The equation is used for all i SSIs. The results must be summed for all i SSIs to derive the total change in profit for the health system or hospital due to SSIs. ICU indicates intensive care unit and LOS, length of stay.
eAppendix 1. Definitions and descriptions.
eAppendix 2. Statistical analysis (text with 6 tables).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Shepard J, Ward W, Milstone A, et al. Financial Impact of Surgical Site Infections on Hospitals: The Hospital Management Perspective. JAMA Surg. 2013;148(10):907–914. doi:10.1001/jamasurg.2013.2246
Surgical site infections (SSIs) may increase health care costs, but few studies have conducted an analysis from the perspective of hospital administrators.
To determine the change in hospital profit due to SSIs.
Retrospective study of data from January 1, 2007, to December 31, 2010.
The study was performed at 4 of The Johns Hopkins Health System acute care hospitals in Maryland: Johns Hopkins Bayview (560 beds); Howard County General Hospital (238 beds); The Johns Hopkins Hospital (946 beds); and Suburban Hospital (229 beds).
Eligible patients for the study included those patients admitted to the 4 hospitals between January 1, 2007, and December 31, 2010, with complete data and the correct International Classification of Diseases, Ninth Revision code, as determined by the infection preventionist. Infection preventionists performed complete medical record review using National Healthcare Safety Network definitions to identify SSIs. Patients were stratified using the All Patient Refined Diagnosis Related Groups to estimate the change in hospital profit due to SSIs.
Surgical site infections.
Main Outcomes and Measures
The outcomes of the study were the difference in daily total charges, length of stay (LOS), 30-day readmission rate, and profit for patients with an SSI when compared with patients without an SSI. The hypothesis, formulated prior to data collection, that patients with an SSI have higher daily total costs, a longer LOS, and higher 30-day readmission rates than patients without an SSI, was tested using a nonpaired Mann-Whitney U test, an analysis of covariance, and a Pearson χ2 test. Hospital charges were used as a proxy for hospital cost.
The daily total charges, mean LOS, and 30-day readmission rate for patients with an SSI compared with patients without an SSI were $7493 vs $7924 (P = .99); 10.56 days vs 5.64 days (P < .001); and 51.94 vs 8.19 readmissions per 100 procedures (P < .001). The change in profit due SSIs was $2 268 589.
Conclusions and Relevance
The data suggest that hospitals have a financial incentive to reduce SSIs, but hospitals should expect to see an increase in both cost and revenue when SSIs are reduced.
The US Health and Human Services Agency for Healthcare Research and Quality states that health care–associated infections were the most common serious complication of hospital care in the United States in 2008.1Quiz Ref IDThe Centers for Disease Control and Prevention estimates that there are 45 million inpatient surgical procedures performed annually in the United States.2 Approximately 20% of the estimated 2 million nosocomial infections in the United States each year are surgical site infections (SSIs)3 that have associated costs,3-26 morbidity, and mortality.10,13,18-20,24-27 While most agree about the negative clinical outcomes associated with SSIs, there is little consensus on the financial ramifications to the hospital.7,12,17,24,27Quiz Ref IDThe objective of this study was to estimate the change in hospital profit due to SSIs.
The study was authorized by The Johns Hopkins Hospital internal review board with a waiver of informed consent.
We defined hospital cost as the financial amount a hospital spends to provide services. Hospital revenue is the financial amount earned through business operations. Hospital charges are the amount issued to the payers and patients. Real charges and costs are charges and costs adjusted for inflation. The admission All Patient Refined Diagnosis Related Group (APR-DRG) and complexity score are calculated based on the patient’s present-on-admission diagnoses and other relevant characteristics.28 The APR-DRG is a system used to stratify patients by resource use and has been used to classify patients for reimbursement purposes.28 See eAppendix 1 in Supplement for supplemental definitions.
The primary outcomes of the study were the difference in daily hospital charges, intensive care unit (ICU) length of stay (LOS), floor LOS, 30-day readmission rate, and hospital profit for patients with an SSI when compared with patients without an SSI. Hospital charges were assumed to be a proxy for hospital cost; the terms will be used interchangeably. The daily charges were analyzed as a total and in 8 subcategories: room and board charges; operating room charges; pharmacy charges; radiology charges; laboratory charges; supply charges; therapy charges; and other charges. Using the Producer Price Index for hospitals from the US Bureau of Labor Statistics, all financial values referenced are measured in real dollars as of December 2010.29
A retrospective study was performed at 4 of The Johns Hopkins Health System acute care hospitals in Maryland: Johns Hopkins Bayview, a 560-bed academic tertiary care center; Howard County General Hospital, a 238-bed tertiary care center; The Johns Hopkins Hospital, a 946-bed academic tertiary care center; and Suburban Hospital, a 229-bed tertiary care center. Billing, laboratory, and medical record data were compiled for patients admitted to the hospitals or having a surgical procedure, according to International Classification of Diseases, Ninth Revision procedure codes, between January 1, 2007, and December 31, 2010. Suburban Hospital’s data were compiled for January 1, 2008, to December 31, 2010, only. The surgical procedures were identified using electronic health records.
Complete medical record review was performed by trained infection preventionists using National Healthcare Safety Network surveillance definitions to identify SSIs, as previously described.30-32 Follow-up contact with the surgeon was used to increase the detection rate for SSIs at all the hospitals except The Johns Hopkins Hospital. Patients undergoing the following inpatient surgical procedures were included in the study30:
Nonpediatric coronary artery bypass graft chest and/or leg incision.
Spinal fusion and refusion.
Eligible patients for the study included those patients with complete data and the correct International Classification of Diseases, Ninth Revision code, as determined by the infection preventionists. The data were stored in a Microsoft Access database.
The ICU LOS; floor LOS; daily total charges; daily room and board charges; daily operating room charges; daily pharmacy charges; daily radiology charges; daily laboratory charges; daily supply charges; daily therapy charges; and daily other charges were compiled for all patients. The daily hospital charges were calculated for each encounter by dividing the hospital charges by the patient’s LOS. These continuous variables were used to test the hypothesis that the case patients, patients with an SSI, will have higher daily hospital costs, a longer LOS, and higher 30-day readmission rates than the control patients, patients undergoing the same procedure who did not contract an SSI.
For the procedures with a sample size of 25 or more, normality was tested using an Anderson-Darling test. If normally distributed, a nonpaired single-tailed Welch t test was used to test the hypothesis; otherwise, a single-tailed Mann-Whitney U test was used. For procedures with a sample size less than 25, a single-tailed Mann-Whitney U test was used to test the hypothesis. An analysis of covariance was also used to test the hypothesis while controlling for surgical procedure, admission APR-DRG, and admission complexity score. A Pearson χ2 test was used to test the hypothesis that patients with an SSI will have a higher 30-day readmission rate. An α level of .05 was used. The statistical analysis was conducted using Minitab 16 Statistical Software (Minitab Inc) and SPSS (IBM).
For the financial analysis, a second control group was formed. This second control group comprised all patients admitted to the health system during the study period who did not contract an SSI. The control group patients were grouped by their admission APR-DRG and complexity score. Eligible patients for the financial analysis were those patients with all required data available in their electronic health record; admission APR-DRG and complexity score were available only for patients admitted after June 1, 2007.
After being grouped by their admission APR-DRG and complexity score, the mean floor LOS, ICU LOS, and daily total hospital charges were calculated for each admission APR-DRG and complexity score present in the control group. The difference between the floor LOS, ICU LOS, and daily total hospital charges was taken between the case patients (patients with an SSI) and the average of the control patients (patients without an SSI who had the same admission APR-DRG and complexity score as the case patient).
The differences in floor and ICU LOS found in the previous step for all patients with an SSI were multiplied by the mean charge for a floor day and ICU day from the patient’s respective hospital. Concurrently, the difference in daily total hospital charges found in the previous step was multiplied by the total LOS for all patients with an SSI. These results were used to provide an estimate of the change in revenue and cost, respectively, for the health system due to SSIs. The change in revenue minus the change in cost was calculated to estimate the change in health system profit due to SSIs.
Over the study period, there were 399 627 inpatient admissions, 25 849 surgical procedures of interest, and 618 SSIs identified, resulting in an SSI rate of 2.76 per 100 surgical procedures. Twenty-two thousand three hundred seventy-eight procedures and 618 SSIs were eligible for the hypothesis testing while 348 445 inpatient admissions, 17 392 procedures, and 547 SSIs were eligible for the covariance and financial analysis. Fifty-one thousand one hundred eighty-two admissions, 4896 procedures, and 71 SSIs were not eligible for the covariance and financial analysis since their record did not contain the admission APR-DRG and complexity score (Table 1).
Quiz Ref IDThe daily total charges for patients with an SSI were $7493 (95% CI, $7101 to $7884) vs $7924 (95% CI, $7788 to $8060) for patients without an SSI (P = .99). The patients with an SSI had a mean LOS of 10.56 days (95% CI, 9.50 to 11.62) vs 5.64 days (95% CI, 5.34 to 5.95) for patients without an SSI (P < .001). After adjusting for surgical procedure, admission APR-DRG, and admission complexity score, there remained an increased LOS and lower daily total charges in patients with an SSI compared with those without SSIs (eAppendix 2 in Supplement). The 30-day readmission rate for patients with an SSI vs patients without an SSI was 51.94 vs 8.19 readmissions per 100 procedures (odds ratio, 12.12; 95% CI, 10.27 to 14.29) (P < .001). Among patients with an SSI, there were 321 patients with at least 1 thirty-day readmission and 402 thirty-day readmissions in total. Among patients without an SSI, there were 1782 patients with at least 1 thirty-day readmission and 2139 thirty-day readmissions in total. The 547 patients with an SSI and eligible for the financial analysis had 2081 more hospital days (693 ICU days and 1389 non-ICU days) but lower daily total hospital charges of −$132 (95% CI, −$476 to $212) (P = .45) when compared with patients without an SSI and the case patients’ same admission APR-DRG and complexity score (Table 2).
If all 547 SSIs were eliminated, the data suggest that The Johns Hopkins Health System would experience a cost increase of $9 124 029 ($2 606 865.43 annually) and a billable capacity increase of 362 admissions (103 annually), equating to a revenue increase of $11 392 618 ($3 255 034 annually). Quiz Ref IDAdditionally, if it is assumed that payers refuse to reimburse for 30-day readmissions9 related to SSIs, then the elimination of SSIs would provide The Johns Hopkins Health System an increase in revenue of approximately $21 288 486 ($6 082 425 annually) over the study period by increasing their available billable capacity by 922 admissions (264 annually). The data suggest that the total change in profit over the period for the health system, if they eliminated all SSIs, would be $2 268 589, $12 164 457 if it is assumed 30-day readmissions would not be reimbursed (Table 3).
While the topic of financial impact of SSIs may seem trivial, it is not. These infections can be a source of readmissions and a driver of hospital performance. Hence, hospitals have a mandate to improve patient care and safety, which requires infrastructure that can support interventions focused on decreasing adverse events such as SSIs. To support these interests, hospitals and their administrators need data to help balance budgets and support infection prevention and other groups focused on improving performance. Unfortunately, hospitals are failing to see the major reduction in their cost by reducing SSIs.12 We attempted to rethink the approach to the financial calculations of SSIs and help demonstrate the financial ramifications associated with SSIs.
A number of previous publications have cited that the change in incidence of health care–associated infections is directly related to the change in hospital cost.3-11,13-26 This paradigm is the status quo and what much of the infection control and quality improvement community are basing their return on investment calculations on. The current method of calculating the change in hospital cost due to SSIs that we have observed most frequently is as follows. Assume patients with an SSI have costs of approximately $79 134, and the average patient without an SSI had total hospital costs of approximately $44 727. Taking the difference ($79 134 − $44 727), we derive the cost savings of preventing an SSI as $34 407. However, recent publications have challenged this paradigm.12,33 A recent article claims that quality improvement leads to improvements in profit for hospitals by creating additional capacity to treat patients, but quality improvement will not drastically alter the “typically rigid” hospital costs.33 This is further supported by a publication citing that hospitals have high fixed costs, up to 84%.34
The interpretation of cost often leads administrators and clinicians to confusion. The “lost opportunity to house new patients or increase capacity” will often be referenced appropriately by economists as a cost, an opportunity cost, but from a hospital administrator’s or manager’s perspective, this lost opportunity is not a cost, as previously defined, but rather a loss of potential revenue. This concept is illustrated in the following example.
For ease of calculation, assume a 25-bed unit has to plan a 100–calendar day budget with these factors:
The 25-bed unit will have a 100% occupancy rate.
The total cost for a bed day is fixed at $5000 per bed, yielding the total cost for the unit of $12 500 000 ($5000 per bed × 25 beds × 100 days).
All patients admitted have the same APR-DRG with reimbursement fixed at $30 000 per admission.
Patients with an SSI have a mean LOS of 10 days and patients without an SSI have a mean LOS of 5 days.
For a unit where 50% of the patients contract an SSI, the mean LOS will be 7.5 days, the maximum number of admissions will be 333, total revenue will be $9 900 000, and total profit will be ($2 510 000). However, if the SSI rate is 0%, the mean LOS will be 5 days, the maximum number of admissions will be 500, total revenue will be $15 000 000, and total profit will be $2 500 000 (Table 4).
Clarifying the semantics and concepts will provide administrators and clinicians the opportunity to understand their financial calculations. The method used under the current paradigm, as previously described, will provide the change in costs due to SSIs from the payer’s perspective, not the hospital’s perspective. For example, the data suggest a patient with an SSI will have an LOS of 10 days and total charges of approximately $79 134. Assume that SSI was prevented. The data suggest that the LOS for the patient would decrease to 5 days and the patient would accrue a charge of $44 727. The difference in charges, $34 307, would be the amount that the payer, the insurance company, or the patient would not have to expend. So, the payer would reduce their costs by $34 307.
In the same situation where the SSI was prevented, the hospital would reduce their revenue by $34 307 and would spend approximately $34 307 less on the patient since they only cared for the patient for 5 days. However, the hospital has a bed that is now empty for 5 days. This bed cannot be instantly staffed or unstaffed, so there is a cost to keeping the bed. This leaves 2 possible scenarios: (1) the hospital closes the bed or (2) the hospital uses the 5 days of empty bed space and admits an additional patient, known as backfilling the bed.
In situation 1, the hospital can reduce their cost if they reduce SSIs. When an SSI is prevented, the hospital will not have a patient to backfill the bed, so the hospital can choose to lay off or repurpose the staff; sell the capital equipment; or eliminate all expenditure associated with the bed. The closing of the bed will lead to a reduction in hospital cost and increase hospital profit in the short term. In the long term, closing the bed will reduce the maximum possible revenue the hospital can receive since fewer beds will be available to patients.
In situation 2, the hospital may have the ability to backfill the bed if an SSI is prevented. Under a case-based payment system, the data suggest the hospital will receive additional revenue if an SSI is prevented, since the hospital could admit 2 patients at $44 727 per case ($89 454 total charges) instead of a single patient with an SSI at $79 134 in charges. In this scenario, hospital revenue would increase when an SSI is prevented, but it is unclear how hospital cost is affected.
It was not surprising that patients with an SSI had higher total costs than patients without an SSI, but it was surprising that patients with an SSI had lower daily costs (Figure 1). The data suggest that the reduction of SSIs will decrease the hospital’s mean LOS, which could lead to an increase in daily total hospital cost and hence an increase in total hospital cost.
We propose that the change in hospital profit due to the prevention of 1 SSI can be described in the equation in Figure 2. This equation can be summed across i patients with an SSI in the health system/hospital to determine the change in profit.
We calculate the change in hospital cost and revenue by stratifying by the admission APR-DRG and complexity score, as previously described. We then subtract the cost to obtain the backfill patients. This cost will vary widely because some hospitals have easy access to additional market share, making this cost low, where other hospitals will need to embark on a costly marketing campaign to attract additional customers. We then subtract the cost of the intervention that prevented the SSI. The cost of the intervention is divided by the number of SSIs prevented because the cost should be distributed equally across all the SSIs prevented. We then add the missed reimbursement if payers refused to reimburse for 30-day readmission related to SSIs (Figure 3).
Taking the difference between the change in hospital cost and hospital revenue due to preventing a single SSI, we sum across all i SSIs to derive the change in hospital profit due to SSIs. The data in this study suggest that the net loss in profits due to SSIs for The Johns Hopkins Health System was between $4147 and $22 239 per SSI, not accounting for the cost to backfill patients or the cost of the intervention to prevent the SSIs.
Quiz Ref IDThis study had a couple of limitations. First, hospital charges were assumed to be an accurate proxy for hospital costs. This is not a suggested method, but the study was conducted in the state of Maryland, which was assumed to provide an accurate proxy given Maryland’s all-payer reimbursement system. Second, a time-dependent bias, as described by Barnett et al,35 can lead to overestimating the financial impact of SSIs. We adjusted for the time-dependent bias by selecting controls with the same admission APR-DRG and complexity score as the case patients for the financial analysis, but more accurate methods may be available.
Clinicians can spur hospital executives to invest in costly interventions or technology aimed at the reduction of SSIs by providing a cost-benefit analysis. When conducting such an analysis, the use of proper financial terminology is crucial. With the increasing need for additional infection prevention initiatives, which tend to require additional funding, clinicians must take care in presenting accurate financial figures to maintain the financial well-being of health care institutions and promote the safety of patients. Infection control and quality improvement professionals can use the equation given in Figure 2 to develop interventions that they project are cost appropriate. Payers should look to withhold reimbursement for SSI-related readmissions, as this makes SSI prevention more cost-effective.
Corresponding Author: John Shepard, MBA, Stanford Hospital and Clinics, Quality Improvement, 940 Oakes St E, Palo Alto, CA 94303 (firstname.lastname@example.org).
Accepted for Publication: January 22, 2013.
Published Online: August 21, 2013. doi:10.1001/jamasurg.2013.2246.
Author Contributions: Study concept and design: Shepard, Milstone, Frederick, Hadhazy.
Acquisition of data: Shepard, Carlson, Hadhazy.
Analysis and interpretation of data: Shepard, Ward, Milstone, Frederick, Perl.
Drafting of the manuscript: Shepard, Frederick, Perl.
Critical revision of the manuscript for important intellectual content: Ward, Milstone, Carlson, Frederick, Hadhazy, Perl.
Statistical analysis: Shepard, Frederick.
Administrative, technical, or material support: Shepard, Ward, Carlson, Frederick, Hadhazy, Perl.
Study supervision: Milstone, Perl.
Conflict of Interest Disclosures: None reported.
Additional Contributions: We acknowledge the assistance provided by Paul Allen, BS, and Kathleen Speck, MPH, from The Johns Hopkins Hospital; the infection preventionists throughout The Johns Hopkins Health System; John McGready, PhD, from the Johns Hopkins Bloomberg School of Public Health; and Jamie Shepard, BA, MS.
Create a personal account or sign in to: