Skip Navigation U.S. Department of Health and Human Services
Agency for Healthcare Research Quality
Archive print banner

Research Briefs

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to for current information.

Benesch, C., Witter, Jr., D.M., Wilder, A.L., and others (1997, September). "Inaccuracy of the International Classification of Diseases (ICD-9-CM) in identifying the diagnosis of ischemic cerebrovascular disease." (AHCPR PORT contract 290-91-0028). Neurology 49, pp. 660-664.

Using administrative databases to analyze a specific procedure or medical condition typically involves using standardized classification schemes such as the International Classification of Diseases, Version 9, Clinical Modification (ICD-9-CM). However, certain conditions may not be accurately reflected by the ICD-9 codes. In fact, this study shows that the ICD-9 coding scheme may be inaccurate in the classification of patients with ischemic cerebrovascular disease. The researchers compared ICD-9 codes in an administrative database of patients hospitalized in five academic medical centers in 1992 with clinical findings from the patients medical records. For example, for ICD-9 code 434, 85 percent of patients were classified as having a stroke, and for ICD-9 code 435, 77 percent had transient ischemic attacks. For code 436, 77 percent of patients were classified as having strokes. According to the authors, the results from this study raise doubts about the diagnostic accuracy of ICD-9 discharge codes for identifying patients with ischemic cerebrovascular disease. Selection based on specific ICD-9 codes for this disease may be affected by the number of discharge diagnoses used, the concomitant procedures performed during the index admission, and most importantly by the actual clinical condition most likely represented by the identifying code. When ICD-9 codes are the sole basis for patient selection from administrative databases, these limitations may confound conclusions drawn concerning outcomes, variations in health care delivery, and cost effectiveness of treatment, caution the researchers.

Elixhauser, A., Steiner, C., Harris, R., and Coffey, R. (1997). "Comorbidity measures for use with administrative data." Medical Care 36(1), pp. 8-27.

When using administrative data, preexisting conditions (or comorbidities) are handled analytically by stratifying patients into groups—those with coexisting medical conditions and those without; employing separate binary indicators for discrete conditions; or summarizing comorbidity information into an index or score that provides a single parameter for measuring multiple comorbidities. Since comorbidities affect outcomes differently among different patient groups, they probably should not be simplified as an index, conclude these authors. They developed a comprehensive set of 30 comorbidity measures for use with administrative inpatient databases to control for a broad array of patients underlying preexisting conditions in many types of studies. The comorbidities were associated with substantial increases in length of stay, hospital charges, and mortality both for heterogeneous and homogeneous disease groups. The authors point out several comorbidities that are important predictors of outcomes, yet are not commonly measured. These include: mental disorders, drug and alcohol abuse, obesity, coagulopathy, weight loss, and fluid and electrolyte disorders. Reprints (AHCPR Publication No. 98-R013) are available from the AHCPR Publications Clearinghouse.

Keeler, E.B., Park, R.E., Bell, R.M., and others (1997, October). "Adjusting cesarean delivery rates for case mix." (AHCPR PORT contract 290-90-0039). Health Services Research 32(4), pp. 511-528.

Cesarean delivery rates are one of the first measures used to judge hospital and health plan performance. But to properly judge quality of care, cesarean delivery rates must be adjusted for the mix of patients. Hospitals with more women with problem pregnancies would be expected to have higher cesarean delivery rates. These researchers analyzed merged hospital and birth certificate data for singleton births greater than 2,500 grams in Washington State in 1989 and 1990 to develop models to predict the probability that any given mother would have a cesarean delivery.

They found that four factors—prior cesarean, breech but no prior cesarean, first birth, and other (factors)—explained 30 percent of the variance in individual cesarean rates. The full clinical model fit the data well and explained 37 percent of the variance. For instance, women who had delivered previously and had no serious complications accounted for 35 percent of the mothers and averaged less than 2 percent of cesareans. The authors conclude that adjustments for case mix should not include all variables related to cesarean delivery rates, only the most predictive ones. Proper adjustments may not alter hospital rankings greatly, but they will improve the validity and acceptability of the reports.

Parmigiani, G., Samsa, G.P., Ancukiewicz, M., and others (1997, October)."Assessing uncertainty in cost-effectiveness analyses: Application to a complex decision model." (AHCPR contract 290-91-0028). Medical Decision Making 17, pp. 390-401.

A medical intervention's average cost and average effectiveness may be used to derive the marginal cost-effectiveness ratio (MCER) for any two interventions. The MCER is defined as the difference in average cost divided by the difference in average effectiveness. Estimating the precision of MCERs is critical to clinical decisionmaking. These authors present a framework for quantifying uncertainty about costs, effectiveness measures, and MCERs in complex decision models. They discuss two alternative approaches, one based on Bayesian inference and the other on resampling. While computationally intensive, these models are flexible in handling complex distributional assumptions and a variety of outcome measures, according to the authors. They conclude by extending their simplified models to a complex decision model using the stroke-prevention policy model.

Preisser, J.S., and Koch, G.G. (1997). "Categorical data analysis in public health." (AHCPR grant HS06992). Annual Review of Public Health 18, pp. 51-82.

This article surveys categorical data methods widely applied in public health research. Whereas large sample chi-square methods, logistic regression analysis, and weighted least squares modeling of repeated measures once were the primary analytic tools for categorical data problems, today's methodology includes a much broader range of tools made available by increasing computational efficiency. These include computational algorithms for exact inference of small samples and sparsely distributed data and generalized estimating equations for cluster samples. The researchers illustrate the various methods with examples, including a study of the prevalence of cerebral palsy in very low birthweight infants and a study of cancer screening in primary care settings.

Scharfstein, J.A., Paltiel, A.D., and Freedberg, K.A. (1997, October). "The cost-effectiveness of fluconazole prophylaxis against primary systemic fungal infections in AIDS patients." (AHCPR grant HS07595). Medical Decision Making 17, pp. 373-381.

People with AIDS are susceptible to systemic fungal infections, a relapsing and life-threatening group of illnesses that increases in incidence as CD4 lymphocyte counts decrease. However, the drug fluconazole is unlikely to be cost effective in preventing these infections unless its cost ($206) is lowered or it is focused on patients in regions with endemic fungal infections, conclude these researchers. They used a Markov model with data from the research literature to project the cost-effectiveness of fluconazole for prophylaxis against AIDS-related primary systemic fungal infections in a hypothetical group of 100,000 AIDS patients. A no-prophylaxis policy was associated with a discounted life expectancy of 28.2 months and direct medical costs of $36,100 per person. The strategy of treating patients with CD4 counts less than 200 increased costs to $40,500 and life expectancy to 28.4 months, producing a cost-effectiveness ratio (CER) of $240,000 per year of life saved (YLS). Compared with these two approaches, the intermediate alternatives (fluconazole when CD4 counts reached less than 100 or less than 50) were less economically efficient. A reduction in fluconazoles cost from $206 to $80 decreased the CER to $50,000 for the under 200 CD4 strategy. Doubling fungal infection incidence lowered this ratio to $96,000 per YLS.

Stineman, M.G., Goin, J.E., Granger, C.V., and others (1997, September). "Discharge motor FIM-function related groups." (AHCPR grant HS07595). Archives of Physical Medicine and Rehabilitation 78, pp. 980-985.

These researchers describe the development of a classification system for medical rehabilitation patients that is based on the Functional Independence Measure (FIM), which is designed to identify individuals likely to achieve similar motor FIM scores at discharge from rehabilitation. The Discharge Motor FIM-Function Related Groups (DMF-FRGs) are part of the expanding FIM-FRG system that includes an array of prediction tools for medical rehabilitation. The researchers grouped 84,492 rehabilitation inpatients discharged in 1992 into 20 impairment categories and then into FRGs by their admission motor FIM scores. Some FRGs were also subdivided on the basis of admission cognitive FIM scores and age. The entire system explained 63 percent of the variation in motor FIM discharge scores in the validation data set. The researchers conclude that clinicians can use the DMF-FRGs to identify groups of patients whose motor FIM scores at discharge are below, within, or above nationally established ranges of values for the purpose of outcomes management, guideline development, and quality improvement.

Stineman, M.G., Tassoni, C.J., Escarce, J.J., and others (1997, October). "Development of function-related groups version 2.0: A classification system for medical rehabilitation." (AHCPR grant HS07595). Health Services Research 32(4), pp. 529-548.

This paper describes development of a new version of the system to classify medical rehabilitation patients according to their use of rehabilitation resources. The system, the Functional Independence Measure-Function Related Groups (FIM-FRGs), classifies patients into groups that are similar with respect to rehabilitation length of stay. The researchers used data from 85,447 patient discharges in 1992 from 252 freestanding rehabilitation facilities and hospital units to create FIM-FRGs Version 2.0. This version incorporates clinical and statistical criteria to increase the percentage of patients classified, expand the impairment categories, and evaluate the incremental predictability of coexisting medical diagnoses. This updated system explained 32 percent of the variance in length of stay in the 1992 validation sample and 31 percent in the 1990 discharges. The researchers conclude that FIM-FRGs Version 2.0 includes more specific impairment categories, classifies a higher percentage of patient discharges, and appears sufficiently stable over time to form the basis of a payment system for inpatient medical rehabilitation.

Young, Jr., R.C., Rachal, R.E., Bailey, S.B., and others (1997). "Strategies for suppression, containment, and eradication of resurgent tuberculosis." (AHCPR grant HS07387). Journal of Health Care for the Poor and the Underserved 8(4), pp. 424-436.

The incidence of tuberculosis (TB) in this country surged 20 percent between 1985 and 1992 in association with the AIDS epidemic. Monodrug resistance to TB has been as high as 33 percent and multidrug-resistant TB (MDRTB) as high as 7 percent. In this study by researchers at the MEDTEP Minority Research Center at Meharry Medical College, strategies are described that could decrease and eventually eradicate the resurgent TB epidemic, which disproportionately affects minorities. These involve public health policy measures for communities and health care organizations and guidelines for health care providers. Most important is patient education about the need to cover coughs and sneezes, which carry Mycobacterium tuberculosis, and to take all prescribed medications to prevent TB disease, reinfection, or drug resistance. Education of health care workers to better screen, diagnose, and treat TB is also needed. Another step is to place TB patients on directly observed therapy (DOT), which has been shown to reduce the relapse rate and the frequency of both primary and acquired drug resistance. If DOT is not feasible, the researchers recommend monitoring patient compliance through interviews, pill counts, and urine tests.

Return to Contents

AHCPR Publication No. 98-0012
Current as of January 1998

The information on this page is archived and provided for reference purposes only.


AHRQ Advancing Excellence in Health Care