Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative.

AIM: To improve the accuracy and completeness of reporting of studies of diagnostic accuracy in order to allow readers to assess the potential for bias in a study and to evaluate the general isability of its results. METHODS: The standards for reporting of diagnostic accuracy (STARD) steering committee searched the literature to identify publications on the appropriate conduct and reporting of diagnostic studies and extracted potential items into an extensive list. Researchers, editors, and members of professional organisations shortened this list during a 2 day consensus meeting with the goal of developing a checklist and a generic flow diagram for studies of diagnostic accuracy. RESULTS: The search for published guidelines about diagnostic research yielded 33 previously published checklists, from which we extracted a list of 75 potential items. At the consensus meeting, participants shortened the list to a 25-item checklist, by using evidence whenever available. A prototype of a flow diagram provides information about the method of recruitment of patients, the order of test execution and the numbers of patients undergoing the test under evaluation, the reference standard, or both. CONCLUSIONS: Evaluation of research depends on complete and accurate reporting. If medical journals adopt the checklist and the flow diagram, the quality of reporting of studies of diagnostic accuracy should improve to the advantage of clinicians, researchers, reviewers, journals, and the public.

[Reporting studies of diagnostic accuracy according to a standard method; the Standards for Reporting of Diagnostic Accuracy (STARD)]. / Verslaglegging van diagnostisch evaluatieonderzoek volgens een standaardmethode; de 'standards for reporting of diagnostic accuracy' (STARD).

The objective of the 'Standards for Reporting of Diagnostic Accuracy' (STARD) initiative is to improve the reporting of studies of diagnostic accuracy, so as to allow readers to assess the potential for bias in a study and to evaluate the generalibility of its results. The group searched the literature to identify publications on the appropriate conduct and reporting of diagnostic studies. This was used to draw up a list of potential items. During a consensus meeting, a group of researchers, medical journal editors, and members of professional organisations reduced this list to a usable checklist. Wherever possible, evidence from the literature was used to justify the decisions made. The search for published guidelines about diagnostic research yielded 33 previously published checklists, from which a list of 75 potential items was extracted. At the consensus meeting, participants shortened the list to a 25-item checklist. A generic flow diagram was drawn up to provide guidance on the method for including patients, the order in which tests were to be conducted and the number of patients to undergo the test being evaluated, the reference standard, or both. A scientific publication can only be assessed when the reporting is both correct and complete. Use of the checklist and flow diagram will improve the quality of reports produced, to the advantage of clinicians, researchers, reviewers, journal editors and other interested parties.

A hierarchical regression approach to meta-analysis of diagnostic test accuracy evaluations.

An important quality of meta-analytic models for research synthesis is their ability to account for both within- and between-study variability. Currently available meta-analytic approaches for studies of diagnostic test accuracy work primarily within a fixed-effects framework. In this paper we describe a hierarchical regression model for meta-analysis of studies reporting estimates of test sensitivity and specificity. The model allows more between- and within-study variability than fixed-effect approaches, by allowing both test stringency and test accuracy to vary across studies. It is also possible to examine the effects of study specific covariates. Estimates are computed using Markov Chain Monte Carlo simulation with publicly available software (BUGS). This estimation method allows flexibility in the choice of summary statistics. We demonstrate the advantages of this modelling approach using a recently published meta-analysis comparing three tests used to detect nodal metastasis of cervical cancer.

Fine-needle aspiration biopsy of nonpalpable breast lesions in a multicenter clinical trial: results from the radiologic diagnostic oncology group V.

PURPOSE: To determine the diagnostic accuracy of ultrasonographically (US) and stereotactically guided fine-needle aspiration biopsy (FNAB) in the diagnosis of nonpalpable breast lesions. MATERIALS AND METHODS: At 18 institutions, 442 women who underwent 22-25-gauge imaging-guided FNAB were enrolled. Definitive surgical, core-needle biopsy, and/or follow-up information was available for 423 (95.7%) of these women. The reference standard was established from additional clinical and imaging information for an additional six (1.4%) women who did not undergo further histopathologic evaluation. The FNAB protocol was standardized at all institutions, and all specimens were reread by one of two expert cytopathologists. RESULTS: When insufficient samples were included in the analysis and classified as positive, the sensitivity and specificity of FNAB were 85%-88% and 55.6%-90.5%, respectively; accuracy ranged from 62.2% to 89.2%. The diagnostic accuracy of FNAB was significantly better for detection of masses than for detection of calcifications (67.3% vs. 53.8%, P =.006) and with US guidance than with stereotactic guidance (77.2% vs. 58.9%; P =.002). CONCLUSION: FNAB of nonpalpable breast lesions has limited value given the high insufficient sample rate and greater diagnostic accuracy of other interventions, including core-needle biopsy and needle-localized open surgical biopsy.

Rate of insufficient samples for fine-needle aspiration for nonpalpable breast lesions in a multicenter clinical trial: The Radiologic Diagnostic Oncology Group 5 Study. The RDOG5 investigators.

BACKGROUND: Radiologic Diagnostic Oncology Group 5 is a multicenter clinical trial designed to evaluate fine-needle aspiration (FNA) of nonpalpable breast lesions performed by multiple operators using the same protocol. METHODS: Four hundred and nineteen women with mammographically detected nonpalpable breast lesions were enrolled on the trial at 18 institutions. Group A institutions randomized women to stereotactically guided FNA (SFNA) followed by stereotactically guided core needle biopsy (SCNB), or SCNB only. Group B institutions randomized women to SFNA and SCNB, SCNB, or ultrasonographically guided FNA followed by ultrasonographically guided core needle biopsy (USCNB), or USCNB only. A total of 377 women were eligible for analysis. RESULTS: FNA yielded 128 insufficient samples for the 377 patients (33.95%; 95% confidence interval, 29.2-38.7%). The rate of insufficient samples varied by type of lesion with calcified lesions associated with a significantly higher rate of insufficient sampling than masses (P < 0.001). The radiologist's level of suspicion of the lesion was not a statistically significant predictor of insufficient samples for mass lesions, but was a predictor for calcified lesions. For the 336 lesions for which histologic information was available, insufficient samples occurred in significantly more benign than malignant lesions. CONCLUSIONS: The high rate of insufficient samples for FNA of nonpalpable breast lesions in this multicenter trial makes its use impractical in this setting. Because of this factor, the study was terminated early.

A simple method for comparing correlated ROC curves using incomplete data.

Comparative studies of the accuracy of diagnostic procedures often use a paired design to gain in efficiency. Standard methods for analysing data from paired designs require complete observations. In many studies, however, one of the test results may be missing for some patients. In this paper, we propose a simple correction to the existing complete data methods to compare areas under ROC curves derived from paired designs. The approach makes it possible to use the entire available data set in carrying out the comparison, provided that the probability of having both tests does not depend on the test results. As an illustration, we apply our method to the analysis of data from prospective comparison of MRI and ultrasound in detecting periprostatic invasion.

Variations in the utilization of coronary angiography for elderly patients with an acute myocardial infarction. An analysis using hierarchical logistic regression.

This article reports a study of variations in the utilization of angiography for Medicare recipients who had an acute myocardial infarction. The study cohort consisted of 1987 Medicare beneficiaries who had a recent acute myocardial infarction. Variations were examined from three perspectives: patient characteristics, regional practice patterns, and on-site availability of the procedure. Factors associated with variation within and among states were incorporated into the analysis using hierarchical logistic regression models. The probability of angiography during the first 90 days after an acute myocardial infarction was estimated as a function of patient age, gender, race, and comorbidity for patients in 51 states (including the District of Columbia). Interstate differences were examined in relation to geographic region and on-site availability of angiography. Observed rates of angiography ranged between 13.8% and 38.3% (median, 24.7%). Variation was nearly threefold based on estimated state probabilities of angiography for a patient with characteristics set at the national average. Observed and estimated rates were lower in northeastern states than in other parts of the United States. States with more extensive onsite availability of angiography tended to have higher angiography rates after adjusting for patient characteristics and geographic region. Adjusted angiography rates were on average higher for younger patients, males, and nonblacks. There was substantial interstate variation in race differences, with states in the Southeast generally having the largest differences. The adjusted black-to-nonblack odds ratio ranged from a low of 0.41 to a high of 0.94. Interstate variation in age and gender differences was moderate. The work reported in this article illustrates the potential of hierarchical regression modeling as a framework for the analysis of variations and some methodologic issues connected with its implementation. Our results show that large variations in the utilization of procedures can exist, despite uniform insurance coverage and a relatively homogeneous patient cohort. Aggressive use of angiography was highly variable across states as was the degree of access to the procedure for blacks and nonblacks. The state rate of on-site availability of angiography facilities was an important predictor of utilization. Increased on-site availability of angiography, however, was not associated with a reduction of differences in access to the procedure.

Changes in insurance status and access to care for persons with AIDS in the Boston Health Study.

The purpose of this study was to measure unmet needs and changes in insurance status for persons with acquired immunodeficiency syndrome (AIDS). Thirty-six percent of the study's Boston-area respondents (n = 305) had a change in insurance coverage between AIDS diagnosis and interview. Medicaid coverage increased from 14% to 41%. Pneumocystis carinii pneumonia prophylaxis was nearly universal. Only 5% did not receive zidovudine, and intravenous drug users were at higher risk. Approximately 14% to 15% of patients reported problems in obtaining medical and dental services; Blacks, homeless persons, and those who were not high school graduates were at higher risk. Use of selected treatments for which there were clear clinical guidelines was adequate, yet disadvantaged groups were more likely than other persons with AIDS to face obstacles to other services.

Racial differences in the use of revascularization procedures after coronary angiography.

OBJECTIVE: To assess whether rates of coronary revascularization procedures differ between blacks and whites after coronary angiography is performed and to assess the relationship of these rates to hospital characteristics. DESIGN: A retrospective cohort study using 1987 and 1988 data on hospital claims and characteristics from the Health Care Financing Administration. SETTING: One thousand four hundred twenty-nine acute care hospitals that provide coronary angiography in the United States. PATIENTS: A national sample of 27,485 Medicare Part A enrollees, aged 65 to 74 years, who underwent inpatient angiography for coronary heart disease in 1987. MAIN OUTCOME MEASURE: The adjusted odds of revascularization with either coronary angioplasty or bypass graft surgery within 90 days of angiography for whites relative to blacks, controlling for age, sex, region, Medicaid eligibility, principal diagnosis, comorbid diagnoses, and hospital characteristics of ownership, teaching status, urban/suburban or rural location, and availability of revascularization procedures. RESULTS: White men and women were significantly more likely than black men and women, respectively, to receive a revascularization procedure after coronary angiography (57% and 50% vs 40% and 34%, both P < .001). The adjusted odds of receiving a revascularization procedure after coronary angiography were 78% higher for whites than blacks (95% confidence interval for odds ratio, 1.56 to 2.03). Statistically significant racial differences in the adjusted odds of receiving a revascularization procedure were present in all types of hospitals except rural hospitals, and these differences did not vary significantly by any of the four hospital characteristics (all P > .20 for interaction terms). CONCLUSIONS: Among Medicare enrollees, whites are more likely than blacks to receive revascularization procedures after coronary angiography. Racial differences of similar magnitude occur in all types of hospitals. These differences may reflect overuse in whites or underuse in blacks, but they are unlikely to reflect access to cardiologists or hospitals that perform revascularization procedures. Potential explanations include unmeasured clinical or socioeconomic factors, differing patient preferences, and racial bias at the hospitals performing angiography.

Comparison of magnetic resonance imaging and ultrasonography in staging early prostate cancer. Results of a multi-institutional cooperative trial.

BACKGROUND: In 1987, a cooperative study group consisting of five institutions was formed to determine the relative benefits of magnetic resonance imaging (MRI) and endorectal (transrectal) ultrasonography in evaluating patients with clinically localized prostate cancer (stage Ta or Tb). METHODS: Over a period of 15 months, 230 patients were entered into the study and evaluated with identical imaging techniques. We compared imaging results with information obtained at the time of surgery and on pathological analysis. RESULTS: MRI correctly staged 77 percent of cases of advanced disease and 57 percent of cases of localized disease; the corresponding figures for ultrasonography were 66 and 46 percent (P not significant). These figures did not vary significantly between readers; moreover, simultaneous interpretation of MRI and ultrasound scans did not improve accuracy. In terms of detecting and localizing lesions, MRI identified only 60 percent of all malignant tumors measuring more than 5 mm on pathological analysis and ultrasonography identified only 59 percent. CONCLUSIONS: The MRI and ultrasonography equipment that is currently available is not highly accurate in staging early prostate cancer, mainly because neither technique has the ability to identify microscopic spread of disease. Further evaluation with improved equipment may improve the accuracy of these techniques.

Low-level lead exposure and the IQ of children. A meta-analysis of modern studies.

We identified 24 modern studies of childhood exposures to lead in relation to IQ. From this population, 12 that employed multiple regression analysis with IQ as the dependent variable and lead as the main effect and that controlled for nonlead covariates were selected for a quantitative, integrated review or meta-analysis. The studies were grouped according to type of tissue analyzed for lead. There were 7 blood and 5 tooth lead studies. Within each group, we obtained joint P values by two different methods and average effect sizes as measured by the partial correlation coefficients. We also investigated the sensitivity of the results to any single study. The sample sizes ranged from 75 to 724. The sign of the regression coefficient for lead was negative in 11 of 12 studies. The negative partial r's for lead ranged from -.27 to -.003. The power to find an effect was limited, below 0.6 in 7 of 12 studies. The joint P values for the blood lead studies were less than .0001 for both methods of analysis (95% confidence interval for group partial r, -.15 +/- .05), while for the tooth lead studies they were .0005 and .004, respectively (95% confidence interval for group partial r, -.08 +/- .05). The hypothesis that lead impairs children's IQ at low dose is strongly supported by this quantitative review. The effect is robust to the impact of any single study.