Extended Use of Urinary Catheters in Older Surgical Patients: A Patient Safety Problem?
Objectives. To explore the relationship between the extended postoperative use of indwelling urinary catheters and outcomes for older patients who have undergone cardiac, vascular, gastrointestinal, or orthopedic surgery in skilled nursing facilities and to describe patient and hospital characteristics associated with the extended use of indwelling urinary catheters.
Design. Retrospective cohort study.
Setting. US acute care hospitals and skilled nursing facilities.
Patients. A total of 170,791 Medicare patients aged 65 years or more who were admitted to skilled nursing facilities after discharge from a hospital with a primary diagnosis code indicating major cardiac, vascular, orthopedic, or gastrointestinal surgery in 2001.
Main outcome measures. Patient‐specific 30‐day rate of rehospitalization for urinary tract infection (UTI) and 30‐day mortality rate, as well as the risk of having an indwelling urinary catheter at the time of admission to a skilled nursing facility.
Results. A total of 39,282 (23.0%) of the postoperative patients discharged to skilled nursing facilities had indwelling urinary catheters. After adjusting for patient characteristics, the patients with catheters had greater odds of rehospitalization for UTI and death within 30 days than patients who did not have catheters. The adjusted odds ratios (aORs) for UTI ranged from 1.34 for patients who underwent gastrointestinal surgery (
) to 1.85 for patients who underwent cardiac surgery (
); the aORs for death ranged from 1.25 for cardiac surgery (
) to 1.48 for orthopedic surgery (
) and for gastrointestinal surgery (
). After controlling for patient characteristics, hospitalization in the northeastern or southern regions of the United States was associated with a lower likelihood of having an indwelling urinary catheter, compared with hospitalization in the western region (
vs 
).
Conclusions. Extended postoperative use of indwelling urinary catheters is associated with poor outcomes for older patients. The likelihood of having an indwelling urinary catheter at the time of discharge after major surgery is strongly associated with a hospital's geographic region, which reflects a variation in practice that deserves further study.
Received April 24, 2007; accepted October 9, 2007; electronically published December 31, 2007.
Indwelling urinary catheters are widely used as part of perioperative care to address associated perioperative voiding dysfunction.1,2 The benefits of indwelling catheters do not clearly justify their current level of use, however, given the substantial risk of nosocomial urinary tract infection (UTI) that accrues in direct relationship to the duration of catheterization—the risk increases by 5%‐10% per catheter‐day beyond the first 48 hours of catheterization.3,4 The consequences of nosocomial UTI include prolonged length of stay (LOS), bacteremia, prosthetic joint–associated infection, and death.2,5 In addition to the associated risk of infection, indwelling catheters may dangerously limit patient mobility.6
To date, there is limited research on perioperative bladder management, despite its relevance for nearly all patients who undergo major surgery. Most investigation has concerned patients who underwent orthopedic surgeries in single institutions and has examined outcomes in the immediate postoperative period.2,7‐9 In our previous work, however, we demonstrated an association between extended use of indwelling urinary catheters and poor outcomes in a national sample of patients who had undergone surgery for hip fractures, and we found that indwelling catheter use varies among hospitals in different regions of the country; it was also found to vary depending on whether a hospital was in an urban location and whether a hospital had a teaching affiliation.10 We define “extended” catheter use as the continued presence of an indwelling urinary catheter at the time of discharge from acute care.
Given the results above, our objectives in the present study are to expand on our previous work by (1) exploring the relationship between the extended used of indwelling urinary catheters and outcomes for older patients who have undergone cardiac, vascular, gastrointestinal, or orthopedic surgery and been discharged to skilled nursing facilities; and, (2) describing the patient and hospital characteristics associated with the extended use of indwelling urinary catheters among this population.
Methods
Sample
The study population comprised all Medicare patients aged 65 years or more who were admitted to skilled nursing facilities after discharge from and acute care hospital with a primary diagnosis code of major cardiac, vascular, gastrointestinal, or orthopedic surgery between January 1, 2001, and December 31, 2001. The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9‐CM) codes for cardiac surgery included coronary artery bypass grafting (36.10‐36.17, 36.19, 36.2, 36.3, 36.31, 36.32, and 36.39), open heart valve procedures (35.00‐35.28, 35.96, and 35.99), and other open chest cardiac surgeries excluding transplantation (35.31‐35.32, 35.39‐35.95, 35.98, 36.00, 36.03, 36.09, 36.91, 36.99, 37.10‐37.12, 37.31‐37.35, 37.4, 37.52‐37.54, 37.61‐37.68, and 37.91‐37.99). ICD9‐CM codes for gastrointestinal surgery procedures included colectomy (45.71‐45.76, 45.79, 45.8, 48.41, 48.49, 48.5, 48.61‐48.66, and 48.69), lysis of adhesions (54.5, 54.51, and 54.59), small bowel resection (45.61, 45.62, and 45.63), colostomy (46.10‐46.14), open cholecystectomy (51.21, 51.22, 51.41‐51.43, 5149, 51.51, and 51.59), and exploratory laparotomy (54.11). The codes for vascular surgery procedures included peripheral vascular bypass (39.25 and 39.29) and amputation of lower extremity (84.10‐84.19). Codes for orthopedic surgery procedures included total knee replacement (81.41‐81.47, 81.54, and 81.55) and total or partial elective hip replacement (any of 81.51‐81.53, 81.61‐81.4, and 81.69, excluding hip fracture codes 820.0, 820.01‐820.03, 820.09, 820.20‐820.22, and 820.8).
Patients less than 65 years of age, patients who had already been admitted to a skilled nursing facility or rehabilitation facility after one of the procedures of interest in the same calendar year, and patients whose records were missing information on the presence of a urinary catheter were excluded from analysis.
Data Sources
The Data Analysis PRO Skilled Nursing Facility National Stay File, version 4.0 (DataPRO), was used to obtain information about patients' demographic and clinical characteristics, the index hospitalization, and the skilled nursing facility stay.11 The DataPRO file contained linked hospital stay claims, skilled nursing facility claims, the Minimum Data Set (MDS), and the Medicare Online Survey, Certification, and Reporting database in a single record for each patient admitted to a skilled nursing facility in the year 2001 (
). The MDS is a component of the federally mandated, periodic, comprehensive clinical assessment of residents of nursing facilities.12 The construction of the DataPRO file is described in detail elsewhere.11
Information about hospitals was obtained from the 2001 Healthcare Cost and Utilization Project Nationwide Inpatient Sample.13 The Nationwide Inpatient Sample is a discharge data set containing data on all patients discharged from a representative sample of 20% of hospitals from 33 participating states. The DataPRO file was linked to the 986 sampled hospitals by use of the American Hospital Association and Medicare hospital identification numbers by means of a “crosswalk” file provided by the American Hospital Association.
Variables
Outcomes. The patient‐specific outcomes of rehospitalization for UTI and/or mortality, within 30 days after admission to a skilled nursing facility were determined on the basis of the DataPRO file. Rehospitalization for UTI was defined by ICD9‐CM codes 590.0‐590.9, 601.0‐601.9, 595.0‐595.4, 595.89, 595.9, 597.0, 598.0, and/or 599.0 listed as the primary diagnosis for a qualifying rehospitalization claim within 30 days after of admission to a skilled nursing facility. Mortality was determined from the Medicare Enrollment Database.
Presence of a catheter. Urinary catheterization at the time of admission to a skilled nursing facility was determined from the 5‐day MDS assessment item H.3.d, which is the first assessment for Medicare skilled nursing facility stays. Section H addresses resident continence. Item H.3.d consists of a check box to indicate the presence of an indwelling catheter, which is distinguished from other urinary appliances such as external (ie, “condom”) catheters and intermittent catheters.
Covariates. Outcome models were adjusted for patient‐level and skilled nursing facility–level independent risk factors. Patient‐level independent risk factors included the following: patient demographic characteristics (ie, age, sex, and race), hospital LOS, comorbid conditions (ie, dementia, heart failure, chronic respiratory condition, stroke, cancer, and/or hypertension), in‐hospital complications (ie, delirium, UTI, coma, and/or pressure ulcer), and diagnoses justifying extended use of an indwelling urinary catheter (ie, urinary retention; bladder outlet obstruction; quadri‐, para‐, or hemiparesis; multiple sclerosis; other neurogenic bladder condition; and prostate disease). Diagnoses justifying extended use of an indwelling urinary catheter were determined on the basis of clinical considerations guided by published guidelines that consider complicated urinary retention to be the major indication for extended use of indwelling catheters.14 Overall comorbidity was measured by the Charlson‐Deyo index.15
Also included was information about the patient's postoperative functional status (obtained from the admission MDS at the skilled nursing facility), which was determined on the basis of 5 individual measures of activities of daily living,16 as well as by the modified Barthel index, which is a composite measure of function obtained by summing the scores for activities of daily living.17 The modified Barthel index is scored from 0 to 90, with higher numbers indicating higher levels of function. Finally, postoperative cognitive function at the time of admission MDS assessment was determined by use of the cognitive performance scale score. The cognitive performance scale is a validated measure derived from 5 individual measures on the MDS that reflect cognitive function. It is scored from 0 to 6; higher numbers indicate poorer cognition.18
For the model of rehospitalization for UTI, several additional covariates were included in the predictive models on the basis of previous work (Appendix, Table A)19,20: marital status, renal failure, musculoskeletal disease, dysphagia, presence of a feeding tube, need for assistance with eating, whether the patient was restricted to bed, and presence of a do‐not‐resuscitate order. The overall rehospitalization rate for patients in each skilled nursing facility, generated from the 2001 DataPRO file, was also included in the model of rehospitalization for UTI, because this rate varies with skilled nursing facility characteristics.21,22 In addition, for each surgical group, major surgical procedures and several additional covariates (Appendix, Table A) were included in the predictive models on the basis of a review of the literature.23‐29
Hospital characteristics entered into the model included size, urban location, geographic region, teaching status, and ownership. Geographic regions were the Northeast, South, West, and Midwest; the Midwest served as the reference category. Ownership categories were government, not‐for‐profit, and for‐profit; the for‐profit hospitals served as the reference category.
Analysis
Bivariate comparisons of patient characteristics were conducted for patients with and without indwelling urinary catheters at the time of admission to a skilled nursing facility. The Fisher exact test was used for categorical variables and the Wilcoxon rank sum test was used for continuous variables.
Outcomes. We used logistic regression to evaluate the association between the presence of a catheter and 30‐day outcomes (rehospitalization for UTI and death) for each of the 4 surgical groups, controlling for patient characteristics. The choice of independent variables from the patient‐level risk factors was made on the basis of clinical hypotheses and informed by prior work.10,30 Variable reduction was done on the basis of the significance of each variable in the model, as well as to minimize multicollinearity in the models. The final set of covariates for each of the 8 models appears in the Appendix. The fit of the model for each outcome was assessed by evaluating the area under the receiver–operating characteristic curve (or C statistic), for which a value of 1.0 is ideal, whereas a value of 0.5 is no better than chance.31
Model for catheter presence. To account for the clustering of observations in hospitals, a hierarchical random effects model was used to predict the risk of having an indwelling urinary catheter for an extended postoperative period for all surgical patients, controlling for the characteristics of the patient, surgical procedure, and hospital. This model was estimated using the subset of surgical patients who had been discharged to a skilled nursing facility from 1 of the 538 hospitals for which Nationwide Inpatient Sample data were identifiable. The choice of independent variables from patient‐level and hospital‐level risk factors was made on the basis of clinical hypotheses. The number of variables was reduced on the basis of variable contribution to the model and to minimize multicollinearity in the model. Two skilled nursing facility measures were also included in this model (the number of days before admission MDS assessment for each patient; and the ratio of patients observed to be discharged to the community within 30 days to patients expected to be discharged to the community within 30 days, for all patients admitted to each skilled nursing facility32) because these measures might be related to the chance that the skilled nursing facility would remove the catheter prior to administration of the MDS assessment.
The goodness of fit of this model was assessed by 2 measures. First, the scaled deviance adjusted for degrees of freedom was calculated; desirable values were near 1. Second, a pseudo–multivariate coefficient of determination (pseudo‐R2) was calculated by comparing the log likelihood of the full model to the log likelihood of an intercept‐only model. Statistical analyses were performed with SAS, version 8.02 (SAS Institute).
Results
Patient Characteristics
There were 170,791 patients who met the inclusion criteria for the study, after we excluded those whose records were missing data on the presence of a urinary catheter (12,805 [6.6%]). The group of postoperative patients discharged to skilled nursing facilities following cardiac, vascular, gastrointestinal, and orthopedic surgery yielded 4 distinct patient populations (Table 1). Orthopedic surgery patients had the best overall health status, the shortest hospital LOS, and the lowest mortality and rehospitalization rates. Vascular surgery patients had the worst overall health status and the highest mortality and rehospitalization rates.
A total of 39,282 (23.0%) of all postoperative patients discharged to skilled nursing facilities had urinary catheters at the time of admission to the nursing facility. The catheterization rate at the time of nursing facility admission varied from 18.8% (4,690 patients) among cardiac surgery patients to 28.5% (8,241) among vascular surgery patients and 28.5% (7,379) gastrointestinal surgery patients.
Outcomes
Table 2 presents the results of the logistic regression models for the 2 outcomes in each of the 4 surgical categories. After adjustment for patient characteristics, the analysis shows that patients with catheters in all surgical categories had significantly greater odds of rehospitalization for UTI within 30 days after admission to a skilled nursing facility (adjusted odds ratios [aORs] ranged from 1.34 for gastrointestinal surgery patients [
] to 1.85 for cardiac surgery patients [
]), compared with patients without catheters. Similarly, patients with catheters in all surgical categories had significantly greater odds of death within 30 days after admission to a skilled nursing facility (aORs ranged from 1.25 for cardiac surgery patients [
] to 1.48 for orthopedic surgery patients [
] and 1.48 for gastrointestinal surgery patients [
]). The fit of the logistic regression models for each outcome was good, with C statistics ranging from 0.69 to 0.76 for rehospitalization for UTI and from 0.78 to 0.88 for mortality.
Hospital Characteristics
Data on hospital characteristics were obtained from the 986 hospitals in the 2001 Healthcare Cost and Utilization Project Nationwide Inpatient Sample data set that could be identified with an American Hospital Association number (
) and which had a postoperative patient of interest discharged to a skilled nursing facility during 2001 (
). The descriptive characteristics of these 541 hospitals are shown in Table 3. The study hospitals were large, with an average of 206 staffed beds. The majority were urban (68%) and operated under not‐for‐profit ownership (72%). Twenty‐one percent were teaching hospitals. Study hospitals were evenly distributed across the 4 geographic regions.
Characteristics Associated With Extended Catheterization
Table 4 presents the results of the hierarchical random effects model that used patient and hospital characteristics to predict the risk of a patient having an indwelling urinary catheter at the time of admission to a skilled nursing facility. The model was estimated using data on the subset of 29,719 cardiac, vascular, gastrointestinal, and orthopedic surgery patients who were admitted to 1 of the 541 hospitals for which Nationwide Inpatient Sample data were identifiable (level 1 [patient‐level] effects, 29,719; level 2 [hospital‐level] effects, 541). The sign of the coefficient reflects the direction of, and the size of the coefficient reflects the magnitude of, the association between the covariate and the outcome. The patient characteristics of congestive heart failure, diabetes, a diagnosis that justified catheterization, a pressure ulcer, dependence in toileting, and a cognitive performance scale score reflecting worse cognitive function all had a statistically significant positive association with the extended use of an indwelling urinary catheter. A Barthel score indicating worse physical function and a greater number of days at the skilled nursing facility prior to MDS assessment had a statistically significant negative association with the extended use of an indwelling urinary catheter. LOS demonstrated a U‐shaped association with the extended use of an indwelling urinary catheter: both a short LOS (3 days) and a long LOS (14 days or longer) were associated with having an indwelling catheter (
and 
).
After accounting for patient characteristics, the hospital's geographic region was strongly predictive of the extended use of an indwelling urinary catheter for patients discharged to a skilled nursing facility after major surgery: hospitalization in the Northeast or South was associated with a significantly lower likelihood of having an indwelling urinary catheter, compared with hospitalization in the West (
vs 
). The surgical service in which a patient underwent surgery was not associated with extended indwelling urinary catheterization. The scaled deviance of the model was 1.01. The pseudo‐R2 was 0.12.
Discussion
Adverse outcomes or injuries stemming from the processes of health care are considered threats to patient safety.33 Certainly, the overuse of Foley catheters is a patient safety problem according to this definition, given the known risk of UTI when urinary catheterization extends beyond 48 hours. In this study, we replicated the association between indwelling catheters and UTI in a national sample of surgical patients who underwent extended postoperative catheterization (aORs ranged from 1.34 for gastrointestinal surgery patients [
] to 1.85 for cardiac surgery patients [
]). Further, we have identified an association between the extended use of an indwelling urinary catheter and 30‐day postoperative mortality (aORs ranged from 1.25 for cardiac surgery patients [
] to 1.48 for orthopedic surgery patients [
] and gastrointestinal surgery patients [
]). This observed association is consistent with data suggesting that 15% of patients with nosocomial UTI will develop bacteremic UTI, a known risk factor for death in older patients.5 Despite our exhaustive risk adjustment, we cannot exclude unmeasured confounding as a source of the observed associations. Nonetheless, all of the associations are large and statistically significant for each population studied and consistent with our previous findings in patients with hip fractures.
Although extended use of an indwelling urinary catheter puts patients at risk for adverse outcomes, it is unknown whether reductions in duration of catheterization can be achieved. Our data suggest, however, that extended use of an indwelling urinary catheter is not only associated with patient characteristics but also with characteristics of the surgical hospitalization (LOS and the geographic region of the hospital in which surgery was performed). Thus, variation in postoperative catheter use may reflect practice differences that contribute to poor outcomes. Furthermore, these practices may be amenable to interventions to decrease variation in the duration of catheterization, resulting in a reduced incidence of postoperative catheter‐associated UTI and improved surgical outcomes for older patients.
Significantly, this investigation identified a strong association between a hospital's geographic region and the extended use of indwelling urinary catheters in postoperative patients after major surgery, and this association mirrored the geographic variation we have described for patients with hip fractures.10 Specifically, hospitalization in the Northeast and South resulted in a lower likelihood of extended catheterization, compared with hospitalization in the West. This finding is consistent with regional variations in the use of many healthcare services by the Medicare population, which have been variously attributed to patient demographic characteristics, regional practice styles, local regulations, and the availability of technologies and facilities.34,35 The stable, longstanding differences in hospital LOS among the US geographic regions are one well‐known example of such variation.36
It is also interesting to note that, after risk adjustment for patient characteristics and hospital LOS, the surgical service in which a patient underwent surgery was not predictive of extended use of an indwelling urinary catheter. This finding suggests that the assessment of the performance of this process of care might be done at the hospital level and not for each surgical service.
This study has several limitations. First, its observational nature prevents us from concluding that extended catheter use is a cause of excess deaths. In particular, our risk adjustment cannot account for unmeasured patient characteristics associated with catheterization that may confound the observed relationship between catheter use and death. In addition, the impact of skilled nursing facility practices may be incompletely represented in the models, despite our attempts to include it.
Second, the results of our study may not be generalizable to all surgical patients, primarily because of referral bias regarding the site of postacute care. The subset of patients referred to skilled nursing facilities is likely to be sicker and have worse functional status than the population of all surgical patients. The degree to which this is the case may vary with surgical service. Among skilled nursing facility patients, the distinct characteristics of the 4 postoperative populations reflect both the differential use of subacute rehabilitation and underlying differences in surgical populations. For instance, elective orthopedic surgery patients are more representative of the ambulatory elderly population than the sicker vascular surgery patient population. In addition, subacute rehabilitation following orthopedic surgery is a standard option for many elective orthopedic patients. Accordingly, the mean hospital LOS prior to skilled nursing facility admission was 4.9 days for these patients, similar to the mean hospital LOS for patients who underwent hip and/or knee replacement procedures in 2000.37
Third, as a result of the use of the Healthcare Cost and Utilization Project Nationwide Inpatient Sample to obtain hospital‐level information, the conclusions of this study about hospital practice are based on a small subset of hospitals that may differ significantly from the group of all US hospitals. Consequently, the generalizability of our results to all hospitals, particularly smaller ones, is limited.
Finally, we inferred that catheterization status and the functional and cognitive measures obtained from the 5‐day MDS assessment reflected the patient’s status at the time of transition from the hospital to a skilled nursing facility. As there may have been variability in the timing of the MDS assessment at the skilled nursing facility, we used a “days to MDS assessment” covariate to adjust for this factor for each patient. In addition, this administrative data set is limited by the accuracy of and reliability of the coding reported by the various institutions.
Despite these limitations, the strong association between the extended postoperative use of an indwelling urinary catheterization and both mortality and 30‐day rehospitalization for UTI, as well as the marked regional variation in use, is cause for concern. Future work should confirm this variation in practice, determine why this variation exists, and design interventions to reduce excess postoperative indwelling catheter–days.
Acknowledgments
We thank David Ciesla, MD, Denver Health and Hospital Association.
Financial support. National Research Service Award Primary Care Research Fellowship and the Hartford/Jahnigen Center of Excellence in Aging (to H.L.W.).
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article.
Disclaimer. The data file was created under Centers for Medicare and Medicaid Services Contract 500‐99‐CO01 for the Data Analysis PRO. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, and the mention of trade names, commercial products, or organizations does not imply endorsement by the US government. The authors assume full responsibility for the accuracy and completeness of the ideas presented. This article is a direct result of the Health Care Quality Improvement Program initiated by Centers for Medicare and Medicaid Services, which has encouraged identification of quality improvement projects derived from analyses of patterns of care, and therefore required no special funding on the part of the contractor. Ideas and contributions to the authors concerning experience in engaging with issues presented are welcomed.
Appendix
References
- 1. Wells DL, Saltmarche A. Voiding dysfunction in geriatric patients with hip fracture: prevalence rate and tentative nursing interventions. Orthop Nurs 1986; 5:25‐28.
- 2. Michelson JD, Lotke PA, Steinberg ME. Urinary‐bladder management after total joint‐replacement surgery. N Engl J Med 1988; 319:321‐326.
- 3. Schaeffer AJ. Catheter‐associated bacteriuria. Urol Clin North Am 1986; 13:735.
- 4. Stamm WE. Guidelines for prevention of catheter‐associated urinary tract infections. Ann Intern Med 1975; 82:386.
- 5. Saint S. Clinical and economic consequences of nosocomial catheter‐related bacteriuria. Am J Infect Control 2000; 28:68‐75.
- 6. Saint S, Lipsky B, Goold S. Urinary catheters: a one‐point restraint? Ann Intern Med 2002; 137:125‐127.
- 7. Skelly JM, Guyatt GH, Kalbfleisch R, Singer J, Winter L. Management of urinary retention after surgical repair of hip fracture. CMAJ 1992; 146:1185‐1188.
- 8. Knight RM, Pellegrini JV Jr. Bladder management after total joint arthroplasty. J Arthroplasty 1996; 11:882‐888.
- 9. Johansson I, Athlin E, Frykholm L, Bolinder H, Larsson G. Intermittent versus indwelling catheters for older patients with hip fractures. J Clin Nurs 2002; 11:651‐656.
- 10. Wald H, Epstein A, Kramer A. Extended use of indwelling urinary catheters in postoperative hip fracture patients. Med Care 2005; 43:1009‐1017.
- 11. Kramer AM, DataPRO Project Investigators at the Colorado Foundation for Medical Care, the University of Colorado Health Sciences Center Division of Health Care Policy and Research, Fu Associates, Stepwise Systems. Skilled Nursing Facilities Prospective Payment System Quality Medical Review Data Analysis PRO: Final Report. 7‐31‐2002. Baltimore, MD: Centers for Medicare and Medicaid Services; 2002.
- 12. Hawes C, Morris JN, Phillips C, Mor V, Fries BE, Nonemaker S. Reliability estimates for the minimum data set for nursing home resident assessment and care screening (MDS). Gerontologist 1995; 35:172‐178.
- 13. Agency for Healthcare Research and Quality. NIS Technical Documentation. 2000. Available at: http://www.hcup‐us.ahrq.gov/db/nation/nis/nisdbdocumentation.jsp. Accessed December 20, 2007.
- 14. American Medical Directors Association (AMDA). Clinical Practice Guideline: Urinary Incontinence. 1996. Available at: http://www.amda.com/tools/cpg/incontinence.cfm. Accessed December 20, 2007.
- 15. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD‐9‐CM administrative databases. J Clin Epidemiol 1992; 45:613‐619.
- 16. Lawton MP, Brody EM. Assessment of older people: self maintaining and instrumental activities of daily living. Gerontologist 1969; 9:179‐86.
- 17. Mahoney FL, Barthel DBW. Functional evaluation: the Barthel index. Md State Med J 1965; 14:61‐65.
- 18. Morris JN, Fries BE, Mehr DR, et al. MDS cognitive performance scale. J Gerontol 1994; 49: M174‐M182.
- 19. Fish RH, DataPRO Project Investigators at the Colorado Foundation for Medical Care, the University of Colorado Health Sciences Center Division of Health Care Policy and Research, Fu Associates, Stepwise Systems. Findings from Rehospitalization Transfer within 30 Days Analysis: Summary and Selected Results. Baltimore, MD: Centers for Medicare and Medicaid Services; 2002.
- 20. Eilertsen TB, DataPRO Project Investigators at the Colorado Foundation for Medical Care, the University of Colorado Health Sciences Center Division of Health Care Policy and Research, Fu Associates, Stepwise Systems. Findings from Analysis of Residents Discharged to the Community. Baltimore, MD: Centers for Medicare and Medicaid Services; 2002.
- 21. Carter MW, Porell FW. Variations in hospitalization rates among nursing home residents: the role of facility and market attributes. Gerontologist 2003; 43:175‐191.
- 22. Zimmerman S, Gruber‐Baldini AL, Hebel JR, Sloane PD, Magaziner J. Nursing home facility risk factors for infection and hospitalization: importance of registered nurse turnover, administration, and social factors. J Am Geriatr Soc 2002; 50:1987‐1995.
- 23. Khuri SF, Daley J, Henderson W, et al. Risk adjustment of the postoperative mortality rate for the comparative assessment of the quality of surgical care: results of the National Veterans Affairs Surgical Risk Study. J Am Coll Surg 1997; 185:315‐327.
- 24. Daley J, Khuri SF, Henderson W, et al. Risk adjustment of the postoperative morbidity rate for the comparative assessment of the quality of surgical care: results of the National Veterans Affairs Surgical Risk Study. J Am Coll Surg 1997; 185:328‐340.
- 25. Collins TC, Johnson M, Daley J, Henderson W, Khuri SF, Gordon HS. Preoperative risk factors for 30‐day mortality after elective surgery for vascular disease in Department of Veterans Affairs hospitals: is race important? J Vasc Surg 2001; 34:634‐640.
- 26. Collins TC, Daley J, Henderson WH, Khuri SF. Risk factors for prolonged length of stay after major elective surgery. Ann Surg 1999; 230:251‐259.
- 27. Peterson ED, Delong ER, Muhlbaier LH, et al. Challenges in comparing risk‐adjusted bypass surgery mortality: results from the Cooperative Cardiovascular Project. J Am Coll Cardiol 2000; 36:2174‐2184.
- 28. Dawson I, van Bockel JH. Reintervention and mortality after infrainguinal reconstructive surgery for leg ischemia. Br J Surg 1999; 86:38‐44.
- 29. O'Brien MM, Gonzales R, Shroyer AL, et al. Modest serum creatinine elevation affects adverse outcome after general surgery. Kidney Int 2002; 62:585‐592.
- 30. Levy CR, Fish R, Kramer AM. Site of death in the hospital versus nursing home among Medicare skilled nursing facility residents admitted under Medicare's Part A Benefit. J Am Geriatr Soc 2004; 52:1247‐1254.
- 31. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver–operating characteristic (ROC) curve. Radiology 1982; 143:29‐36.
- 32. Eilertsen TB, DataPRO Project Investigators at the Colorado Foundation for Medical Care, University of Colorado Health Sciences Center Division of Health Care Policy and Research, Fu Associates, Stepwise Systems. Rehabilitation Data Flag Analysis. Baltimore, MD: Centers for Medicare and Medicaid Services; 2004.
- 33. Meyer G, Lewin DI, Eisenberg JM. To err is preventable: medical errors and academic medicine. Am J Med 2001; 110:597‐603.
- 34. The Dartmouth Atlas of Health Care in the United States. Chicago: American Hospital Publishing; 1996.
- 35. Kane RL, Lin W, Blewett LA. Geographic variation in the use of post‐acute care. HSR: Health Services Research 2002; 37:667‐682.
- 36. US Congress, Office of Technology Assessment. Hip Fracture Outcomes In People Age 50 And Over—Background Paper. OTA‐BP‐H‐120. Washington, DC:US Government Printing Office; 1994.
- 37. National Center for Health Statistics. National Hospital Discharge Survey. 2000.
-
Presented in part: National Research Service Award Annual Meeting; Boston, MA; June 2005.