Understanding readmissions in children undergoing surgery: A pediatric NSQIP analysis.

BACKGROUND: Readmission is increasingly being utilized as an important clinical outcome and measure of hospital quality. Our aim was to delineate rates, risk factors, and reasons for unplanned readmission in pediatric surgery. MATERIALS AND METHODS: Retrospective review of pediatric patients (n=130,274) undergoing surgery (2013-2014) at hospitals enrolled in the Pediatric National Surgical Quality Improvement Program (NSQIP-P) was performed. Logistic regression was used to model factors associated with unplanned 30-day readmission. Reasons for readmission were reviewed to determine the most common causes of readmission. RESULTS: There were 6059 (n=4.7%) readmitted children within 30days of the index operation. Of these, 5041 (n=3.9%) were unplanned, with readmission rates ranging from 1.3% in plastic surgery to 5.2% in general pediatric surgery, and 10.8% in neurosurgery. Unplanned readmissions were associated with emergent status, comorbidities, and the occurrence of pre- or postdischarge postoperative complications. Overall, the most common causes for readmission were surgical site infections (23.9%), ileus/obstruction/gastrointestinal (16.8%), respiratory (8.6%), graft/implant/device-related (8.1%), neurologic (7.0%), or pain (5.8%). Median time from discharge to readmission was 8days (IQR: 3-14days). Reasons for readmission, time until readmission, and need for reoperative procedure (overall 28%, n=1414) varied between surgical specialties. CONCLUSION: The reasons for readmission in children undergoing surgery are complex, varied, and influenced by patient characteristics and postoperative complications. These data inform risk-stratification for readmission in pediatric surgical populations, and help to identify potential areas for targeted interventions to improve quality. They also highlight the importance of accounting for case-mix in the interpretation of hospital readmission rates. LEVEL OF EVIDENCE: 3.

Costs of Clostridium difficile infection in pediatric operations: A propensity score-matching analysis.

BACKGROUND: The purpose of this analysis was to assess the burden of Clostridium difficile infection in the hospitalized pediatric surgical population and to characterize its influence on the costs of care. METHODS: There were 313,664 patients age 1-18 years who underwent a general thoracic or abdominal procedure in the Kids' Inpatient Database during 2003, 2006, 2009, and 2012. Logistic regression was used to model factors associated with the development of C difficile infection. A propensity score-matching analysis was performed to evaluate the influence of C difficile infection on mortality, duration of stay, and costs in similar patient cohorts. Population weights were used to estimate the national excess burden of C difficile infection on these outcomes. RESULTS: The overall prevalence of C difficile infection in the sampled cohort was 0.30%, with an increasing trend of C difficile infection over time in non-children's hospitals (P < .001). C difficile infection was associated with younger age, nonelective procedures, increasing comorbidities, and urban teaching hospital status (P < .001). An estimated 1,438 children developed C difficile infection after operation. After propensity score matching, the mean excess duration of stay and costs attributable to C difficile infection were 5.8 days and $12,801 (P < .001), accounting for 8,295 days spent in the hospital and $18.4 million (2012 USD) in spending annually. CONCLUSION: C difficile infection is a relatively uncommon but costly complication after pediatric operative procedures. Given the increasing trend of C difficile infection among hospitalized surgical patients, there is substantial opportunity for reduction of inpatient burden and associated costs in this potentially preventable nosocomial infection.

Timing of postoperative infections after colectomy: evidence from NSQIP.

BACKGROUND: We studied whether risk factors for infectious complications differed between inpatient (IP) and postdischarge (PD) periods in patients undergoing colon surgery. METHODS: Among partial colon resection patients in the National Surgical Quality Improvement Program (2005 to 2010), we identified risk factors for superficial and deep (D-SSI) surgical site infections and urinary tract infections in the IP and PD phases of care. RESULTS: Obesity was associated with higher risk of both IP superficial surgical site infections and D-SSI (odds ratio [OR] 1.41, P < .0001 and OR 1.28, P < .0001) and increasing to OR 1.73 (P < .0001) and OR 1.83 (P < .0001), respectively, in the PD period. Smoking was associated with development of D-SSI, and this risk increased from IP to PD phases of care (OR 1.15, P = .02 to OR 1.54, P < .0001). CONCLUSIONS: Risk factors for infections differ between IP and PD phases of care in colon surgery patients. Earlier discharge from the hospital may shift recognition of an SSI to the outpatient setting.

Comparing 30-day outcomes between thoracoscopic and open approaches for resection of pediatric congenital lung malformations: Evidence from NSQIP.

BACKGROUND/PURPOSE: Thoracoscopic surgery has been increasingly utilized in treating pediatric congenital lung malformations (CLM). Comparative studies evaluating 30-day outcomes between thoracoscopic and open resection of CLM are lacking. METHODS: There were 258 patients identified in pediatric NSQIP with a CLM and pulmonary resection in 2012-2013. Comparisons of patient characteristics and outcomes between surgical approaches were made using standard univariate statistics. In addition, a propensity score match was performed to evaluate outcomes in similar patient cohorts. RESULTS: One-hundred twelve patients (43.4%) received thoracoscopic resections and 146 patients (56.6%) received open resections. Patients undergoing open resections were more likely to be less than 5 months of age and have a comorbidity/preoperative condition (47.3% vs. 25.0%, p<0.001). The extent of resection was a lobectomy in 84.8% of thoracoscopic and 92.5% of open resection patients. Median operative time was similar between both groups (thoracoscopic 172 vs. open 153.5 minutes). On univariate analysis, thoracoscopic resection was associated with decreased postoperative complications (9.8% vs. 25.3%, p=0.001) and LOS (3 vs. 4 days, p<0.001). However, after adjusting for similar patient and operative characteristics, no significant differences were encountered between techniques. CONCLUSIONS: Thoracoscopic and open resection provide comparable 30-day outcomes and safety in the management of congenital lung malformations.

Length of Stay and Readmissions in Mastectomy Patients.

Interest is growing in preventing readmissions as payers start to link reimbursement to readmission rates. The purpose of this study was to assess factors that contribute to 30-day readmission rates for women undergoing mastectomy for breast cancer. Data from the Pennsylvania Health Care Cost Containment Council were queried for women undergoing mastectomy for breast cancer during 2011 (n = 2,919). The outcomes measured were length of stay (LOS) and 30-day readmission. Univariate comparisons between characteristics of readmitted (n = 172) and nonreadmitted patients were performed using t-tests and chi-square tests. Readmission was modeled using logistic regression; LOS was modeled using linear regression and controlled for potential confounders. In multivariate analyses, patients with peripheral vascular disease were more likely to be readmitted (OR 4.36, p = 0.002). Increased LOS was also associated with increased odds of readmission (OR 1.26, p = <0.0001). Since LOS was an important predictor of readmission we also estimated determinants of LOS using linear regression. The occurrence of reconstructive surgery (p = <0.0001) and renal disease (p < 0.0001) were highly predictive of longer LOS. This study showed peripheral vascular disease and longer lengths of stay were associated with higher odds of readmission in women undergoing mastectomy. Clinicians should be cognizant that optimizing a patient's vascular status before mastectomy may lead to lower rates of readmission. Additional research is needed to determine whether the relationship between readmissions and length of hospital stay is a causative versus associative phenomenon since LOS is a modifiable factor that may lead to lower readmissions.

Reframing Surgical Care: Understanding Complexity and Promoting Teaming.

Reframing healthcare delivery in terms of the principles of complex adaptive systems has practical implications for addressing the challenges in improving surgical care. In an Integrated Practice Unit (IPU) - such as a surgical service line, a surgical in-patient floor, or an acute care unit - a diverse group of caregivers must interact in a highly interdependent fashion in an environment characterized by ambiguity, uncertainty, and time constraints. Understanding of the concept of teaming and the tenets of relational coordination are crucial to the promotion of a successful patient-centric approach to surgical care.

The impact of children's hospital designation on outcomes in children with malrotation.

BACKGROUND: The benefit of Ladd's procedure for malrotation at a Children's Hospital (CH) has not previously been established. Our aim was to characterize the potential variations in management and outcomes between CH and Non-Children's Hospitals (NCH) in the treatment of malrotation with Ladd's procedure. METHODS: There were 2827 children identified with malrotation and complete information from the Kids' Inpatient Database (2003, 2006, 2009). Outcomes were compared between CH and NCH and evaluated with logistic and linear regressions. Additional propensity score matching was used to balance covariates between CH and NCH. RESULTS: There were 2261 (80.0%) children with malrotation undergoing Ladd's procedures treated at CH; 566 (20.0%) were treated at NCH. In multivariate analysis, CH was associated with a 39% lower odds of resection (p=0.004), with no differences observed for mortality, morbidity and LOS. Comparison of a propensity score matched cohort confirmed these findings, as well as demonstrated no significant differences in associated costs. CONCLUSIONS: The majority of pediatric intestinal malrotation is managed at CH. While measured outcomes of mortality, morbidity, LOS, and costs were not different at NCH, CH was less likely to perform intestinal resection during Ladd's procedure.

An implemented MRI program to eliminate radiation from the evaluation of pediatric appendicitis.

BACKGROUND: Recent efforts have been directed at reducing ionizing radiation delivered by CT scans to children in the evaluation of appendicitis. MRI has emerged as an alternative diagnostic modality. The clinical outcomes associated with MRI in this setting are not well-described. METHODS: Review of a 30-month institutional experience with MRI as the primary diagnostic evaluation for suspected appendicitis (n=510). No intravenous contrast, oral contrast, or sedation was administered. Radiologic and clinical outcomes were abstracted. RESULTS: MRI diagnostic characteristics were: sensitivity 96.8% (95% CI: 92.1%-99.1%), specificity 97.4% (95% CI: 95.3-98.7), positive predictive value 92.4% (95% CI: 86.5-96.3), and negative predictive value 98.9% (95% CI: 97.3%-99.7%). Radiologic time parameters included: median time from request to scan, 71 minutes (IQR: 51-102), imaging duration, 11 minutes (IQR: 8-17), and request to interpretation, 2.0 hours (IQR: 1.6-2.6). Clinical time parameters included: median time from initial assessment to admit order, 4.1 hours (IQR: 3.1-5.1), assessment to antibiotic administration 4.7 hours (IQR: 3.9-6.7), and assessment to operating room 9.1 hours (IQR: 5.8-12.7). Median length of stay was 1.2 days (range: 0.2-19.5). CONCLUSION: Given the diagnostic accuracy and favorable clinical outcomes, without the potential risks of ionizing radiation, MRI may supplant the role of CT scans in pediatric appendicitis imaging.

Endobronchial occlusion with one-way endobronchial valves: a novel technique for persistent air leaks in children.

PURPOSE: In children, persistent air leaks can result from pulmonary infection or barotrauma. Management strategies include surgery, prolonged pleural drainage, ventilator manipulation, and extracorporeal membrane oxygenation (ECMO). We report the use of endobronchial valve placement as an effective minimally invasive intervention for persistent air leaks in children. METHODS: Children with refractory prolonged air leaks were evaluated by a multidisciplinary team (pediatric surgery, interventional pulmonology, pediatric intensive care, and thoracic surgery) for endobronchial valve placement. Flexible bronchoscopy was performed, and air leak location was isolated with balloon occlusion. Retrievable one-way endobronchial valves were placed. RESULTS: Four children (16 months to 16 years) had prolonged air leaks following necrotizing pneumonia (2), lobectomy (1), and pneumatocele (1). Patients had 1-4 valves placed. Average time to air leak resolution was 12 days (range 0-39). Average duration to chest tube removal was 25 days (range 7-39). All four children had complete resolution of air leaks. All were discharged from the hospital. None required additional surgical interventions. CONCLUSION: Endobronchial valve placement for prolonged air leaks owing to a variety of etiologies was effective in these children for treating air leaks, and their use may result in resolution of fistulae and avoidance of the morbidity of pulmonary surgery.

Determinants of 30-d readmission after colectomy.

BACKGROUND: Readmission after colectomy has become an important metric for measuring quality of care. Our aim was to investigate the impact of patient and hospital characteristics on 30-d readmission rates among patients undergoing colectomies in Pennsylvania. METHODS: Data were obtained from the Pennsylvania Health Care Cost Containment Council, which included all patients undergoing colectomy during 2011 (n = 10,155). Characteristics of non-readmitted and readmitted patients were compared with univariate tests. The primary outcome was 30-d readmission, which was modeled using multivariable logistic regression. RESULTS: Of the 10,155 patients who underwent colectomy, 1492 (14.7%) were readmitted within 30 d of discharge. Readmission was influenced by the underlying diagnosis (P < 0.001). Additionally, readmission was more likely with a Charlson comorbidity index ≥ 2 (odds ratio [OR] = 1.57, P < 0.001), emergent admission (OR = 1.26, P = 0.001), an in-hospital complication (OR = 1.46, P < 0.001), lowest quartile for surgeon volume (OR = 1.24, P = 0.01), and construction of an ileostomy (OR = 2.31, P < 0.001). Factors associated with decreased likelihood of readmission included laparoscopic surgery (OR = 0.73, P < 0.001). No association with hospital volume was found. CONCLUSIONS: A 30-d readmission after colectomy is influenced by numerous patient- and surgeon-related factors. Reducing in-hospital complications, and improving patient education after ileostomy construction, provide substantial targets for intervention. Our data also suggest that there may be a critical range of colectomies performed annually by surgeons, greater than which no additional benefit is conferred in reducing readmissions, but below which there is an increased risk of readmission. Further research is needed to determine the influence of laparoscopic surgery in reducing readmission in equally matched patient populations.

Measuring surgical quality: a national clinical registry versus administrative claims data.

BACKGROUND: This study compared postoperative complications of patients who underwent pancreaticoduodenectomy (PD) recorded in the National Surgical Quality Improvement Program (NSQIP) to patients who underwent PD recorded in the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS). METHODS: Data included 8,822 PD cases recorded in NSQIP and 9,827 PD cases recorded in NIS performed between 2005 and 2010. Eighteen postoperative adverse outcomes were identified in NSQIP and then matched to corresponding International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes in NIS. Using logistic regression, the relationship between database and postoperative complications was determined while accounting for patient factors. RESULTS: Patients undergoing PD in the NIS were more likely to have several adverse outcomes, including urinary tract infection (odds ratio (OR) = 1.42, p < 0.001), pneumonia (OR = 1.51, p < 0.001), renal insufficiency (OR = 2.39, p < 0.001), renal failure (OR = 1.67, p = 0.005), graft/prosthetic failure (OR = 9.35, p < 0.001), and longer length of stay (1.1 days, p < 0.001). They were less likely to have cardiac arrest (OR = 0.45, p = 0.002), postoperative sepsis (OR = 0.38, p < 0.001), deep vein thrombosis (OR = 0.18, p < 0.001), and cerebrovascular accident (OR = 0.04, p = 0.003). CONCLUSIONS: There is considerable discordance between NSQIP and NIS in the assessment of postoperative complications following PD, which underscores the value of recognizing the capabilities and limitations of each data source.

How do risk factors for mortality and overall complication rates following laparoscopic and open colectomy differ between inpatient and post-discharge phases of care? A retrospective cohort study from NSQIP.

BACKGROUND: Risk factors for complications differ between laparoscopic (LC) and open colectomy (OC) patients, given the selection bias between these groups. How risk factors for these outcomes differ between inpatient and post-discharge phases of care requires further study. METHODS: A retrospective cohort study (2005-2010) using NSQIP data was performed comparing OC and LC patients. Multivariable logistic regression was used to compare covariates associated with mortality and overall complication rates both before and after hospital discharge. RESULTS: Patients in the LC cohort were younger (64.2 vs. 62.5 years; P < 0.0001) with a lower incidence of comorbidities. OC was associated with a higher incidence of mortality compared to LC among inpatients (3.3 vs. 0.61%, P < 0.0001) and following discharge (0.88 vs. 0.29%, P < 0.0001). OC also demonstrated a higher incidence of overall complication rates for both inpatients (22.32 vs. 9.36%, P < 0.0001) and following discharge (8.83 vs. 7.24%, P < 0.0001). Risk factors (P < 0.05) for mortality following LC included age and emergency procedures for inpatients; pre-operative SIRS was associated with mortality occurring after discharge. For the OC cohort, risk for mortality was increased with smoking and contaminated/dirty wounds for inpatients; pre-operative weight loss was associated with death following discharge. Factors associated with increased risk of morbidity following LC included smoking history for inpatients and pre-operative steroid therapy following discharge. Following OC, morbidity was strongly associated with ASA scores for inpatients; pre-operative steroid therapy was a risk factor following discharge. Obesity was strongly associated with non-mortal complications in both cohorts following discharge. CONCLUSIONS: (1) LC is associated with a lower incidence of post-operative mortality and complications. (2) Risk factors associated with adverse post-operative outcomes change during the post-operative period; surveillance for these outcomes should be tailored by operative technique and phase of post-operative care (3) Obesity is an underappreciated risk for complications following discharge for both LC and OC.

Surgical specialty and outcomes for carotid endarterectomy: evidence from the National Surgical Quality Improvement Program.

BACKGROUND: Carotid endarterectomy (CEA) has been performed since the 1950s and remains one of the most common surgical procedures in the United States. The procedure is performed by cardiothoracic, general, neurologic, and vascular surgeons. This study uses data from the National Surgical Quality Improvement Program (NSQIP) to examine the outcomes after CEA when performed by general or vascular surgeons. MATERIALS AND METHODS: Data included 34,493 CEAs from years 2005 to 2010 recorded in the NSQIP database. Primary outcomes measured were length of stay, 30-d mortality, surgical site infection, cerebrovascular accident, myocardial infarction, and blood transfusion requirement. Secondary outcomes measured were the remaining intraoperative outcomes from the NSQIP database. RESULTS: After controlling for patient and surgical characteristics, patients treated by general surgeons did not have a significantly different LOS or 30-d mortality than those treated by vascular surgeons. Patients of general surgeons had nearly twice the risk of acquiring a surgical site infection (odds ratio [OR] = 1.94; P = 0.012), >1.5 times the risk of cerebrovascular accident (OR = 1.56; P = 0.008), and >1.8 times the risk of blood transfusion (OR = 1.85; P = 0.017) than those of vascular surgeons. Patients of general surgeons had less than half the risk of having a myocardial infarction (OR = 0.34; P = 0.031) than those of vascular surgeons. CONCLUSIONS: Surgical specialty is associated with a wide range of postoperative outcomes after CEA. Additional research is needed to explore practice and cultural differences across surgical specialty that may lead to outcome differences.

Collaborating to increase access to clinical and educational resources for surgery: a case study.

A case study is described in which collaborations between a Department of Surgery, a Department of Information Technology, and an academic health sciences library resulted in the development of an electronic surgical library available at the bedside, the deployment of tablet devices for surgery residents, and implementation of a tablet-friendly user interface for the institution's electronic medical record.

Impact of surgeon and hospital volume on mortality, length of stay, and cost of pancreaticoduodenectomy.

BACKGROUND: Improved mortality rates following pancreaticoduodenectomy by high-volume surgeons and hospitals have been well documented, but less is known about the impact of such volumes on length of stay and cost. This study uses data from the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) to examine the effect of surgeon and hospital volume on mortality, length of stay, and cost following pancreaticoduodenectomy while controlling for patient-specific factors. METHODS: Data included 3,137 pancreaticoduodenectomies from the NIS performed between 2004 and 2008. Using logistic regression, the relationship between surgeon volume, hospital volume, and postoperative mortality, length of stay, and cost was estimated while accounting for patient factors. RESULTS: After controlling for patient characteristics, patients of high-volume surgeons at high-volume hospitals had a significantly lower risk of mortality compared to low-volume surgeons at low-volume hospitals (OR 0.32, p < 0.001). Patients of high-volume surgeons at high-volume hospitals also had a five day shorter length of stay (p < 0.001), as well as significantly lower costs (US$12,275, p < 0.001). CONCLUSIONS: The results of this study, which simultaneously accounted for surgeon volume, hospital volume, and potential confounding patient characteristics, suggest that both surgeon and hospital volume have a significant effect on outcomes following pancreaticoduodenectomy, affecting not only mortality rates but also lengths of stay and costs.

American College of Surgeons National Surgical Quality Improvement Program Pediatric: a beta phase report.

PURPOSE: The American College of Surgeons (ACS) National Surgical Quality Improvement Program Pediatric (NSQIP-P) expanded to beta phase testing with the enrollment of 29 institutions. Data collection and analysis were aimed at program refinement and development of risk-adjusted models for inter-institutional comparisons. METHODS: Data from the first full year of beta-phase NSQIP-P were analyzed. Patient accrual used ACS-NSQIP methodology tailored to pediatric specialties. Preliminary risk adjusted modeling for all pediatric and neonatal operations and pediatric (excluding neonatal) abdominal operations was performed for all cause morbidity (other than death) and surgical site infections (SSI) using hierarchical logistic regression methodology and eight predictor variables. Results were expressed as odds ratios with 95% confidence intervals. RESULTS: During calendar year 2010, 29 institutions enrolled 37,141 patients. 1644 total CPT codes were entered, of which 456 accounted for 90% of the cases. 450 codes were entered only once (1.2% of cases). For all cases, overall mortality was 0.25%, overall morbidity 7.9%, and the SSI rate 1.8%. For neonatal cases, mortality was 2.39%, morbidity 18.7%, and the SSI rate 3%. For the all operations model, risk-adjusted morbidity institutional odds ratios ranged 0.48-2.63, with 9/29 hospitals categorized as low outliers and 9/29 high outliers, while risk-adjusted SSI institutional odds ratios ranged 0.36-2.04, with 2/29 hospitals low outliers and 7/29 high outliers. CONCLUSION: This report represents the first risk-adjusted hospital-level comparison of surgical outcomes in infants and children using NSQIP-P data. Programmatic and analytic modifications will improve the impact of this program as it moves into full implementation. These results indicate that NSQIP-P has the potential to serve as a model for determining risk-adjusted outcomes in the neonatal and pediatric population with the goal of developing quality improvement initiatives for the surgical care of children.

Synergistic implications of multiple postoperative outcomes.

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.

Cost-effectiveness of the National Surgical Quality Improvement Program.

OBJECTIVE: The purpose of this study was to compare the cost-effectiveness of the National Surgical Quality Improvement Program (NSQIP) at an academic medical center between the first 6 months and through the first and second years of implementation. BACKGROUND: The NSQIP has been extended to private-sector hospitals since 1999, but little is known about its cost-effectiveness. METHODS: Data included 2229 general or vascular surgeries, 699 of which were conducted after NSQIP was in place for 6 months. We estimated an incremental cost-effectiveness ratio (ICER) comparing costs and benefits before and after the adoption of NSQIP. Costs were estimated from the perspective of the hospital and included hospital costs for each admission plus the total annual cost of program adoption and maintenance, including administrator salary, training, and information technology costs. Effectiveness was defined as events avoided. Confidence intervals and a cost-effectiveness acceptability curve were computed by using a set of 10,000 bootstrap replicates. The time periods we compared were (1) July 2007 to December 2007 to July 2008 to December 2008 and (2) July 2007 to June 2008 to July 2008 to June 2009. RESULTS: The incremental costs of the NSQIP program were $832 and $266 for time periods 1 and 2, respectively, yielding ICERs of $25,471 and $7319 per event avoided. The cost-effectiveness acceptability curves suggested a high probability that NSQIP was cost-effective at reasonable levels of willingness to pay. CONCLUSIONS: In these data, not only did NSQIP appear cost-effective, but also its cost-effectiveness improved with greater duration of participation in the program, resulting in a decline to 28.7% of the initial cost.

Electronic measures of surgical site infection: implications for estimating risks and costs.

OBJECTIVE: Electronic measures of surgical site infections (SSIs) are being used more frequently in place of labor-intensive measures. This study compares performance characteristics of 2 electronic measures of SSIs with a clinical measure and studies the implications of using electronic measures to estimate risk factors and costs of SSIs among surgery patients. METHODS: Data included 1,066 general and vascular surgery patients at a single academic center between 2007 and 2008. Clinical data were from the National Surgical Quality Improvement Program (NSQIP) database, which includes a nurse-derived measure of SSI. We compared the NSQIP SSI measure with 2 electronic measures of SSI: MedMined Nosocomial Infection Marker (NIM) and International Classification of Diseases, Ninth Revision (ICD-9) coding for SSIs. We compared infection rates for each measure, estimated sensitivity and specificity of electronic measures, compared effects of SSI measures on risk factors for mortality using logistic regression, and compared estimated costs of SSIs for measures using linear regression. RESULTS: SSIs were observed in 8.8% of patients according to the NSQIP definition, 2.6% of patients according to the NIM definition, and 5.8% according to the ICD-9 definition. Logistic regression for each SSI measure revealed large differences in estimated risk factors. NIM and ICD-9 measures overestimated the cost of SSIs by 134% and 33%, respectively. CONCLUSIONS: Caution should be taken when relying on electronic measures for SSI surveillance and when estimating risk and costs attributable to SSIs. Electronic measures are convenient, but in this data set they did not correlate well with a clinical measure of infection.