PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
J Vasc Surg. Author manuscript; available in PMC Aug 1, 2012.
Published in final edited form as:
PMCID: PMC3150426
NIHMSID: NIHMS271869
Time of Year Does Not Influence Mortality for Vascular Operations at Academic Centers
Damien J. LaPar, MD,* Castigliano M. Bhamidipati, DO,* Glbert R. Upchurch, Jr., MD,* John A. Kern, MD,* Irving L. Kron, MD,* Kenneth J. Cherry, MD,* and Gorav Ailawadi, MD
Division of Thoracic and Cardiovascular Surgery, Department of Surgery, University of Virginia Health System, Charlottesville, VA, USA
*Division of Vascular and Endovascular Surgery, Department of Surgery, University of Virginia Health System, Charlottesville, VA, USA
Correspondence to: Gorav Ailawadi, MD Assistant Professor of Surgery Department of Surgery University of Virginia School of Medicine PO Box 800679 Charlottesville, VA 22908-0679 Fax: (434)-982-3885 Phone: (434)-924-5052 ; gorav/at/virginia.edu
Objectives
Studies in general surgery have suggested worse outcomes due to the presence of new trainees. We hypothesized that outcomes for vascular operations would be equal regardless of teaching hospital status or academic quarter within the United States.
Methods
From 2003-2007, 264,374 vascular operations were evaluated using the Nationwide Inpatient Sample database. Patients were stratified according to Non-Teaching (NTH, n=137,406), Teaching (TH, n=126,968), and Teaching with Vascular Surgery Training Program (VSH, n=28,730) hospital status. Multivariate analyses were used to examine the effect of academic quarter on mortality.
Results
Unadjusted mortality was higher at TH compared to NTH (2.5% vs. 2.0%, p<0.001). Aortic and peripheral vascular operations were more common at TH, while carotid endarterectomy (P<0.001) was more frequent at NTH (P<0.001). After risk adjustment, the odds of death were significantly (P<0.001) increased for aortic and peripheral vascular operations but were similar at both TH (1.11 [0.98-1.25], P=0.10) and VSH (1.16 [0.98-1.37, P=0.08) compared to NTH. Importantly, academic quarter was not associated with increased risk of mortality at either TH (AQ1 OR=0.95 [080-1.13], AQ2 OR=1.08 [0.91-1.28], AQ3 OR= 1.13 [0.96-1.34], AQ4=Reference, P=0.19) or VSH (AQ1 OR=1.02 [0.81-1.29], AQ2 OR=0.99 [0.79-1.25], AQ3 OR=1.02 [0.81-1.28], AQ4= Reference, P=0.99) .
Conclusions
Mortality is not significantly influenced by operative time of year following vascular operations at academic centers. Teaching hospitals perform more high-risk operations compared to non-teaching hospitals with similar risk adjusted mortality.
Keywords: Academic Season, Vascular Surgery, July Effect, July Phenomenon, Outcomes
Academic medical centers in the United States (U.S.) experience significant transitions in patient care at the beginning of each academic year with the arrival of new interns, residents, fellows, and other medical personnel. The term “July phenomenon or effect” at academic medical centers is often used to refer to the influence of new trainees and inexperienced medical staff on patient outcomes. The presence of this influence at training institutions has been the focus of significant investigation.1-4 According to the Institute of Medicine, as well as other studies, inexperienced health practitioners are often targeted as a primary source of increased medical errors and compromised patient outcomes.5-7 Moreover, recent evidence suggests an increase in fatal medical errors during the beginning of the academic year.8
Patient outcomes following surgical operations represent the manifestation of both operative and postoperative events. In addition, the influence of new surgical trainees may also impact various surgical populations differently. Although studies have suggested compromised patient outcomes among general surgical populations,9 the effect of new trainees may be mitigated by a higher level of resident supervision and experience within surgical subspecialties.
The objectives of this study were to examine the effect of academic season on postoperative outcomes and resource utilization among vascular surgery patients and to assess differences in outcomes between teaching and non-teaching hospitals. We utilized a national administrative database to examine the impact of operative time of year within a large, vascular surgery patient population and tested the null hypothesis that outcomes following vascular operations within the United States would be not be affected by teaching hospital status or academic quarter.
Data Source
Due to the absence of patient identifiers and the fact that the data is collected for purposes other than research, the University of Virginia Institutional Review Board (IRB) did not perform a formal review of this study, as the contents did not meet the regulatory definition of human subjects research. This study utilized data obtained from the Nationwide Inpatient Sample (NIS) database for the years 2003-2007. The NIS database is the largest, publicly available all-payer, inpatient care database in the U.S., representing an approximate 20% stratified random sample of all U.S. hospital discharges. NIS includes those hospitals designated as “community hospitals” (“all non-Federal, short-term, general, and other specialty hospitals, excluding hospital units of institutions”) within the American Hospital Association (AHA) Annual Survey. Data includes in-patient hospital discharge records collected for patients of all ages and sources of insurance. A specific discharge weight is assigned for each discharge record, representing the respective relative proportion of the total U.S. in-patient hospital population represented by each record.
Patients
International Classification of Diseases-Ninth Revision, Clinical Modifications (ICD-9-CM) procedure and diagnostic codes were used to identify all patients in the NIS datasets undergoing one of five major vascular operations: carotid endarterectomy (ICD-9-CM codes 3812), open abdominal aortic aneurysm (AAA) repair (ICD-9-CM codes 3844), endovascular AAA repair (ICD-9-CM codes 3971), aortoiliac bypass (ICD-9-CM codes 3925), and peripheral vascular bypass (ICD-9-CM codes 3929). Patients were primarily stratified by teaching and non-teaching hospital status. Academic quarters were defined as follows: Academic Quarter 1 (AQ1: July-September), Academic Quarter 2 (AQ2: October-December), Academic Quarter 3 (AQ3: January-March), and Academic Quarter 4 (AQ4: April-June). The presence of co-morbid disease was assessed for each patient using the different AHRQ co-morbidity categories developed by Elixhauser et. al. 10 The inclusion of all Elixhauser categories during data analysis has been demonstrated to provide effective adjustments for mortality risk, and this approach has been shown to be superior to the use of other comorbidiy indexs, including the Charlson/Deyo method. 11
Hospitals
Hospital related details were obtained through a combination of data available in the NIS database, as well as from the Association of American Medical College's (AAMC) Graduate Medical Education Tracking System. Teaching hospital (TH) and non-teaching hospital (NTH) status were defined by criteria in the NIS dataset as any hospital with the presence of any Accreditation Council for Graduate Medical Education (ACGME) resident training program regardless of specialty. Vascular surgery teaching hospitals (VTH) were identified through linking of AHA identification numbers within the NIS dataset with the AAMC's Graduate Medical Education Tracking System. Hospital operative volume was categorized into quartiles based on frequency of vascular operations (Low [<25th percentile], Medium [26-49th percentile], High [50-74th percentile], and Very High [>75th percentile]. Remaining hospital variables reflect definitions utilized within the NIS database.
Outcomes Measured
All measured outcomes were established a priori before data collection and analysis. The primary outcomes of interest in this study were in-hospital mortality and complications. Secondary outcomes were hospital length of stay and total costs. In-hospital complications were categorized according to the ICD-9-CM based coding scheme developed by Guller and colleagues,12 including wound, infections, urinary, pulmonary, gastrointestinal, cardiovascular, systemic and procedure related complications. Death occurring during the in-patient stay was identified from the patients’ discharge status. Length of stay and total costs were determined from discharge records.
Statistical Analysis
The associations between patient risk factors, hospital features, normally distributed continuous variables and unadjusted outcomes were assessed using analysis of variance (ANOVA),while the Mann Whitney U test was used for non-normally distributed outcomes. Pearson's χ2 or Fisher's exact tests were utilized for all categorical variable comparisons as appropriate. All group comparisons were unpaired.
Separate multivariable logistic regression models were utilized to estimate confounder adjusted measures of association for each outcome of interest among patients undergoing vascular surgery operations. All model covariates (patient age, gender, elective operative status, hospital geographic region, vascular surgery teaching hospital status, type of operation (carotid endarterectomy, open AAA repair/replacement, endovascular AAA repair, aortoiliac bypass, and peripheral vascular bypass), operative year, operative volume, academic quarter, and Elixhauser categories of co-morbid disease) were selected a priori based upon established clinical risk or were considered potential confounders for the effect of academic quarter on surgical mortality. All covariates contributing cases to each estimated outcome were retained in the final models, including all non-significant variables. All logistic regression models included appropriate adjustments for variance components estimated from the weighted study population. The Wald χ2 test was used to assess statistical significance of the association between academic quarter and in-hospital mortality. Area Under the Receiver Operating Characteristics Curve (AUC) was used to assess the discrimination achieved by logistic regression models. Model calibration across deciles of observed and predicted risk was assessed using the Hosmer-Lemeshow test.
Sensitivity analysis was performed for predictive models. For each model, sensitivity was assessed to evaluate the possibility that the estimated effect of academic quarter on outcomes could be a spurious result, reflecting the influence of a closely related but unmeasured confounder. Each model was re-estimated after removing the most statistically significant covariate as measured by the Wald statistic. With this approach, the potential for spurious results is reduced if the originally observed effect is not substantially attenuated and remains statistically significant after re-estimation.13 After removing the most significant covariate from each logistic regression model, the effect of academic quarter on the estimated odds of patient death was not significantly attenuated (<10%), validating the sensitivity of each original model.
Within-group percentages are used to express the frequency of each categorical variable. Normally distributed continuous variables (age) are reported as means ± standard deviation while the median [interquartile range] is reported for all non-normally distributed data (hospital length of stay and total costs). Confounder adjusted odds ratios (OR) with a 95% confidence interval (CI) are used to report the results of logistic regression models. Reported p-values are two-tailed and reflect adjustment using the Sidak correction for multiple comparisons. Statistical significance was identified by p-values < 0.05. Data manipulation and analysis were performed using SPSS software, version 17 (SPSS, Chicago, IL).
Patient and Hospital Characteristics
During the six-year study period, a total of 264,374 discharge records for patients undergoing vascular surgery operations, representing a weighted estimate of 1,284,801 patients nationwide, were identified. Frequencies of all patient and hospital related characteristics stratified by hospital type are listed in Table 1. Overall, NTH performed slightly more vascular operations compared to TH (52.0% vs. 48.0%, P<0.001) while 10.7% of all vascular procedures were performed at vascular surgery teaching hospitals. Elective operations were slightly more frequent among NTH (75.5% vs. 72.9%, P<0.001). Performance of carotid endarterectomy was more common at NTH while open and endovascular AAA repair, aortoiliac bypass, and peripheral vascular bypasses were more common at TH. Mean patient age was clinically comparable between study groups, and male gender was most common despite hospital teaching status. Patient co-morbidities and risk factors were not affected by teaching hospital status. The overall frequency of major cardiovascular and pulmonary disease states included: congestive heart failure (6.0%), chronic pulmonary disease (25.6%), hypertension (71.3%), peripheral vascular disorders (27.8%), pulmonary circulatory disease (0.5%), and valve disease (4.3%). Performance of vascular surgery operations was similar among mean income groups, and the distribution of operations was similar among academic quarters and operative year regardless of hospital status. The large majority of vascular operations occurred in an urban setting for all patient groups and within large hospital bed size hospitals. The Southern geographic region performed the highest proportion of procedures (South [41.9%], Northeast [18.6%], Midwest [23.1%], West [16.3%], P=0.001). Vascular operations were more commonly performed at very high-volume (>75th percentile operative volume) centers (Very High Volume [71.6%], Low [1.2%], Medium [7.2%], High [20.1%], P <0.001).
Table 1
Table 1
Preoperative and operative characteristics for all patients undergoing vascular operations at non-teaching versus teaching hospitals.
Unadjusted Associations with Postoperative Outcomes
The unadjusted incidence of postoperative outcomes at non-teaching versus teaching hospitals appears in Table 2. Overall, complications were more common at teaching hospitals: infectious (1.2%), genitourinary (1.4%), pulmonary (4.5%), gastrointestinal (1.3%), and procedure related (2.5%). Median hospital length of stay was 3.0 [1.0-6.0] days for non-teaching hospitals and 3.0 [1.0-8.0] days for teaching hospitals, while median total costs were $26,925 $15,497-$50,933] and $32,226 [$17,432-$61,369], respectively. Unadjusted mortality was lower at non-teaching hospitals compared to teaching hospitals (2.0% vs. 2.5%, P<0.001).
Table 2
Table 2
Unadjusted incidence of postoperative outcomes for all patients undergoing vascular operations at non-teaching versus teaching hospitals.
When stratified by major vascular operations (Table 3), non-teaching hospitals were associated with incrementally lower overall, pulmonary and cardiovascular complications for all operations. Mortality following CEA did not differ based on hospital type, while the incidence of death was statistically higher at either teaching or vascular surgery hospitals for all aortic and peripheral vascular operations compared to non-teaching hospitals.
Table 3
Table 3
Relationship between type of surgery and outcomes for patients undergoing vascular operations at teaching (TH), vascular surgery (VSH), and non-teaching (NTH) hospitals.
Unadjusted Comparisons of Academic Quarters
The unadjusted frequency of selected patient risk factors, operative characteristics and incidence of postoperative outcomes are detailed in Table 4. As expected, incremental differences were observed for patient age, female gender and performance of non-elective operations across academic quarters. Carotid endarterectomy was the most commonly performed operation in each academic quarter, followed by peripheral vascular bypass procedures. Unadjusted mortality (2.1%, P<0.05) was lowest during AQ1, while few differences were observed in overall, pulmonary, and cardiovascular complication rates between academic quarters. Hospital length of stay was not affected by academic quarters , while AQ2 incurred the highest total costs among all academic quarters.
Table 4
Table 4
Unadjusted association between academic quarter and select patient risk factors, operative features and postoperative outcomes.
Adjusted Associations for Postoperative Outcomes
After adjustment for the confounding effects of patient risk factors, hospital volume and location, and operations performed, neither teaching hospitals (1.11 [0.98-1.25], P=0.10) nor vascular surgery teaching hospitals (1.16 [0.98-1.37, P=0.08) were associated with increased risk-adjusted mortality compared to NTH. Furthermore, academic quarter was not an independent predictor of in-hospital mortality (P=0.09). In addition, upon testing of the interaction between hospital type and academic quarter, the association between academic quarter and patient mortality did not depend on hospital type: TH vs. NTH (P=0.19) and VSH vs. NTH (P=0.99). Further, as expected, the odds of in-hospital death were significantly increased for open (OR=12.4 [11.2-13.78], P<0.001) and endovascular (OR=7.86 [6.95-8.89], P<0.001) AAA repairs, aortoiliac bypass (OR=2.92 [2.55-3.34], P<0.001), and peripheral vascular bypass (OR=2.63 [2.38-2.91], P<0.001) operations compared to carotid endarterectomy.
Additional multivariable logistic regression models for confounder adjusted mortality and morbidity were used to test the interaction between type of vascular operation and type of hospital (Table 5). After adjustment, the effect of hospital type and type of vascular operation were found not to depend upon each other with respect to mortality. Similarly, few statistical interactions were revealed between the effects of hospital type and type of vascular operation for the composite outcome of any postoperative complication.
Table 5
Table 5
Adjusted odds ratios for the effect of operation type versus carotid endarterectomy (CEA) within either teaching (TH) or vascular surgery (VSH) hospitals related to each outcome of interest.
The present study is a comprehensive review of contemporary outcomes for vascular surgery operations at teaching and non-teaching institutions. Moreover, these findings are the first to examine the effect of time of year of vascular surgery operations at U.S. academic medical centers. These results demonstrate that teaching hospital status has no effect on outcomes after accounting for confounding differences in patient risk factors and operations. Academic quarter and time of year of operation did not independently influence patient mortality following vascular surgical procedures, reflecting the absence of a new-trainee influence on patient outcomes. Moreover, this study demonstrates the performance of more complex, higher risk operations at teaching hospitals. Despite these differences, patients undergoing vascular operations at teaching hospitals accrued only slightly longer hospital lengths of stay compared to patients undergoing operations at non-teaching hospitals.
The influence of time of year of operation has been postulated as a potential explanation for variations in patient outcomes following both medical and surgical hospitalizations at academic centers. This effect has been investigated within various patient populations with mixed results.1-4, 9 In a large, multi-institutional cohort study of 20,254 patients undergoing various surgical operations as collected in the American College of Surgeons-National Surgical Quality Improvement Program (NSQIP), Englesbe and colleagues (2007) demonstrated an impressive 41% increase in the risk of mortality for operations performed during July and August.9 Similarly, Rich and colleagues (1993) demonstrated a similar pattern of outcomes among patients with non-surgical diseases.3 However, several other studies have failed to demonstrate such an effect on surgical outcomes,2-4 including cardiac surgery operations performed at teaching hospitals.1 In another study of over 300,000 Medicare beneficiaries, the authors concluded that the beginning of the academic year was a safe time to undergo major surgical operations, including coronary artery bypass grafting, elective abdominal aortic aneurysm repair, carotid endarterectomy, pancreatectomy, esophagectomy, coloectomy, and hip repair following fracture.2 Similarly, Bakeen and colleagues documented similar risk-adjusted outcomes for over 70,000 Veterans Affairs (VA) patients undergoing cardiac operations during early (July 1-August 31) and late (September 1-June 31) academic periods.1 These results are in agreement with those of the present study.
The findings of the present study extend the examination of resident work experience and the investigation of academic season to reflect contemporary trends at academic medical centers spanning current eighty-hour workweek restrictions for surgical trainees. In our analyses, we controlled for comorbidities and other confounders through the inclusion of over 45 patient, hospital and operation related factors, including vascular surgery teaching hospital status, in each of our predictive models. Importantly, the combined influence of academic quarter and hospital status was assessed through the inclusion of an interaction term between these two variables in each statistical model. Even after these adjustments, performance of operations during a given academic quarter at both teaching and vascular surgery hospitals was not associated with increased patient mortality. In addition, the present results revealed differences in resource utilization between those undergoing operations at TH versus NTH, demonstrating slightly longer median hospital lengths of stay (LOS) and higher median total costs for those undergoing operations at TH. These findings are not surprising given the performance of more complex operations at academic medical centers.
Discrepancies in types of operations performed at teaching and non-teaching hospitals may explain the differences observed. Patients undergoing operations at teaching hospitals more frequently underwent complex and higher risk operations, including open and endovascular AAA repairs and aortoiliac bypass operations. In the Dutch Randomized Endovascular Aneurysm Management (DREAM) trial, 30 day mortality rates for endovascular aneurysm repair (EVAR) were 1.2%, while those for open aneurysm repair (OAR) was 4.6%.14 A second trial from the United Kingdom demonstrated similar early perioperative mortality rates for both EVAR (1.7%) and OAR (4.7%).15 In our study, we observed similar trends, documenting a 12 fold increase in the odds of death for patients undergoing OAR, a 7.9 fold increase in the odds of death for those undergoing EVAR, and 3.9 fold increased risk of death for those undergoing aortoiliac bypass operations compared to carotid enderarterectmy. Alternatively, non-teaching hospitals more frequently performed carotid endarterectomies. As carotid endarterectomy is a commonly performed operation with low perioperative mortality rates of approximately 0.5%,16 the higher frequency of these operations at non-teaching hospitals was postulated to explain the lower mortality on univariate analysis at these institutions. However, after risk adjustment, these discrepancies did not translate into higher adjusted mortality at either teaching or vascular surgery hospitals upon testing the interaction between major vascular operation type and hospital type. In fact, we observed no significant differences in the adjusted odds of mortality when testing this important interaction. Similarly, a slightly higher incidence of non-elective (urgent/emergent) operations at academic centers may have contributed to higher rates of postoperative cardiopulmonary and procedure related complications and increased mortality. However, following risk adjustment, the contribution of these differences on outcomes was small.
This study has important clinical relevance as it provides a nationally representative and broadly generalizable sampling of an increasingly reported epidemiologic phenomenon to surgical outcomes literature. The existing medical and surgical literature is mixed regarding the true influence of seasonality at academic institutions. Our results also provide important insight into surgical subspecialty care. As surgical subspecialty care often requires more specialized postoperative management, we believe that the lack of differences in patient outcomes at the beginning of an academic year may have several potential explanations. First, these findings may represent the result of a higher level of supervision from more senior surgical trainees and attending surgeons. Next, the presence of more mid-level providers often present on subspecialty surgical services can maintain consistent postoperative care during the period of new trainees. Finally, much of the postoperative care is protocolized to avoid variation in postoperative care. Consequently, we believe our data dispels the notion that patient outcomes differ at teaching institutions during the early academic season.
Despite these interesting results, this study has several limitations and potential considerations that warrant further discussion. First, inherent selection bias must be considered due to its retrospective design; however, the strict methodology and randomization of the NIS database likely reduces this bias. Second, the potential for unrecognized miscoding of diagnostic, procedure, and complication codes must be considered. The NIS database designates hospitals as teaching hospitals if there is any ACGME-accredited residency program, regardless of specialty, and these designations serve as a potential source of confounding. Additionally, true perioperative mortality and morbidity rates may be underestimated as they may have occurred following the patient's discharge. In our analyses, we categorized the operative time of year into academic quarters as opposed to analyzing the data by individual operative months. While this approach has been reported in similar series, it may dilute reported outcomes that may otherwise be observed within individual operative month. Nevertheless, we believe that such stratification appropriately captures and reports variations in patient outcomes that they exist throughout the academic year. The potential for a mortality bias in the hospital length of stay analysis should be considered. However because the mortality rates are rather small in both study groups, and fairly similar, such a bias would likely not be large. The de-identified nature of this administrative database limits us to only comment on in-hospital, short-term outcomes. In addition, we are unable to include adjustments for other well-established surgical risk factors, such as low preoperative albumin levels or poor nutrition status, as they are not captured in the NIS databases. However, upon sensitivity analyses our statistical models proved resilient to the presence of a potentially unmeasured confounder.
CONCLUSION
In this study, risk-adjusted mortality is not affected by teaching hospital status, and mortality is not significantly influenced by operative time of year for patients undergoing vascular procedures at academic centers. Moreover, teaching hospitals perform a higher percentage of more complex, high-risk vascular operations compared to non-teaching hospitals. Performance of vascular operations during the early academic year is not found to be associated with worse outcomes than the rest of the year.
ACKNOWLEDGMENTS
We would like to thank George J. Stukenborg, PhD of the University of Virginia Department of Public Health Sciences for his statistical guidance and collaboration in this study.
Funding: This study was supported by Award Number 2T32HL007849-11A1 (DJL, CMB) from the National Heart, Lung, And Blood Institute and the Thoracic Surgery Foundation for Research and Education Research Grant (GA). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Heart, Lung, And Blood Institute or the National Institutes of Health.
Footnotes
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Presented at the 34th Annual Meeting of the Midwestern Vascular Surgical Society, September 11, 2010, Indianapolis, IN
1. Bakaeen FG, Huh J, Lemaire SA, Coselli JS, Sansgiry S, Atluri PV, et al. The July effect: impact of the beginning of the academic cycle on cardiac surgical outcomes in a cohort of 70,616 patients. Ann Thorac Surg. 2009;88(1):70–5. [PubMed]
2. Englesbe MJ, Fan Z, Baser O, Birkmeyer JD. Mortality in medicare patients undergoing surgery in July in teaching hospitals. Ann Surg. 2009;249(6):871–6. [PubMed]
3. Rich EC, Hillson SD, Dowd B, Morris N. Specialty differences in the ‘July Phenomenon’ for Twin Cities teaching hospitals. Med Care. 1993;31(1):73–83. [PubMed]
4. Schroeppel TJ, Fischer PE, Magnotti LJ, Croce MA, Fabian TC. The “July phenomenon”: is trauma the exception? J Am Coll Surg. 2009;209(3):378–84. [PubMed]
5. Altman DE, Clancy C, Blendon RJ. Improving patient safety--five years after the IOM report. N Engl J Med. 2004;351(20):2041–3. [PubMed]
6. Phillips J, Beam S, Brinker A, Holquist C, Honig P, Lee LY, et al. Retrospective analysis of mortalities associated with medication errors. Am J Health Syst Pharm. 2001;58(19):1835–41. [PubMed]
7. Weingart SN, Wilson RM, Gibberd RW, Harrison B. Epidemiology of medical error. BMJ. 2000;320(7237):774–7. [PMC free article] [PubMed]
8. Phillips DP, Barker GE. A july spike in fatal medication errors: a possible effect of new medical residents. J Gen Intern Med. 2010;25(8):774–9. [PMC free article] [PubMed]
9. Englesbe MJ, Pelletier SJ, Magee JC, Gauger P, Schifftner T, Henderson WG, et al. Seasonal variation in surgical outcomes as measured by the American College of Surgeons-National Surgical Quality Improvement Program (ACS-NSQIP). Ann Surg. 2007;246(3):456–62. discussion 63-5. [PubMed]
10. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27. [PubMed]
11. Southern DA, Quan H, Ghali WA. Comparison of the Elixhauser and Charlson/Deyo methods of comorbidity measurement in administrative data. Med Care. 2004;42(4):355–60. [PubMed]
12. Guller U, Hervey S, Purves H, Muhlbaier LH, Peterson ED, Eubanks S, et al. Laparoscopic versus open appendectomy: outcomes comparison based on a large administrative database. Ann Surg. 2004;239(1):43–52. [PubMed]
13. Lin DY, Psaty BM, Kronmal RA. Assessing the sensitivity of regression results to unmeasured confounders in observational studies. Biometrics. 1998;54(3):948–63. [PubMed]
14. Prinssen M, Verhoeven EL, Buth J, Cuypers PW, van Sambeek MR, Balm R, et al. A randomized trial comparing conventional and endovascular repair of abdominal aortic aneurysms. N Engl J Med. 2004;351(16):1607–18. [PubMed]
15. Greenhalgh RM, Brown LC, Kwong GP, Powell JT, Thompson SG. Comparison of endovascular aneurysm repair with open repair in patients with abdominal aortic aneurysm (EVAR trial 1), 30-day operative mortality results: randomised controlled trial. Lancet. 2004;364(9437):843–8. [PubMed]
16. Harthun NL, Baglioni AJ, Jr., Kongable GL, Meakem TD, Cherry KJ. Carotid endarterectomy: update on the gold standard treatment for carotid stenosis. Am Surg. 2005;71(8):647–51. discussion 51-2. [PubMed]