Selected Publications by Grantees, 2002

Center for Outcomes and Effectiveness Research


Pelvic Inflammatory Disease

Title: Douching and endometritis: results from the PID evaluation and clinical health (PEACH) study.
Publication: Sexually Transmitted Diseases 2001 Apr;28(4):240-5.
Authors: Ness RB et al.
Grant Number: HS08358.
Principal Investigator: Roberta Ness.
Project Officer: Heddy Hubbard.
Abstract: BACKGROUND: Douching has been related to risk of pelvic inflammatory disease (PID). GOAL: To examine the association between douching and PID in a large, multicenter, clinical trial of PID after adjustment for race/ethnicity. STUDY DESIGN: Interviews were conducted with 654 women who had signs and symptoms of PID. Vaginal Gram stains and upper genital tract pathology/cultures were obtained from all the women. Women with evidence of plasma cell endometritis and/or gonococcal or chlamydial upper genital tract infections were compared with women who had neither endometritis nor upper genital tract infection. RESULTS: Women with endometritis or upper genital tract infection were more likely to have douched more than once a month or within 6 days of enrollment than women who never douched. These associations remained after adjustment for confounding factors, after analysis of black women only; and among women with normal or intermediate vaginal flora but not bacterial vaginosis. CONCLUSION: Among a predominantly black group of women with clinical PID, frequent and recent douching was associated with endometritis and upper genital tract infection.

Title: Clinical predictors of endometritis in women with symptoms and signs of pelvic inflammatory disease.
Publication: American Journal of Obstetrics and Gynecology 2001;184:856-64.
Authors: Peipert JF et al.
Grant Number: HS08358.
Principal Investigator: Roberta Ness.
Project Officer: Heddy Hubbard.
Abstract: OBJECTIVE: Careful detection and treatment of pelvic inflammatory disease are essential for the prevention of adverse sequelae. The purpose of this study was to evaluate the diagnostic test characteristics of clinical criteria for the diagnosis of pelvic inflammatory disease. STUDY DESIGN: We performed a cross-sectional analysis of the baseline characteristics of 651 patients enrolled in a multicenter randomized treatment trial for pelvic inflammatory disease. Clinical and laboratory findings were recorded for all patients, and endometrial sampling was performed. We calculated sensitivity and specificity and performed receiver operating characteristic curve analysis and multivariate logistic regression, using histologic endometritis as the criterion standard. RESULTS: The minimal criteria for pelvic inflammatory disease, as recommended by the Centers for Disease Control and Prevention, had a sensitivity of 83 percent, in comparison with a 95 percent sensitivity for adnexal tenderness (P =.001). Of the supportive clinical criteria, the finding most highly associated with endometritis was a positive test result for Chlamydia trachomatis or Neisseria gonorrhoeae (adjusted odds ratio, 4.3; 95 percent confidence interval, 2.89--6.63). A multivariate logistic regression model indicated that combinations of criteria significantly improve the prediction of endometritis. CONCLUSION: Sensitivity can be maximized by using the presence of adnexal tenderness as a minimal criterion for the diagnosis of pelvic inflammatory disease, and supportive criteria are helpful in estimating the probability of endometritis.

Title: Hormonal and barrier contraception and risk of upper genital tract disease in the PID Evaluation and Clinical Health (PEACH) study.
Publication: American Journal of Obstetrics and Gynecology 2001 Jul;185(1):121-7.
Grant Number: HS08358.
Principal Investigator: Roberta Ness.
Project Officer: Heddy Hubbard.
Abstract: OBJECTIVE: Among women diagnosed with pelvic inflammatory disease, we examined the associations between hormonal or barrier methods of contraception and upper genital tract infection or inflammation. METHODS: Participants were 563 patients from a treatment trial for pelvic inflammatory disease. All had pelvic pain; pelvic organ tenderness; and leukorrhea, mucopurulent cervicitis, or untreated cervicitis. Contraceptive use within the prior 4 weeks was compared among women with baseline upper genital tract gonorrhea or chlamydia, women with endometritis without upper genital tract gonorrhea or chlamydia, and women with neither upper genital tract gonorrhea or chlamydia nor endometritis. RESULTS: Inconsistent condom use was significantly and independently associated with a two to three times elevated risk for upper genital tract infection. Upper genital tract gonorrhea or chlamydia was not significantly associated with use of oral contraceptives, use of medroxyprogesterone, condoms used consistently, nor other barrier methods. CONCLUSION: No hormonal or barrier contraceptive method was related to a reduction in upper genital tract disease among women with clinical pelvic inflammatory diseases.

Return to Topic Index


Title: What factors influence physicians' decisions to switch from intravenous to oral antibiotics for community-acquired pneumonia?
Publication: Journal of General Internal Medicine 2001 Sep;16(9): 599-605.
Authors: Halm EA et al.
Grant Number: HS08282.
Principal Investigator: Michael Fine.
Project Officer: Joanne Book.
Abstract: OBJECTIVE: One of the major factors influencing length of stay for patients with community-acquired pneumonia is the timing of conversion from intravenous to oral antibiotics. We measured physician attitudes and beliefs about the antibiotic switch decision and assessed physician characteristics associated with practice beliefs. DESIGN: Written survey assessing attitudes about the antibiotic conversion decision. SETTING: Seven teaching and non-teaching hospitals in Pittsburgh, Pa. PARTICIPANTS: Three hundred forty-five generalist and specialist attending physicians who manage pneumonia in 7 hospitals. MEASUREMENTS AND RESULTS: Factors rated as "very important" to the antibiotic conversion decision were: absence of suppurative infection (93percent), ability to maintain oral intake (79percent), respiratory rate at baseline (64percent), no positive blood cultures (63percent), normal temperature (62percent), oxygenation at baseline (55percent), and mental status at baseline (50percent). The median thresholds at which physicians believed a typical patient could be converted to oral therapy were: temperature < or =100 degrees F (37.8 degrees C), respiratory rate < or =20 breaths/minute, heart rate < or =100 beats/minute, systolic blood pressure > or =100 mm Hg, and room air oxygen saturation > or =90percent. Fifty-eight percent of physicians felt that "patients should be afebrile for 24 hours before conversion to oral antibiotics," and 19percent said, "patients should receive a standard duration of intravenous antibiotics." In univariate analyses, pulmonary and infectious diseases physicians were the most predisposed towards early conversion to oral antibiotics, and other medical specialists were the least predisposed, with generalists being intermediate (P <.019). In multivariate analyses, practice beliefs were associated with age, inpatient care activities, attitudes about guidelines, and agreeableness on a personality inventory scale. CONCLUSIONS: Physicians believed that patients could be switched to oral antibiotics once vital signs and mental status had stabilized and oral intake was possible. However, there was considerable variation in several antibiotic practice beliefs. Guidelines and pathways to streamline antibiotic therapy should include educational strategies to address some of these differences in attitudes.

Title: Arterial blood gas and pulse oximetry in initial management of patients with community-acquired pneumonia.
Publication: Journal of General Internal Medicine 2001 Sep;16(9):590-8.
Authors: Levin KP et al.
Grant Number: HS06468.
Principal Investigator: Wishwa Kapoor.
Project Officer: Lynn Bosco.
Abstract: OBJECTIVE: To identify the factors associated with the use of arterial blood gas (ABG) and pulse oximetry (PO) in the initial management of patients with community-acquired pneumonia (CAP) and arterial hypoxemia at presentation. PARTICIPANTS: A total of 944 outpatients and 1,332 inpatients with clinical and radiographic evidence of CAP prospectively enrolled from 5 study sites in the United States and Canada. ANALYSES: Separate multivariate logistic regression analyses were used to 1) compare measurement of ABG and PO within 48 hours of presentation across sites while controlling for patient differences, and 2) identify factors associated with arterial hypoxemia (PaO2 <60 mm Hg or SaO2 <90percent for non-African Americans and <92percent for African Americans) while breathing room air. RESULTS: Range of ABG use by site was from 0percent to 6.4percent (P =.06) for outpatients and from 49.2percent to 77.3percent for inpatients (P<.001), while PO use ranged from 9.4percent to 57.8percent for outpatients (P <.001) and from 47.9percent to 85.1percent for inpatients (P <.001). Differences among sites remained after controlling for patient demographic characteristics, comorbidity, and illness severity. In patients with 1 or more measurements of oxygenation at presentation, hypoxemia was independently associated with 6 risk factors: age >30 years (odds ratio [OR], 3.2; 95percent confidence interval [CI], 1.7 to 5.9), chronic obstructive pulmonary disease (OR, 1.9; 95percent CI, 1.4 to 2.6), congestive heart failure (OR, 1.5; 95percent CI, 1.0 to 2.1), respiratory rate >24 per minute (OR, 2.3; 95percent CI, 1.8 to 3.0), altered mental status (OR, 1.6; 95percent CI, 1.1 to 2.3), and chest radiographic infiltrate involving >1 lobe (OR, 2.2; 95percent CI, 1.7 to 2.9). The prevalence of hypoxemia among those tested ranged from 13percent for inpatients with no risk factors to 54.6percent for inpatients with > or =3 risk factors. Of the 210 outpatients who had > or =2 of these risk factors, only 64 (30.5percent) had either an ABG or PO performed. In the 48 outpatients tested without supplemental O2 with > or =2 risk factors 8.3percent were hypoxemic. CONCLUSIONS: In the initial management of CAP, use of ABG and PO varied widely across sites. Increasing the assessment of arterial oxygenation among patients with CAP is likely to increase the detection of arterial hypoxemia, particularly among outpatients.

Return to Topic Index


Title: Treatments for prostate cancer in older men: 1984-1997.
Publication: Urology 2001 Dec;58(6):977-82.
Authors: Bubolz T et al.
Grant Number: HS08397.
Principal Investigator: Michael Barry.
Project Officer: Steven Fox.
Abstract: OBJECTIVES: To examine the temporal trends in radical prostatectomy (RP),brachytherapy (BT), and external beam radiotherapy (EBRT) rates among men aged 65 years or older for the period 1984 to 1997. METHODS: We used the retrospective population-based analysis of treatments for prostate cancer among Medicare beneficiaries. The rates of RP were obtained from Part A (hospital) Medicare data for 20percent of the national sample for 1984 to 1997. The BT and EBRT rates for the period 1993 to 1997 were obtained from a 5percent national sample of Physician/Supplier Part B data. The rates of treatment, 30-day mortality, and readmissions were included. RESULTS: The rate of RP peaked in 1992. From 1993 to 1997, its use decreased by 6percent among men aged 65 to 69 years, 34percent among men aged 70 to 74 years, and 50percent for men aged 75 years or older. However, by 1997, the RP + BT treatment rate again approached the 1992 levels of RP alone; BT was used twice as often as RP in men aged 75 years or older. By1997, the RP + BT + EBRT rate exceeded the 1993 rate for men aged 65 to 69 years and was again approaching the 1993 rate for men aged 70 to 74 years. From 1984 to 1997, the presence of comorbid conditions gradually declined for RP and accounted for more than 60percent of the decrease in the short term mortality during this period. Variations in RP use by geographic region have also decreased. CONCLUSIONS: RP is now more selectively targeted for treatment of prostate cancer in men older than 70 years than in the past. However, since BT has been substituted for radical surgery in many of these older men, the total population-based treatment rates have changed very little over time.

Title: A frequently used patient and physician-directed educational intervention does nothing to improve primary care of prostate conditions.
Publication: Urology 2001 Dec;58(6):875-81.
Authors: Hammond CS et al.
Grant Number: HS08397.
Principal Investigator: Michael Barry.
Project Officer: Steven Fox.
Abstract: OBJECTIVES: To measure the impact of an educational intervention directed at both patients and their primary care physicians about prostate-related conditions. METHODS: We used a randomized, control design for 50 physicians in 33 rural primary care practices from New England and Arkansas and a probability sample of 2402 of their male patients. For the physicians, we mailed two newsletters, conducted two face-to-face research staff visits, and provided printed educational manuals about the management of prostate conditions. For the patients, mailed educational pamphlets were targeted to the baseline symptom levels. After 18 months, 87percent of patients and 92percent of physicians completed a final survey. The patient survey measured health status, urinary symptoms and bother, treatments received, and prostate-related knowledge. The final physician survey asked them about their management of common prostate conditions. RESULTS: Before randomization, most men (59percent) said they knew little or nothing about prostate problems that affect urination, and 63percent also reported "little" or "no" knowledge about prostate-specific antigen testing. Eighteen months later, we observed no differences between the intervention and control patients in the measures of health status, urinary symptoms and bother, treatments received, and prostate-related knowledge. The intervention, physicians' knowledge, and self-reported practices for managing common prostate conditions were no better than the control physicians'. CONCLUSIONS: This commonly used education strategy had no measurable impact on prostate-related care.

Title: Determining cause of death in prostate cancer: are death certificates valid?
Publication: J Natl Cancer Inst 2001 Dec 5;93(23): 1822-3.
Authors: Penson DF et al.
Grant Number: HS08397.
Principal Investigator: Michael Barry.
Project Officer: Steven Fox.
Abstract: It is worth noting that, in our study, in the cases in which there was disagreement between clinician assessment and death certificate assignment of COD, there was always clinically significant respiratory and cardiac comorbidity, which may have affected the coding algorithm for assigning COD using death certificates. This observation is consistent with the findings of Satariano et al. (13), who noted that prostate cancer patients with concurrent cardiovascular disease were less likely to be coded as having died of prostate cancer than if they had other comorbidities. This observation underscores the need for health care providers to carefully document all relevant details when completing death certificates, particularly in men with multiple comorbidities (14). Our study is limited, in that only 32percent of all men who died in 1995 of prostate cancer in King County did so in a hospital. However, because medical records pertaining to death are only available in the inpatient setting, the validity of underlying COD assigned by death certificate to inpatients who die at home or in a nursing home remains unclear. This should be an area of further research. In summary, this study indicates that there is an excellent agreement between the underlying COD from death certificates and medical records in prostate cancer patients. Researchers and clinicians alike should be comfortable accepting the validity of disease-specific mortality rates derived from datasets that use death certificate data to evaluate these outcomes.

Title: Risk factors for clinical benign prostatic hyperplasia in a community-based population of healthy aging men.
Publication: Journal of Clinical Epidemiology 2001 Sep;54(9):935-44.
Authors: Meigs JB et al.
Grant Number: HS08397.
Principal Investigator: Michael Barry.
Project Officer: Steven Fox.
Abstract: We defined risk factors for a clinical diagnosis of benign prostatic hyperplasia (BPH) among subjects of the population-based Massachusetts Male Aging Study. In 1987-89, 1,709 men aged 40-70 provided baseline risk factor data and were followed for a mean of 9 years; 1,019 men without prostate cancer provided followup data. We classified men with clinical BPH at followup if they reported (1) frequent or difficulty urinating and were told by a health professional that they had an enlarged or swollen prostate or (2) if they reported having surgery for BPH. At followup the prevalence of clinical BPH was 19.4 percent, increasing from 8.4 percent of men aged 38-49 years to 33.5 percent of men aged 60-70 years (P < 0.001 for trend). Elevated free PSA levels (age-and total PSA-adjusted OR, top vs. bottom quartile ng/mL 4.4, 95 percent CI 1.9-10.5), heart disease (age-adjusted OR 2.1, CI 1.3-3.3), and use of beta-blocker medications (OR 1.8, CI 1.1-3.0) increased odds for BPH, while current cigarette smoking (OR 0.5, CI 0.3-0.8) and high levels of physical activity (top vs. bottom quartile kcals/day OR 0.5, CI 0.3-0.9) decreased odds of BPH. All but the medication effects persisted in fully adjusted multivariable models. Total or fat calorie intake, sexual activity level, alcohol intake, body mass index, waist-hip ratio, diastolic blood pressure, a history of diabetes, hypertension, vasectomy, or serum levels of androgens or estrogens did not individually predict clinical BPH. We conclude that physical exercise and cigarette smoking appear to protect against development of clinical BPH. Elevated free PSA levels predict clinical BPH independent of total PSA levels. Risk associated with heart disease does not appear to be due solely to detection bias or to effects of heart disease medications. A wide variety of other characteristics appear to have no influence on risk for clinical BPH.

Title: Symptom indexes to assess outcomes of treatment for early prostate cancer.
Publication: Med Care 2001 Oct;39(10):1118-30.
Authors: Clark JA, Talcott JA.
Grant Number: HS6824.
Principal Investigator: James Talcott.
Project Officer: Mary Cummings.
Abstract: BACKGROUND: Patients' perceptions of treatment outcomes are important in the management of early prostate cancer, but few studies have offered reliable and responsive measures to assess the likely side effects of the most common treatments. OBJECTIVE: To develop indexes of urinary, bowel, and sexual function, and related distress. RESEARCH DESIGN: Prospective cohort study of the outcomes of treatment for early prostate cancer, with self-administered questionnaires completed before treatment, and 3 and 12 months afterward. Hypothesized indexes, based on a clinical model of pathophysiological side effects of treatment, were defined and evaluated with respect to reliability and validity. SUBJECTS: Patients (n = 184) undergoing radical prostatectomy or external beam radiotherapy for early prostate cancer. MEASURES: Urinary and bowel items pertained to frequency or intensity of symptoms of dysfunction; parallel items assessed symptom-related distress. Sexual dysfunction items assessed the quality of erections, orgasm, and ejaculation; distress was assessed by 2 items adapted from the MOS Sexual Problems (MOS-SP) scale. HRQoL was assessed by the SF-36 and Profile of Mood States. RESULTS: Symptom and symptom-related distress indexes for urinary incontinence, urinary obstruction/irritation, bowel dysfunction, and sexual dysfunction were defined. Symptom and distress indexes in each domain were highly correlated. Responsiveness was substantial and varied by treatment in ways consistent with clinical experience. The indexes accounted for significant proportions of the variance in HRQoL measures. CONCLUSIONS: These indexes may be used in monitoring outcomes of treatment for early prostate cancer.

Title: Outcomes for men with clinically nonmetastatic prostate carcinoma managed with radical prostactectomy, external beam radiotherapy, or expectant management: a retrospective analysis.
Publication: Cancer 2001 Jun 15;91(12):2302-14.
Authors: Barry MJ et al.
Grant Number:HS08397.
Principal Investigator: Michael Barry.
Project Officer: Steven Fox.
Abstract: BACKGROUND: With a lack of data from randomized trials, the optimal management of men with nonmetastatic prostate carcinoma is controversial. The authors sought to define the outcomes of three common strategies for managing patients with nonmetastatic prostate carcinoma: expectant management, radiotherapy, and radical prostatectomy. METHODS: The authors conducted a retrospective cohort study with standardized collection of key prognostic data, including centralized assignment of Gleason grades from original biopsy specimens. Participants included all Connecticut hospitals (the expectant management cohort) and three academic medical centers in other states (the radiotherapy and surgery cohorts). Two thousand three hundred eleven consecutive men ages 55-74 years who were diagnosed during 1971-84 with nonmetastatic prostate carcinoma and were treated at the participating sites were included. RESULTS: Kaplan-Meier estimates with 95 percent confidence intervals (95 percent CI) of overall survival at 10 years for each cohort were as follows: expectant management cohort, 42 percent of patients (95 percent CI, 38-46 percent); radiotherapy cohort, 52 percent of patients (95 percent CI, 46-58 percent); and radical prostatectomy cohort, 69 percent of patients (95 percent CI, 67-71 percent); for disease specific mortality, the estimates were as follows: expectant management cohort, 75 percent of patients (95 percent CI, 71-79 percent); radiotherapy cohort, 67 percent of patients (95 percent CI, 61-73 percent); and radical prostatectomy cohort, 86 percent of patients (95 percent CI, 84-88 percent). There were large differences in distributions of important prognostic factors among men in the different treatment groups. CONCLUSIONS: These data provide precise estimates of the outcomes of patients who have been treated with different modalities for nonmetastatic prostate carcinoma in the recent past. Direct comparisons of outcomes between treatment groups are inadvisable because of the different characteristics of patients who select these alternative management strategies.

Title: Prostate-specific-antigen testing for early diagnosis of prostate cancer.
Publication: New England Journal of Medicine 2001 May;344(18):1373-7.
Authors: Barry MJ.
Grant Number: HS08397.
Principal Investigator: Michael Barry.
Project Officer: Steven Fox.
Abstract: The PSA test detects prostate cancer at an early stage in many cases. At present, data are not yet available from large, well-designed, randomized trials to determine whether early detection is beneficial or harmful or has not effects. As a result, the optimal strategy for early detection with PSA testing remains unknown. Decision analyses suggest that given certain assumptions about its effectiveness, PSA screening could be cost effective, at least for younger men. On the basis of available data, men who are approximately 50 to75 years of age (depending on the presence of risk factors at the lower age limit and the general state of health a the upper age limit) should be made aware of the availability of the PSA test and its potential harms and benefits, so that they can make an informed choice about screening. A discussion about testing should include the following points: the likelihood that prostate cancer will be diagnosed, the possibilities of false positive and false negative results, the anxiety associated with a positive test, and the uncertainty regarding whether screening reduces the risk of death from prostate cancer. In a recent study, these points were among those that men and their wives thought all men should know before undergoing a PSA test. Randomized trials have indicated that routinely providing such information reduces the proportion of men who decide to be tested, although substantial proportions of men still elect to do so. Clinicians should not be dismayed by either choice.

Return to Topic Index


Title: Variables Associated with Disparities in Treatment of Patients with Schizophrenia and Comorbid Mood and Anxiety Disorders.
Publication: Psychiatr Serv 2001 Sep;52(9):1216-22.
Authors: Dixon L et at.
Grant Number: 292-92-0054.
Principal Investigator: Anthony Lehman.
Project Officer: Charlotte Mullican.
Abstract: OBJECTIVE: The study compared self-reported comorbid affective and anxiety disorder diagnoses and treatments of African-American and Caucasian subjects in a large sample of patients who had a diagnosis of schizophrenia. METHODS: A total of 685 patients receiving treatment for schizophrenia were interviewed as part of the Schizophrenia Patient Outcomes Research Team study. The associations of race with past and current diagnoses and with current treatment for depression, mania, and anxiety disorders were assessed with multivariate analyses. RESULTS: African Americans were significantly less likely than Caucasians to report having a past or current diagnosis of depression, manic-depression, or anxiety disorder and to be receiving current treatment for these disorders. Gender, education, and marital status were also associated with presence of a comorbid diagnosis and receipt of treatment. CONCLUSIONS: The study suggests the possibility of racial and other disparities in the diagnosis and treatment of patients with schizophrenia and comorbid affective and anxiety disorders. Although various causal explanations are plausible, all point toward the need for enhanced cross-cultural competence at all levels of mental health care, especially in the diagnosis and treatment of comorbid psychiatric illnesses.

Title: Use and costs of ambulatory care services among medicare enrollees with schizophrenia.
Publication: Psychiatric Services 2001; 52:786-92.
Authors: Dixon L et al.
Grant Number: 290-92-0054.
Principal Investigator: Anthony Lehman.
Project Officer: Charlotte Mullican.
Abstract: OBJECTIVE: The objective of this study was to identify predictors of the use and cost of ambulatory care services among Medicare recipients with schizophrenia. METHODS: The design was a cross-sectional analysis of Medicare claims in 1991. The study subjects were a 5 percent random sample of all persons in the United States who had at least one Medicare service claim in 1991 and who were diagnosed as having schizophrenia in any care setting. Outcome measures included use and cost of any ambulatory care service, individual therapy, psychiatric somatotherapy, group therapy, or family therapy. RESULTS: For nearly 25 percent of the total sample of 12,440, no claims were filed for ambulatory care services in 1991. The mean+/-SD number of ambulatory care visits during the year was 7.9+/-21. The most frequently used type of therapy was individual therapy (5+/-14 visits). The mean+/-SD yearly cost of care for persons who received ambulatory care services was 470+/-$1,028. Among persons under 65 years of age, Caucasians were about 1.5 times as likely as African Americans to have received an ambulatory care service and 1.3 times as likely to have received individual therapy. Persons who were 65 or older were less likely to have received any service. Among service recipients, costs of care were lower for African Americans and for older people. CONCLUSIONS: The use of Medicare-funded ambulatory care services by persons with schizophrenia varied by race and age. Further investigation is required to determine whether subgroups of individuals who do not have additional insurance coverage or access to services are receiving substandard care.

Return to Topic Index


Title: Improving the reliability of stroke subgroup classification using the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) criteria.
Publication: Stroke 2001 May;32(5):1091-8.
Authors: Goldstein LB et al.
Grant Number: 290-91-0028.
Principal Investigator: Yen-pin Chiang.
Project Officer: David Matchar.
Abstract: BACKGROUND AND PURPOSE: We sought to improve the reliability of the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification of stroke subtype for retrospective use in clinical, health services, and quality of care outcome studies. The TOAST investigators devised a series of 11 definitions to classify patients with ischemic stroke into 5 major etiologic/pathophysiological groupings. Interrater agreement was reported to be substantial in a series of patients who were independently assessed by pairs of physicians. However, the investigators cautioned that disagreements in subtype assignment remain despite the use of these explicit criteria and that trials should include measures to ensure the most uniform diagnosis possible. METHODS: In preparation for a study of outcomes and management practices for patients with ischemic stroke within Department of Veterans Affairs hospitals, 2 neurologists and 2 internists first retrospectively classified a series of 14 randomly selected stroke patients on the basis of the TOAST definitions to provide a baseline assessment of interrater agreement. A 2-phase process was then used to improve the reliability of subtype assignment. In the first phase, a computerized algorithm was developed to assign the TOAST diagnostic category. The reliability of the computerized algorithm was tested with a series of synthetic cases designed to provide data fitting each of the 11 definitions. In the second phase, critical disagreements in the data abstraction process were identified and remaining variability was reduced by the development of standardized procedures for retrieving relevant information from the medical record. RESULTS: The 4 physicians agreed in subtype diagnosis for only 2 of the 14 baseline cases (14percent) using all 11 TOAST definitions and for 4 of the 14 cases (29percent) when the classifications were collapsed into the 5 major etiologic/pathophysiological groupings (kappa=0.42; 95percent CI, 0.32 to 0.53). There was 100percent agreement between classifications generated by the computerized algorithm and the intended diagnostic groups for the 11 synthetic cases. The algorithm was then applied to the original 14 cases, and the diagnostic categorization was compared with each of the 4 physicians' baseline assignments. For the 5 collapsed subtypes, the algorithm-based and physician-assigned diagnoses disagreed for 29percent to 50percent of the cases, reflecting variation in the abstracted data and/or its interpretation. The use of an operations manual designed to guide data abstraction improved the reliability subtype assignment (kappa=0.54; 95percent CI, 0.26 to 0.82). Critical disagreements in the abstracted data were identified, and the manual was revised accordingly. Reliability with the use of the 5 collapsed groupings then improved for both interrater (kappa=0.68; 95percent CI, 0.44 to 0.91)and intrarater (kappa=0.74; 95percent CI, 0.61 to 0.87) agreement. Examining each remaining disagreement revealed that half were due to ambiguities in the medical record and half were related to otherwise unexplained errors in data abstraction. CONCLUSIONS: Ischemic stroke subtype based on published TOAST classification criteria can be reliably assigned with the use of a computerized algorithm with data obtained through standardized medical record abstraction procedures. Some variability in stroke subtype classification will remain because of inconsistencies in the medical record and errors in data abstraction. This residual variability can be addressed by having 2 raters classify each case and then identifying and resolving the reason(s) for the disagreement.

Return to Topic Index

Current as of January 2002

Internet Citation:

Selected Publications by Grantees, 2002. Center for Effectiveness and Outcomes Research. January 2002. Agency for Healthcare Research and Quality, Rockville, MD.

Return to COER
About AHRQ
AHRQ Home Page
Department of Health and Human Services