The association between operating margin and surgical diversity at Critical Access Hospitals

Data

The study design was a cross-sectional secondary data analysis that obtained data from the American Hospital Directory (AHD), a private company that publishes information on over 7,000 hospitals nationwide within an online database. Primary data was originally reported by the Center for Medicare and Medicaid Services (CMS) as publicly available Medicare Cost Reports and Medicare Provider Analysis and Review (MEDPAR) files from financial years 2020 and 2021. Cost reports yield annual financial data from Medicare-certified providers. MEDPAR files contain information on specific inpatient service categories (e.g., surgical specialties offered per hospital) as well as Medicare case mix index (CMI) data. The study utilized information on hospital system affiliation that the AHD obtained through proprietary in-house methods that were not provided by Medicare Cost Reports.

All 1354 CAHs (as of July 2021) were initially included in data collection. Only the most recent fiscal year (FY) per hospital was utilized for analysis. The earliest FY utilized spanned from May 1, 2019 to April 30, 2020, and the latest FY utilized spanned from March 1, 2020 to February 28, 2021. The final data set therefore represented the period of May 1, 2019 to February 28, 2021. Hospitals were excluded if they had incomplete cost report data. The final data set included 1299 CAHs. The data utilized for this report did not meet the regulatory definition of human subjects research; therefore, institutional review board (IRB) approval was deferred by the REDACTED College of Osteopathic Medicine IRB Board.

Outcome measures

Many factors contribute to a hospital’s overall financial performance [20]. Pink et al. [21] presents a framework specific to CAHs that includes six dimensions of financial performance: profitability, liquidity, capital structure, revenue, cost, and utilization Profitability is measured by ratios that reflect profit relative to costs. Operating margin is the most frequently utilized measure of profitability in the hospital finance literature [20]. Operating margin is measured as operating revenue minus operating expense, divided by operating revenue and multiplied by 100. Because of surgery’s direct contribution to internal operations, and because surgery does not contribute to external operations, operating margin was chosen as the preferred measure of the effect of surgery on overall hospital finances. The use of operating margin rather than net margin also disregards potentially confounding factors such as hospital debt, taxes, and capital structures.

We did not perform regression analysis on operating cost and operating revenue because these two values are subsumed within the calculation of operating margin. The regression model accounted for the financial dimension of utilization by including occupancy rate as a covariate. The dimensions of liquidity and capital structure were not represented in the model. Descriptive statistics were reported for net income, total margin, and net patient revenue, but these financial indicators were not included in linear regression.

Regression model

The dependent variable was operating margin, and the independent variable was the number of surgical services offered per CAH. As defined by the American College of Surgeons, surgical specialties include: cardiothoracic surgery, colon and rectal surgery, general surgery, obstetrics and gynecology, gynecologic oncology, neurological surgery, ophthalmic surgery, oral and maxillofacial surgery, orthopedic surgery, otorhinolaryngology, pediatric surgery, plastic and maxillofacial surgery, urology, and vascular surgery [22].

Covariates included system affiliation (yes or no), ownership (public, private, or nonprofit), occupancy rate, CMI, and geographic region. Occupancy rate was calculated as inpatient days divided by available beds multiplied by 365, then multiplied again by 100 to generate a percentile. CMI is the one-year average of a hospital’s diagnosis-related group (DRG) weights and reflects the general cost and complexity of the Medicare and Medicaid cases that each hospital sees over the course of one year. Membership in a parent system, occupancy rate, and CMI have all shown positive correlations with CAH hospital operating margin [20, 23]. Operating as a “for-profit” hospital is associated with increased operating margins compared to “nonprofit” hospitals; additionally, both for-profit and nonprofit hospitals show higher operating margins than public, governmental hospitals [20]. Finally, each CAH was assigned a geographic region of the United States (Northeast, South, West, or Midwest) as per regions defined by the U.S. Census Bureau [24]. CAH closures occur more frequently in the South, and prior work suggests that the regional location in the South has a negative influence on CAH profitability [15, 19, 25].

Statistics

GraphPad Prism 9.2 software was utilized to perform least-squares multiple linear regression analysis utilizing the outcome variable and the different regressors. Goodness of fit was estimated with R². Standard errors, 95% confidence intervals, and p values were calculated for the intercept and for each regressor. A p value ≤ 0.05 was set as the significance threshold. For categorical variables with multiple degrees of freedom, a reference level was chosen based on the most prevalent categorical variable within each class. For example, when calculating parameter estimates based on geographic region, the Midwest (the most prevalent region) was set as the reference level and other regions were compared to the Midwest. Reference variables did not participate in the regression equation and did not receive parameter estimates. A Pearson correlation test was utilized to obtain correlation coefficients (r) for all continuous variables included in multiple regression. Multicollinearity testing was performed to obtain an R² for each independent variable.

Interpretation of results

We obtained a correlation of determination (R²) of 0.18 which suggests that 18% of the variance among hospital operating margins is explained by the regressors included in our model. Although there is no standardized means of interpreting R², Cohen categorizes an R² of <0.2 as representing a small effect size [26]. Our result is not dissimilar to R² values obtained in prior hospital research involving multiple linear regression utilizing operating margin as a dependent variable; such studies tend to involve R² values from 0.2 to 0.4 [27], [28], [29].

Parameter estimates, also known as beta weights, reflect the contributions of different regressors toward a dependent variable within a multiple linear regression equation while other variables are kept constant. In our model, the parameter estimates reflect the degree of increase in operating margin given a 1-unit increase in any of the regressors. In some cases, such as when hospitals are affiliated with a parent system, this is a binary difference. Considering the number of surgical services, each additional service produced an additional 1.5% increase in operating margin. Occupancy rate was formatted into the model as a percentage, therefore the obtained parameter estimate for occupancy rate reflected an additional 8.9% operating margin per 100% increase in occupancy rate—in more realistic terms, each 10% increase in occupancy rate was correlated with a 0.89% increase in operating margin. Similar manipulation was involved in the interpretation of CMI within the model: each 1-point increase in CMI corresponded to a 9.4% increase in operating margin and therefore each 0.1-point increase in CMI reflected a 0.9% increase in operating margin.

The moderate correlation between surgical services and CMI (r=0.59, p<0.0001) suggests that the presence of surgical services may provide a secondary, indirect effect on CAH profitability. CMI is calculated exclusively utilizing Medicare and Medicaid discharges. Theoretically, in prospective payment models, CMI should not affect operating margin because reimbursements should be proportionate to costs [20]. However, CAHs receive cost-based Medicare reimbursement at a 101% rate and therefore have a greater incentive to take on a high CMI compared to prospectively paid hospitals. Surgical services involve high DRG weights, which seem to help CAHs maximize cost-based Medicare reimbursements.

Occupancy rate was positively correlated with the number of surgical services provided (r=0.23, p<0.0001). Nonsurgical CAHs showed a mean occupancy rate of 23.09 compared to 29.79 at surgical CAHs. CAHs providing two or three surgical services showed occupancy rates of 31.37 and 37.47, respectively. This trend is not well explained by variations in total beds because CAHs are limited to a 25-bed maximum. The data suggest that surgical services help CAHs boost patient volume, and the interaction between surgical services and occupancy rate helps to explain the effects of both variables on operating margin. The effect of occupancy rate on operating margin, as delineated by Oner et al. [20], is attributable to economies of scale, as higher patient volumes lead to lower per-patient fixed operating costs.

Commentary

This study had two important findings reflecting the value of surgical services provided to CAHs. First, the study found that the presence of surgical services was associated with increases in operating margin proportionate to the number of different services provided. Second, the presence of surgical services was positively correlated with occupancy rate and CMI, both of which were independently associated with increases in operating margin. The reported associations of the selected covariates with operating margin also add to the CAH literature.

Low profitability is a predictor of hospital conversion and closure [30, 31]. Generally, rural hospital closures reduce local access to healthcare, and especially to emergency care [32]. Beyond medicine, hospital closures in rural communities can have broad economic impacts, because hospitals are large employers. Closures of sole community hospitals are correlated with reduced local per capita incomes and increased local unemployment rates [33]. The economic impact of CAH closures is observable at the county-wide level [34]. The results of this study suggest that the provision of surgical services can help increase hospital profitability and mitigate potential closures.

However, many CAHs face difficulties recruiting surgeons, and projections indicate that shortages of rural surgeons are likely to continue growing larger [7], [8], [9], [10, 35]. This is problematic for the financial health of hospitals facing surgical recruitment problems because the operating room is a key driver of rural hospital profitability [18]. The question remains: what can be done to address the widespread demand for rural surgeons?

Many osteopathic medical schools have a stated mission to train students who will go on to care for rural and underserved populations, based on an understanding that medical students from rural communities are more likely to practice in rural communities [36, 37]. Osteopathic physicians are overrepresented in the rural workforce compared to the overall workforce [38, 39]. Additionally, osteopathic surgical subspecialists are significantly more likely to practice in rural areas than their rural counterparts [40, 41]. As of 2012, osteopathic physicians comprised 4.9% of the total primary care physician (PCP) workforce yet accounted for 10.4% of the rural PCP workforce [39]. The state of rural primary care is symbiotic with the state of the hospitals that provide acute care to such areas, many of which are CAHs. The continued expansion of osteopathic undergraduate medical education is vital to supplying physicians, including surgeons, to rural populations whose healthcare needs remain unmet.

Student loan repayment poses another salient means of addressing rural surgeon shortages. Approximately 40% of rural hospitals offer some form of educational loan forgiveness in the hope of recruiting general surgeons [8]. Average tuition costs for first-year medical students at private medical schools have risen 22% from 2013 to 2021, outpacing the inflation of the US dollar (14%) across the same time frame [42, 43]. The National Health Service Corps (NHSC) provides a government-funded loan repayment program that extends financial assistance to medical professionals who choose to practice medicine in “communities of need” upon graduation [44]. For physicians, eligibility for NHSC loan repayment is contingent upon practicing in one of five primary care specialties: family medicine, general internal medicine, general pediatrics, obstetrics and gynecology, or geriatrics. The expansion of NHSC scholarship eligibility to include general surgery would represent a large stride toward assisting rural hospitals in recruiting surgeons considering the degree of need. This move would not be unprecedented, because the American College of Surgeons recognizes obstetrics and gynecology as a surgical specialty.

Study strengths and limitations

The regression model’s effect size was small (R²=0.18) and similar to the R² values obtained in prior hospital research utilizing multiple linear regression with operating margin as an outcome variable [27], [28], [29]. The generalizability of the results to CAHs is strong, because nearly all CAHs contributed to the analysis. Furthermore, we observed that occupancy rate, CMI, system affiliation, and geographic region exerted effects on CAH operating margin that were consistent with the current literature surrounding these variables [20, 22, 24]. However, the present study is limited in a few ways. First, the analysis is cross-sectional and can only suggest—not measure—causal relationships. Second, the sample window reflects only one year. The overlap between the study window and the ongoing COVID-19 pandemic presents a significant obstacle to generalizability, because the findings from this study were derived from a mix of prepandemic and pandemic financial data. A prospective follow-up of this study combining financial data across many years would provide a more accurate depiction of the nature of the relationship between surgical breadth and operating margin at CAHs. Inpatient surgical volume uniformly decreased during certain times of the study window due to institutional moratoriums on many elective surgical procedures; despite this nationwide drop in surgical volume, a positive correlation between surgical breadth and operating margin was still observed [45]. Finally, there were many potential covariates which were not included in the multiple regression model. These include market characteristics such as the Herfindahl-Hirschman Index and wage index, as well as hospital characteristics such as average length of stay and percent revenue from Medicare.