($35 870) compared with those without complications ($17 373). Consequently, with risk adjustment, overall profit margin decreased from 5.8% for patients without complications to 0.1% for patients with complications.

Conclusions and Relevance  Hospitals and third-party payers experience increased costs with surgical complications, with hospitals experiencing a reduction in profit margin. Both hospitals and payers appear to currently have financial incentives to promote surgical quality improvement.

In recent decades, widespread efforts have been made to improve the quality of surgical care.^1-6 Recent policy changes have increased the urgency for understanding the association between costs and complications from surgery.^4,7,8 Several emerging policy initiatives shift risk to hospitals by holding them accountable for the costs of complications. Under bundled payment plans, risks are accepted through combining payments for hospitals, physicians, and postacute care services, while accountable care organizations accept a degree of financial responsibility for care in exchange for potential financial incentives for achieving quality standards and savings benchmarks.^9-12 As we move into this new era, surgical leaders need to understand potential liabilities in the face of these shifts.

However, the body of literature relating cost to quality is underdeveloped. Some studies using Medicare data have shown an association between higher complications and costs.^13,14 But, this approach has been widely criticized for 2 reasons. First, this approach relies on the accuracy of diagnostic coding that is intended for billing purposes in the ascertainment of complications. Second, studies using Medicare data only provide the perspective of the payer and not the costs accrued by the hospital or health center that is taking care of a patient. Some studies have been done using more accurate clinical registry data and taking into account the financial perspective for hospitals, but those are now outdated or were limited to a narrow subset of operations.^15,16 To better understand the business case for quality and the implications of health care reform, a more detailed and updated study of the costs associated with complications from both the payer and hospital perspectives are needed.

We evaluated the costs associated with surgical quality and the relative burden that hospitals and payers experience as a result of this. We hypothesized that the effect on hospitals would include increases in costs and decreases in profit margin and contribution margin with complications. We also hypothesized that the effect on insurers would be an increase in reimbursement with complications.

Box Section Ref ID

We merged clinical information from the Michigan Surgical Quality Collaborative (MSQC) and internal cost accounting data available at the University of Michigan Health System. The MSQC is a regional collaborative quality initiative sponsored by Blue Cross Blue Shield of Michigan. Details about data collection and data integrity have been previously described.^5,17 In brief, trained personnel in each hospital collect patient demographic and comorbidity data, intraoperative and postoperative process details, and 30-day outcomes data for patients undergoing general, gynecologic, and vascular surgery throughout the state. The MSQC uses a case sampling method designed to minimize selection bias. This algorithm divides each calendar year into 46 8-day cycles, from which the first 25 consecutive surgical operations that meet MSQC inclusion criteria are selected. The cycle rotates every 8 days to ensure that every cycle begins with a different day of the week. Regular data audits ensure data accuracy. Patients with data collected in the MSQC at this hospital who underwent an operation between January 2, 2008, and April 16, 2015, and had corresponding cost accounting data available were included in our study. There were 744 cases (12.7%) with missing cost data, likely reflecting contractual relationships that are outside of the realm of traditional cost accounting, such as Michigan’s charity support care for patients. We chose to exclude these cases from our analysis, and we were left with 5120 cases with appropriate MSQC and cost accounting data for further analysis.

This study was approved by the University of Michigan institutional review board. Patient consent was waived because we analyzed previously collected anonymized data retrospectively; the use of the data for these purposes was approved by the institutional review board.

We obtained information on total hospital costs and reimbursement for each patient from our hospital cost accounting database. The EPSi system (Allscripts Inc) was used to identify total hospital costs and reimbursements (excluding professional physician fees).^15,18,19 The EPSi system tracks the use of all resources and assigns estimates of cost based on actual payroll and general ledger expenses, categorizing them as fixed or variable costs for each inpatient hospitalization.¹⁵ Available information includes total hospital episode costs, reimbursement by third-party payers, out-of-pocket expenses, and fixed, variable, direct, and indirect costs. We assessed the financial effect on hospitals by determining how the profit (reimbursement less costs) and profit margin (profit divided by reimbursement) change when complications occur. We also performed an analysis stratifying by procedure type, which avoids confounding due to the varying base prices and effect of complications in different procedures. We determined the financial effect on payers by estimating the increase in reimbursement associated with complications.

Contribution margin is hospital revenue minus variable costs. Theoretically, in the presence of operating room and hospital capacity, this measure indicates financial incentives for hospitals to provide care in the near-term.^15,20 Given concerns regarding potential unintended financial incentives associated with complications for hospitals that have been shown using contribution margin, we chose to also study this measure in our patient cohort.¹⁵

We used postoperative complications as a measure of poor-quality care. To determine whether complications occurred, we used definitions developed and recorded by the MSQC, which are prospectively assessed and recorded in a detailed clinical database.^5,17 These include mortality, stroke/cerebrovascular accident, myocardial infarction, cardiac arrest requiring cardiopulmonary resuscitation, unplanned intubation, severe sepsis, pulmonary embolism, organ space surgical site infection (SSI), deep venous thrombosis requiring therapy, deep incisional SSI, anastomotic leak, urinary tract infections, superficial incisional SSI, sepsis, pneumonia, and acute renal insufficiency and/or failure. For the purposes of our study, we combined all complications (major and minor) together to create a dichotomous variable. We used this to overcome the problem of small sample size if each type of complication were studied alone.

The MSQC collects data on procedure types that vary widely in complexity and risk. To ensure that our findings were not confounded by procedure mix, we stratified results by procedure type, and we created risk-adjustment models for hospital costs, variable costs, third-party payments, and actual payments (third-party payer plus patient out-of-pocket costs) using multiple linear regression. These models helped us account for most overall cost variance with R² values of 68.0% (hospital costs model), 67.3% (variable costs model), 61.3% (third-party reimbursement model), and 61.0% (actual payments model). We present results unadjusted as well as log transformed and risk adjusted and stratified by procedure and payer type (for those patients for which payer type data were available). Unadjusted results provide a better representation of high-cost outliers, while adjusted results account for patient and case factors that are potential confounders. Thus, we chose to present both results in this study.

Covariates in models included patient characteristics, comorbidities, primary procedure type, additional and concurrent procedures, and relative value unit for each primary Current Procedural Terminology procedure code (eTable in the Supplement). In colectomy, preoperative use of bowel preparation was also included as a covariate. Because of the right-skewed distribution of all dependent variables, we used log-transformed variables for regression analyses. Regression coefficients were then exponentiated to determine the proportional change associated with each independent variable of interest.

Cost data were retrieved from the University of Michigan Health System Data Warehouse, which runs on an ODBC server (Oracle Corporation) using SQL queries with Microsoft Access software (Microsoft Corporation). Merging data sets and statistical analyses were performed using Stata SE version 13 (StataCorp).

We analyzed 5120 patient episodes of surgical care. Patient characteristics for these episodes are shown in Table 1. Overall, 744 patients (14.5%) experienced complications. Race/ethnicity and body mass index were similar between groups. Patients with complications were more likely to be older (mean [SD] age, 60.5 [14.9] vs 55.2 [16.6] years), male (48.9% vs 42.8%), and have more comorbidities and risk factors including diabetes, hypertension, coronary artery disease, sleep apnea, tobacco use within 1 year, and a less healthy physical status based on the American Society of Anesthesiologists classification than patients who did not experience complications.²¹

When stratified by broad surgical groups, general surgical procedures had a complication rate of 14.7% vs 10.7% for gynecologic procedures and 15.5% for vascular surgery procedures. With regard to payer type, when known, patients with public (government-sponsored) insurance experienced a 19.8% complication rate vs 13.5% for those with private insurance.

Hospital costs, reimbursement from payers, and profit margin all varied with complications. The financial burden absorbed by the hospital is in part represented by changes in hospital costs when complications occur (Figure 1). For all studied procedures combined, unadjusted mean hospital costs were $41 942 higher (P < .001) for patients with complications ($59 205) compared with those without complications ($17 263). Following log transformation and risk adjustment, mean hospital costs were $19 626 higher (P < .001) for patients with complications ($36 060) compared with those without complications ($16 434).

The financial burden absorbed by third-party payers is best represented by changes in reimbursement when complications occur (Figure 1). For all studied procedures, unadjusted mean third-party reimbursement was $39 318 higher (P < .001) for patients with complications ($57 976) compared with those without complications ($18 658). Following log transformation and risk adjustment, mean third-party reimbursement was $18 497 higher (P < .001) for patients with complications ($35 870) compared with those without complications ($17 373).

Costs, reimbursement, and profit margin vary substantially by procedure, as shown in the 5 selected procedure groups (Table 2): cholecystectomy, colectomy, proctectomy, small-bowel procedures, and ventral hernia repair. We chose to present results from these procedures because they are some of the most common procedures performed at our institution. After stratifying by procedure type, some of these procedures on average yield a negative profit margin for the hospital with complications, while others are simply less profitable in the presence of complications.

Costs, reimbursement, and profit margin also vary substantially by complication type, as shown in 7 selected complications (Table 3): pneumonia; urinary tract infection; superficial, deep, and organ space SSI; sepsis; and pulmonary embolus. For superficial SSI, third-party costs increase but hospital profit margin remains at approximately 5% with or without this complication. Hospital profit margin decreases with deep incisional SSI (5% to 3%) but actually increases on average with organ space SSI (5% vs 13%). All other complications on average yield a decreased or negative profit margin for the hospital and increased costs for payers.

Perhaps the best representation of the financial burden absorbed by hospitals is profit margin (Figure 2). Overall, profit margin decreased from 5.8% for patients without complications to 0.1% for patients with complications. For patients with private insurance when insurance status information was available (n = 1330), profit margin decreased from 9.9% without complications to 7.6% with complications. For patients with public insurance (n = 858), a negative hospital profit margin was present with or without complications, but the negative profit margin was much greater in the presence of complications (−1.6% without complications vs −6.3% with complications).

Despite increases in reimbursement and decreases in profit margin, contribution margin (reimbursement less variable costs only) was found to decrease with complications overall, from 58.8% without complications to 56.6% with complications. This same decrease was seen when stratified by procedure type for 17 of 24 studied procedures.

In this study, we sought to enumerate the financial burden of surgical complications for both hospitals and the third-party payers who insure patients. We found that both hospitals and payers appear to have financial incentives to avoid surgical complications. Hospitals demonstrate a reduction in profit margin with complications. Payers demonstrate an increase in reimbursement with complications. Therefore, both of these stakeholders have financial incentives to improve surgical quality. With the demonstration of financial risks associated with complications, this study highlights the potential for attempts to improve quality by patient selection, or “cherry picking.” To avoid this, appropriate risk adjustment must be used in evaluating quality. This is critically important to understand hospital performance accurately.

There have been several previous studies evaluating cost-quality relationships in surgery, but these have had significant limitations.^22-24 Some of them demonstrated a cross-sectional association between complications and cost, but they were limited to data from Medicare billing claims.^13,14 For example, one recent study¹³ of national Medicare data for patients undergoing general and vascular procedures demonstrated that a reduction in hospital complications over time was, in fact, associated with significant reduction in costs to Medicare. However, this study was not able to assess the effect of these savings on hospitals in the form of hospital profit margin. Our study demonstrates the presence of a cost-quality association for both payers and hospitals, in a rigorously assessed clinical data set, and not simply from billing claims, which can be less reliable. While other studies have previously been done in similar clinical data sets, ours included updated data, including from 2015.^15,16,25

In contrast with our findings, one particular high-profile study¹⁵ demonstrated a potential perverse association between complications and costs. This study focused on contribution margin. Unlike overall profit margin, contribution margin does not take into account fixed costs (eg, the purchasing costs of computed tomographic scanners and surgical robots), which are typically divided and incorporated as part of charges for individual patient encounters over time. Looking at contribution margin for a 12-hospital system, these investigators found that it was higher for Medicare and private-payer patients in the presence of complications. For our study, we found a small decrease in contribution margin with complications, but we chose to focus our analysis on overall hospital profit margin, as it is more reflective of a hospital’s long-term financial stability. We found that, on average, cases without complications have a higher profit margin. Thus, although the previous study showed that hospitals may be affected adversely in the short-term by reducing surgical complications, we have shown that profits over the long-term are favorably affected by a reduction in complications.

There were several limitations to our study. First, our cohort represents patients and financial data from a single institution. These results may not be generalizable to smaller nontertiary facilities. Potential differences include case complexity and payer mix. But for hospitals that are similar in these aspects, we would expect trends to be similar. Second, this study retrospectively groups patients based on whether a complication occurred. While our cost models adjust for differences in measured patient factors and case mix between these 2 groups, there may be unmeasured confounders that are different between these 2 groups. Third, the quality of the cost data used is limited to the accuracy provided by the EPSi system.^15,18 While we had no control over the software’s collection and reporting of this data, we believe that the aggregate trends that we observed, comprised from thousands of cases, should be an accurate representation of costs, and this and similar systems have been used to track costs in other previously published work.^15,16,18 Moreover, while MSQC captures complications with a thorough and rigorous review by trained registrars, there is still the potential that these data miss some complications that do occur. Given this, if all complications were in fact appropriately captured, this would perhaps amplify the cost differences between patients with and without complications, and so this difference is perhaps understated. Finally, not all complications are preventable, and the rate of complications will never be zero. Nonetheless, our goal in this work was to better understand financial incentives for improved surgical quality, rather than to identify an exact measure of preventable harm.

Hospital and surgical leaders will be accepting a greater amount of risk under new payment policies, which are aimed at shifting risks to hospitals and physicians. Under bundled payment plans, which are one such policy, risks are accepted through combining payments for hospitals, physicians, and postacute care services. These plans essentially shift responsibility for poor quality and inefficiency to hospitals and physicians.^11,12 In accountable care organizations, another such policy, risks are shifted through the accountable care organization accepting a degree of financial responsibility for care in exchange for potential financial incentives for achieving quality standards and savings benchmarks.^9,10 Given these changes, there is a need to understand the magnitude of those risks for different procedures. The work presented in this study will provide health system leaders with an understanding of the potential financial risks associated with poor quality for a breadth of different procedures.

Corresponding Author: Mark A. Healy, MD, University of Michigan Department of Surgery, Center for Healthcare Outcomes & Policy, 2800 Plymouth Rd, Bldg 16, 016-100N28, Ann Arbor, MI 48109 (healym@umich.edu).

Accepted for Publication: February 22, 2016.

Published Online: May 11, 2016. doi:10.1001/jamasurg.2016.0773.

Author Contributions: Dr Healy had full access to all of 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: All authors.

Acquisition, analysis, or interpretation of data: Healy, Mullard, Dimick.

Drafting of the manuscript: Healy, Mullard, Dimick.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Healy, Mullard, Dimick.

Obtained funding: Healy, Dimick.

Administrative, technical, or material support: Healy, Dimick.

Study supervision: Campbell, Dimick.

Conflict of Interest Disclosures: Dr Dimick is a founder and has equity interest in ArborMetrix Inc, a software and analytics company focused on assessing hospital quality and costs. No other disclosures were reported.

Funding/Support: Dr Healy is supported by grant T32CA009672-25 from the National Institutes of Health. Dr Dimick receives research funding from the National Institutes of Health and the Agency for Healthcare Research and Quality.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: Dr Dimick is the Surgical Innovation Editor of JAMA Surgery but was not involved in the editorial review or the decision to accept the manuscript for publication.

Previous Presentation: This work was presented at the 11th Annual Academic Surgical Congress; February 3, 2016; Jacksonville, Florida.

Additional Contributions: We thank Priya Dedhia, MD, PhD, Lauren Wancata, MD, and Peter White, MD, of the University of Michigan Department of Surgery for internally reviewing and providing comments on the clarity of the text of this manuscript. No compensation was received for these contributions.