Synergistic Implications of Multiple Postoperative Outcomes

Abstract

As payment policies for surgical complications evolve, surgeons and hospitals need to understand the financial implications of postoperative events. Using data from the National Surgical Quality Improvement Program (NSQIP), the authors estimated mortality, length of stay (LOS), and total cost attributable to multiple postoperative events in general and vascular surgery patients. Data were collected using standard NSQIP practices at a single academic center between 2007 and 2009. LOS and costs were fit to linear regression models to determine the effect of 19 postoperative events in the setting of 1, 2, or 3+ events. Of 2250 patients sampled, 457 patients developed at least 1 postoperative event. LOS increased by 2.59, 5.18, and 10.99 days (P < .0001) for 1, 2, and 3+ postoperative events; excess costs were $6358, $12 802, and $42 790 (P < .0001), respectively. Multiple postoperative events have a synergistic effect on mortality, LOS, and the financial cost of patient care.

Keywords

postoperative events quality improvement length of stay cost

The payment landscape is changing dramatically for hospitals and for their surgical outcomes. The Centers for Medicare and Medicaid Services has determined that it will no longer cover the extra costs associated with treating preventable errors, injuries, or infections that occur in the hospital setting.¹ This change provides a strong incentive for hospitals to identify and prioritize postoperative complications and implement quality improvement programs to reduce them.

Rates of many postoperative complications have been well documented in the surgical literature. For example, depending on the surgical procedure and method of identifying the complications, surgical site infections (SSIs) have been estimated to occur in between 1.8% and 8% of cases, myocardial infarction in between 0.5% and 1%, pneumonia in between 0.7% and 2.4%, and cerebrovascular accidents (CVAs) in <1%.² Much less is known about the contribution of these complications to mortality, length of stay (LOS), and costs³ in specific populations, particularly when they occur in combination. Studies have reported that postoperative complications result in significant resource consumption but only reported on complications with single procedures, selected populations, or small groups of patients.^4-6 In previous work, the authors used data from the National Surgical Quality Improvement Program (NSQIP) to estimate the excess costs and LOS associated with SSIs in our institutional general and vascular surgery population and found them to be approximately $10 800 and 4.3 days, respectively. However, SSI is only 1 of 19 surgical complications tracked by NSQIP, and many patients had more than 1 complication.

The objective of this study was to examine mortality, LOS, and the excess cost associated with all 19 measured complications in general and vascular surgery patients using NSQIP data. An important challenge to conducting this analysis was that many of the complications co-occur, making it difficult to isolate the specific impact of a single complication. This was dealt with by estimating the effects of single complications separately from aggregate effects when complications occurred in clusters of 2 and 3 or more.

Methods

Data

Data came from the Penn State Milton S Hershey Medical Center NSQIP and consisted of patient data randomly collected on 2250 adult general and vascular surgical inpatients at a single academic center between 2007 and 2009 using standard NSQIP methodology.^7-11 Selected cases were the first 40 eligible vascular and general surgery operations during an 8-day cycle, where each cycle started on a different day of the week, thus capturing approximately 20% of the hospital’s surgical volume. A trained clinical nurse reviewer collected data on 60 preoperative patient characteristics, 18 intraoperative factors, and 22 postoperative occurrences up to 30 days following the procedure. Chart review, information from morbidity and mortality conferences, and communication with patients by telephone or letter completed the 30th postoperative day data collection.^7-11

Postoperative occurrences under the NSQIP program are divided into 5 categories: wound occurrences (superficial, deep, organ-space SSIs, and fascial dehiscence); respiratory occurrences (pneumonia, unplanned intubation, pulmonary embolism, and intubated >48 hours duration); urinary occurrences (progressive renal insufficiency, acute renal failure requiring dialysis, and urinary tract infection); central nervous system occurrences (cerebral vascular accident, coma >24 hours, and peripheral nerve injury); and cardiac occurrences (cardiac arrest requiring cardiopulmonary resuscitation [CPR] and myocardial infarction). Other occurrences included postoperative blood transfusions, deep venous thrombosis, sepsis/septic shock, and graft/flap failure. To assign a postoperative occurrence, the clinical nurse reviewer identifies medical documentation that specifically meets the American College of Surgeons NSQIP definitions criteria for the selected postoperative occurrence. The definitions criteria are adapted from those used by national regulatory agencies.

The total cost of admission data were obtained from the hospital’s accounting database (Horizon Performance Manager; McKesson, San Francisco, CA). These represented fully loaded operating costs for the surgical admission estimated using a ratio of costs-to-charges (CCR) methodology. The CCR approach is a standard cost accounting technique, whereby costs are estimated as a percentage of hospital charges. Although the CCR approach has limitations for estimating hospital costs, it is by far the most prevalent technique, and previous research has shown department-level CCRs to be a reasonable approach to estimating costs within a hospital and costs across hospitals for patients within a common diagnosis-related group.¹²

Statistical Analysis

The statistical analysis was designed to estimate the excess cost, LOS, and mortality associated with postoperative events controlling for other patient- and surgery-specific confounders. Age, sex, ethnicity, and preoperative patient comorbidities were controlled for because these variables are not only associated with increased risk of developing postoperative complications but contribute to an increase in total hospital costs independent of infection risk.¹³

Patient characteristics were compared using the Student t test for continuous variables and the χ² test for binary and categorical variables. Logistic regression was used to determine which postoperative events had a significant effect on mortality. Total cost and LOS were fit to generalized linear regression models assuming a gamma family of distributions and a log link function. These models were chosen because the cost and LOS data were highly skewed and did not meet the normality assumption of the classical linear regression model. The generalized linear regression models allowed the estimation of the effect of each postoperative event as well as the effect of having more than 1 postoperative event. All statistical analyses were performed using STATA, version 11 (StataCorp LLP, College Station, TX). Statistical significance was defined as a P value <.05.

Results

Demographics

During the period 2007 to 2009, there were 2250 general and vascular surgery inpatients captured in the NSQIP database at our institution. The majority of patients underwent surgery of the colon and rectum (26.2%), the foregut (15.9%), bariatric surgery (11.5%), and vascular procedures (9.0%). Patient characteristics stratified by the occurrence of at least 1 postoperative complication are reported in Table 1. Patients were similar in patient and preoperative characteristics, such as ethnicity, body mass index, smoking status, and alcohol use. The groups also were alike with regard to intraoperative characteristics, including clean and contaminated wound classes and operative time of 4 to 8 hours. There were, however, some significant differences between the 2 groups, suggesting risk factors for developing postoperative complications. For example, the postoperative event group was more likely to be male, to have insulin-dependent diabetes, and to have an operative time longer than 8 hours.

Table 1.

Characteristics of Inpatients Without and With More Than 1 Postoperative Complication

	Without Complications, n = 1793	With Complications, n = 457
Variable	Mean/Percentage	Mean/Percentage	P Value^a
Age (years)	55.4	59.0	<.0001
Sex
Female	56.1%	50.1%	.0230
Male	43.9%	49.9%	.0230
Race/Ethnicity
White	95.6%	93.9%	.1100
Nonwhite	4.4%	6.1%	.1100
Preoperative
Preoperative LOS (days)	0.7	1.8	.0530
Transfer status	3.9%	12.7%	<.0001
BMI (kg/m²)	40.9	41.6	.7305
Nondiabetic	82.2%	78.8%	.0927
Diabetic: insulin dependent	6.3%	9.8%	.0008
Diabetic: non–insulin dependent	11.5%	11.4%	.9210
Smoker	20.2%	24.1%	.0730
History of chronic obstructive pulmonary disease	4.5%	8.8%	<.0001
Current pneumonia	0.7%	3.5%	<.0001
Hypertension	53.1%	53.2%	.9760
Open wound	7.4%	17.9%	<.0001
Steroid use	4.8%	7.9%	.0090
Previous operation within 30 days	2.9%	7.0%	<.0001
Disseminated cancer	4.4%	9.8%	<.001
Alcohol use	1.8%	2.4%	.3860
Dyspnea, none	82.9%	72.9%	<.0001
Dyspnea, moderate	15.6%	15.8%	.9190
Dyspnea, at rest	1.5%	11.4%	<.0001
Congestive heart failure	0.4%	4.2%	<.0001
Myocardial infarction	0.6%	2.2%	.0010
Percutaneous coronary intervention	6.9%	8.5%	.2170
Previous cardiac surgery	7.9%	9.0%	.4630
History of angina	1.2%	3.1%	.0040
Peripheral vascular disease	6.1%	9.2%	.0200
Acute renal failure	0.3%	2.2%	<.0001
Currently requiring hemodialysis	1.5%	3.7%	.0020
Laboratory data
Albumin (g/dL)	3.8 (n = 811)	3.3 (n = 299)	<.0001
Intraoperative
ASA class 1	1.8%	0.4%	.0350
ASA class 2	35.9%	17.3%	<.0001
ASA class 3	57.8%	61.1%	.2050
ASA class 4 or 5	4.6%	21.2%	<.0001
Wound class, clean	31.2%	26.9%	.0770
Wound class, clean–contaminated	47.4%	41.8%	.0320
Wound class, contaminated	11.8%	14.9%	.0720
Wound class, dirty	9.6%	16.4%	<.0001
Time in OR (hours)	3.9	5.0	<.0001
OR time <4 hours	67.8%	56.5%	<.0001
OR time 4-7:59 hours	25.5%	28.0%	.2730
OR time 8-11:59 hours	4.5%	10.1%	<.0001
OR time 12+ hours	2.2%	5.5%	<.0001
Emergent case	9.9%	21.0%	<.0001

Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index; LOS, length of stay; OR, operating room.

Comparison performed using t test for continuous variables and χ² test for categorical variables.

Mortality

In all, 226 patients developed 1 postoperative event; 5 of those events resulted in death (Table 2). Mortality rates for venous thromboembolism, intubated >48 hours, and transfusion requirement were not statistically significant. However, mortality rates for cardiac arrest requiring CPR and CVA were statistically significant. Also reaching statistical significance were the number of patients with 2 and 3 or more postoperative events who died.

Table 2.

Mortality Rate for Postoperative Events

	Mortality Rate
Postoperative Complications^a	Alive	Dead	P Value^b
Venous thromboembolism	95%	5%	.3950
Intubated >48 hours	89%	11%	.1930
Transfusion requirement	67%	33%	.0690
Cardiac arrest requiring CPR	0%	100%	.0000
Cerebrovascular accident	0%	100%	.0240
Postoperative events = 2	92%	8%	.0020
Postoperative events = 3+	78%	22%	.0000

Abbreviation: CPR, cardiopulmonary resuscitation.

The remaining postoperative complications studied resulted in 0% mortality.

Comparison performed using Fisher’s exact test.

Length of Stay

The predicted mean LOS without a postoperative event was 5.5 days. Patients with 1 postoperative event had a LOS that was 2.6 days longer (P < .0001) than patients with no events. The longest statistically significant LOS was attributed to ventilator dependence (Table 3), followed by pneumonia and sepsis.

Table 3.

Multivariate Linear Regression Analysis of Length of Stay Associated With Postoperative Events

			95% Confidence
Events, n (%)	Marginal Effect, days	Postoperative Event	Lower	Upper	P Value
Cardiac arrest requiring CPR	5 (0.22)	−3.90	−4.98	−2.83	<.0001
Cerebrovascular accident	1 (0.04)	−1.31	−7.47	4.84	.6760
Transfusion requirement	3 (0.13)	−0.96	−4.77	2.86	.6230
Intubated >48 hours	9 (0.40)	0.19	−2.58	2.96	.8920
Other cardiac occurrence	3 (0.13)	0.22	−4.64	5.08	.9290
Other wound occurrence	12 (0.53)	0.87	−1.80	3.53	.5240
Graft/prosthesis failure	8 (0.36)	0.98	−2.34	4.30	.5620
Other occurrence	12 (0.53)	1.54	−1.41	4.49	.3070
Wound dehiscence	6 (0.27)	2.07	−2.41	6.54	.3660
Surgical site infection	86 (3.80)	2.14	0.93	3.34	<.0001
Myocardial infarction	2 (0.09)	2.15	−5.75	10.06	.5930
Venous thromboembolism	21 (0.93)	2.45	−0.08	4.99	.0570
Urinary tract infection	16 (0.71)	3.02	−0.07	6.10	.0550
Renal insufficiency	2 (0.09)	3.67	−5.70	13.04	.4430
Sepsis	17 (0.76)	4.94	1.24	8.64	.0090
Pneumonia	8 (0.36)	6.95	0.57	13.34	.0330
Ventilator dependence	6 (0.27)	8.81	0.17	17.46	.0460
Septic shock	4 (0.18)	10.62	−1.14	22.39	.0770
Acute renal failure with hemodialysis	4 (0.18)	11.78	−0.73	24.29	.0650
Postoperative events = 2	104 (4.62)	5.18	3.65	6.71	.0000
Postoperative events = 3+	127 (5.64)	11.01	8.81	13.21	<.0001

Abbreviations: CPR, cardiopulmonary resuscitation.

As suggested in Table 3, many patients had more than 1 postoperative event, which also was significantly associated with excess hospital LOS. Figure 1 (left panel) illustrates the cumulative effect on LOS of incurring 1, 2, or 3 or more postoperative events.

Figure 1.

Cumulative length of stay and cumulative cost of postoperative events

Hospital Costs

The predicted mean total cost of a hospital stay across all procedures without a postoperative complication was $14 839. A single postoperative event added an average of $6358 in additional costs. Results of the regression analysis presented in Table 4 suggest that 5 of the 19 postoperative events studied had a significant effect on total hospital costs. Ventilator dependence was associated with the largest statistically significant attributable cost, followed by pneumonia and sepsis.

Table 4.

Multivariate Linear Regression Analysis of Costs Associated With Postoperative Events

		95% Confidence
Postoperative Complication	Marginal Effect ($)	Lower	Upper	P Value
Cerebrovascular accident	−4097	−19 955	11 762	.6130
Other cardiac occurrence	−1920	−12 826	8987	.7300
Cardiac arrest requiring CPR	766	−9622	11 155	.8850
Other wound occurrence	1822	−5162	8807	.6090
Myocardial infarction	2074	−15 413	19 561	.8160
Other occurrence	2669	−4690	10 028	.4770
Urinary tract infection	2953	−3523	9429	.3710
Intubated >48 hours	3312	−5496	12 119	.4610
Surgical site infection	3349	465	6233	.0230
Transfusion requirement	3598	−11 901	19 097	.6490
Renal insufficiency	6229	−15 321	27 780	.5710
Wound dehiscence	7210	−5860	20 280	.2800
Venous thromboembolism	7357	259	14 456	.0420
Graft/prosthesis failure	8933	−3239	21 106	.1500
Sepsis	12 780	2880	22 680	.0110
Pneumonia	14 617	−511	29 746	.0580
Septic shock	22 129	−4785	49 043	.1070
Acute renal failure with hemodialysis	25 683	−3650	55 017	.0860
Ventilator dependence	49 362	10 421	88 303	.0130
Postoperative events = 2	12 823	8834	16 811	<.0001
Postoperative events = 3+	43 196	35 320	51 073	<.0001

Abbreviations: CPR, cardiopulmonary resuscitation.

The cumulative effect of LOS with 1, 2, and 3 or more postoperative events is shown in Figure 1 (right panel).

Discussion

Despite continued improvement in patient selection, operative techniques, and perioperative management, general and vascular surgery procedures still result in significant perioperative morbidity and mortality.⁵ For this patient population, postoperative complications are associated with an increased risk of death, prolonged hospital LOS, and increased hospital costs. It is important to understand the costs and LOS associated with individual postoperative complications in order to develop quality improvement efforts tailored to prevention strategies, thus improving patient outcomes and more efficiently allocating health care resources.⁵

In this study of 2250 surgical inpatients at a large academic medical center, 457 patients developed at least 1 postoperative complication. Two of the 19 complications resulted in a statistically significant mortality rate and included cardiac arrest requiring CPR and CVAs. The mortality rate also was statistically significant if more than 1 postoperative event occurred. Significant complications associated with both cost and LOS in order from greatest to least resource use were ventilator dependence, pneumonia, sepsis, venous thromboembolism, and SSIs. The development of ventilator dependence longer than 48 hours following a general or vascular surgical procedure incurred additional costs in excess of $49 362 and increased LOS by 8.8 days for the hospital admission in which the surgical procedure was performed. Additionally, developing pneumonia incurred additional costs of $14 617. Patients with sepsis, venous thromboembolism, and SSIs incurred additional costs of $12 780, $7357, and $3349, respectively. In a previous study, we reported excess costs attributable to SSIs as approximately $10 800. Although we had controlled for preoperative characteristics, unlike this study, we did not account for the possibility that the patient may have had multiple postoperative complications in addition to an SSI, thus increasing the cost. More important, this study highlights the financial implications of treating postoperative complications in surgical patients and the effects of multiple complications or events on outcomes and costs. It is not surprising that the same postoperative events associated with significant increases in additional costs also are associated with significant increases in LOS because the most important determinant of health care spending is the intensity of services provided over the course of a hospital stay.¹⁴ It also should be noted that postoperative ileus contributes significantly to postoperative LOS. During the period 2007 to 2009, the NSQIP database did not include postoperative ileus as a postoperative complication; therefore, this variable was not included in the analysis. However, as of January 2011, the NSQIP program instituted a procedure targeted approach, which allows larger facilities or specialty hospitals to focus on high-volume, high-risk procedures, thereby enabling them to target quality improvement efforts in areas that will yield the greatest return on investment. This may include procedure-specific variables and complications such as postoperative ileus.¹⁵

Even though the economic impact of postoperative complications is acknowledged to be significant, few studies specific to general and vascular surgical procedures exist to quantify the costs to treat these complications. Dimick et al⁷ used their NSQIP database to examine costs associated with postoperative events at a single academic medical center. They found that when complications were grouped by infectious, cardiovascular, respiratory, or thromboembolic cause the unadjusted hospital costs increased by $8209, $13 256, $54 430, and $28 355, respectively. However, their study did not isolate individual complications within those groupings for cost comparisons. Another NSQIP study by Davenport et al¹³ examined costs attributable to preoperative risk factors, surgical complexity, and individual postoperative complications. Their study concluded that sepsis, acute renal failure, ventilator dependence, septic shock, and pneumonia were associated with the highest increase in costs. However, their costs were reported as an increase in mean costs associated with complications. When examining metrics of financial performance, marginal costs as opposed to average costs are the key to cost-effectiveness assessments because those are the costs to consider when making budget allocation decisions.¹⁶

A study by Kalish et al⁶ abstracted data from 372 680 major surgery patients in 404 California acute care hospitals and identified 26 complications. They found, on average, that patients with postoperative complications had higher total charges of $30 896 versus $9239 for no complications and longer stays of 13.5 versus 5.4 days. Another study by Khan et al² determined total hospital costs and LOS in all noncardiac surgical patients admitted to a single tertiary care center over a 2-year period. A total of 6.9% of their 7457 patients developed at least 1 postoperative complication. Overall complications increased hospital costs by 78% and LOS by 114%. However, the latter 2 studies relied on administrative databases to assess surgical quality and determine postoperative complications, which may underestimate the true incidence of complications. Administrative databases are not designed to model morbidity directly and, unlike NSQIP, which measures postoperative events directly, rely on billing codes to identify postoperative complications.¹⁷ In a study by Davenport et al,¹⁷ a direct comparison of the NSQIP data set and the University HealthSystem Consortium Clinical database found that, where the data sets described similar severe complications, NSQIP identified them more accurately and reliably.

It is important to identify those postoperative complications that are costly and contribute to prolonged LOS in order to develop the best quality improvement efforts to maximize prevention strategies. At our institution, the first risk-adjusted NSQIP semiannual report for the institution was released in 2008 and indicated postoperative SSIs in general and vascular surgery patients as an area for improvement with an observed to expected (O/E) ratio of 1.31. The NSQIP data indicated that 19.3% of SSIs in general and vascular surgery patients occurred in patients with diabetes, whereas it was 8% in patients without diabetes. A multidisciplinary task force was formed to develop a glucose control protocol, which was implemented for all surgical patients later that year. The following risk-adjusted report yielded an O/E ratio of 0.78 for SSIs.

This study has several limitations. First, all the patients studied were from a single academic medical center, and results may not be representative of general and vascular surgery patients who develop postoperative complications in other institutions. It also is not known whether the costs or LOS associated with the postoperative complications at academic facilities translates to smaller, community-based institutions. In addition, this study only examines cost from the hospital’s perspective. The occurrence of a postoperative complication not only increases hospital costs but has an effect on patient, society, and opportunity costs as well. Therefore, our study almost certainly underestimates the real cost attributable to postoperative complications for the surgical patient.

Postoperative complications are associated with higher mortality and increased LOS and hospital costs. Given the excessive resources required to treat complications, it is necessary to establish quality improvement programs directed at preventing those postoperative events that are the most costly and result in the longest LOS. Reducing the number of postoperative complications may result in cost savings that offset the initiation of the quality improvement programs that rely on clinical data, in particular the NSQIP. This study clearly demonstrates the financial imperative in addition to the patient care imperative for reducing the incidence of postoperative complications.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Pear

. Medicare says it won’t cover hospital errors. http://www.nytimes.com/2007/08/19/washington/19hospital.html?pagewanted=all . Accessed November 2, 2011.

Khan

Quan

Bugar

Lemaire

Brant

Ghali

. Association of postoperative complications with hospital costs and length of stay in a tertiary care center. J Gen Intern Med. 2006;21:177-180.

Ector

Lancellotti

Roberts

. A statement on ethics from the HEART Group. JACC Cardiovasc Imaging. 2008;1:410-412.

Birkmeyer

Dimick

Birkmeyer

. Measuring the quality of surgical care: structure, process, or outcomes? J Am Coll Surg. 2004;198:626-632.

Dimick

Pronovost

Cowan

Lipsett

. Complications and costs after high-risk surgery: where should we focus quality improvement initiatives? J Am Coll Surg. 2003;196:671-678.

Kalish

Daley

Duncan

Davis

Coffman

Iezzoni

. Costs of potential complications of care for major surgery patients. Am J Med Qual. 1995;10:48-54.

Dimick

Chen

Taheri

Henderson

Khuri

Campbell

Jr.

Hospital costs associated with surgical complications: a report from the private-sector National Surgical Quality Improvement Program. J Am Coll Surg. 2004;199:531-537.

Khuri

Daley

Henderson

. The Department of Veterans Affairs’ NSQIP: the first national, validated, outcome-based, risk-adjusted, and peer-controlled program for the measurement and enhancement of the quality of surgical care. National VA Surgical Quality Improvement Program. Ann Surg. 1998;228:491-507.

Khuri

Daley

Henderson

. The comparative assessment and improvement of quality of surgical care in the Department of Veterans Affairs. Arch Surg. 2002;137:20-27.

10.

Fink

Campbell

Jr Mentzer

Jr . The National Surgical Quality Improvement Program in non-Veterans Administration hospitals: initial demonstration of feasibility. Ann Surg. 2002;236:344-353; discussion 353-354.

11.

Khuri

. The NSQIP: a new frontier in surgery. Surgery. 2005;138:837-843.

12.

Shwartz

Young

Siegrist

. The ratio of costs to charges: how good a basis for estimating costs? Inquiry. 1995;32:476-481.

13.

Davenport

Henderson

Khuri

Mentzer

Jr . Preoperative risk factors and surgical complexity are more predictive of costs than postoperative complications: a case study using the National Surgical Quality Improvement Program (NSQIP) database. Ann Surg. 2005;242:463-468; discussion 468-471.

14.

Brasel

Lim

Nirula

Weigelt

. Length of stay: an appropriate quality measure? Arch Surg. 2007;142:461-465; discussion 465-466.

15.

American College of Surgeons National Surgical Quality Improvement Program. About ACS NSQIP. http://www.acsnsqip.org/main/aboutacs/about_procedure_targeted.jsp . Accessed August 27, 2011.

16.

Pronovost

Angus

. Cost reduction and quality improvement: it takes two to tango. Crit Care Med. 2000;28: 581-583.

17.

Davenport

Holsapple

Conigliaro

. Assessing surgical quality using administrative and clinical data sets: a direct comparison of the University HealthSystem Consortium Clinical Database and the National Surgical Quality Improvement Program data set. Am J Med Qual. 2009;24:395-402.