We consider how to combine several independent studies of the same diagnostic test, where each study reports an estimated false positive rate (FPR) and an estimated true positive rate (TPR). We propose constructing a summary receiver operating characteristic (ROC) curve by the following steps. (i) Convert each FPR to its logistic transform U and each TPR to its logistic transform V after increasing each observed frequency by adding 1/2. (ii) For each study calculate D = V - U, which is the log odds ratio of TPR and FPR, and S = V + U, an implied function of test threshold; then plot each study's point (Si, Di). (iii) Fit a robust-resistant regression line to these points (or an equally weighted least-squares regression line), with V - U as the dependent variable. (iv) Back-transform the line to ROC space. To avoid model-dependent extrapolation from irrelevant regions of ROC space we propose defining a priori a value of FPR so large that the test simply would not be used at that FPR, and a value of TPR so low that the test would not be used at that TPR. Then (a) only data points lying in the thus defined north-west rectangle of the unit square are used in the data analysis, and (b) the estimated summary ROC is depicted only within that subregion of the unit square. We illustrate the methods using simulated and real data sets, and we point to ways of comparing different tests and of taking into account the effects of covariates.

BackgroundReading skills are important for accessing health information, using health care services, managing one's health and achieving desirable health outcomes. Our objective was to assess the diagnostic accuracy of the Single Item Literacy Screener (SILS) to identify limited reading ability, one component of health literacy, as measured by the S-TOFHLA.MethodsCross-sectional interview with 999 adults with diabetes residing in Vermont and bordering states. Participants were randomly recruited from Primary Care practices in the Vermont Diabetes Information System June 2003 – December 2004. The main outcome was limited reading ability. The primary predictor was the SILS.ResultsOf the 999 persons screened, 169 (17%) had limited reading ability. The sensitivity of the SILS in detecting limited reading ability was 54% [95% CI: 47%, 61%] and the specificity was 83% [95% CI: 81%, 86%] with an area under the Receiver Operating Characteristics Curve (ROC) of 0.73 [95% CI: 0.69, 0.78]. Seven hundred seventy (77%) screened negative on the SILS and 692 of these subjects had adequate reading skills (negative predictive value = 0.90 [95% CI: 0.88, 0.92]). Of the 229 who scored positive on the SILS, 92 had limited reading ability (positive predictive value = 0.4 [95% CI: 0.34, 0.47]).ConclusionThe SILS is a simple instrument designed to identify patients with limited reading ability who need help reading health-related materials. The SILS performs moderately well at ruling out limited reading ability in adults and allows providers to target additional assessment of health literacy skills to those most in need. Further study of the use of the SILS in clinical settings and with more diverse populations is warranted.

No abstract

No abstract

Reports of diagnostic accuracy often differ. The authors present a method to summarize disparate reports that uses a logistic transformation and linear regression to produce a summary receiver operating characteristic curve. The curve is useful for summarizing a body of diagnostic accuracy literature, comparing technologies, detecting outliers, and finding the optimum operating point of the test. Examples from clinical chemistry and diagnostic radiology are provided. By extending the logic of meta-analysis to diagnostic testing, the method provides a new tool for technology assessment.

No abstract

Because published outcome studies are the only available source of data about the morbidity among surviving very low birthweight infants (VLBW, <1500 g) a search was carried out of 1136 references in the English language. A total of 111 outcome studies were found that reported morbidity data in cohorts of VLBW infants born since 1960. The methods used and results obtained in these studies were systematically assessed. No agreement exists about the definition of study populations, descriptive statistics, or measurement of outcome. Foliow up ranged from six months to 14 years. In 85 cohorts the incidence of cerebral palsy was recorded, and in 106 that of disability. Studies that followed up infants for longer time periods reported higher incidences of disability. Studies from the United States reported higher incidences of disability than those from other countries. The median incidence of cerebral palsy among all the cohorts studied was 7-7%, and the median incidence of disability was 25-0%. Despite substantial improvements in the mortality of VLBW infants, poor outcomes among survivors are common.

We conducted a meta-analysis of 22 randomized, controlled trials in which extended-interval dosing of aminoglycosides was compared with multiple daily dosing. When we classified intermediate outcomes as successes, we found that patients receiving extended-interval dosing were at significantly reduced risk of clinical treatment failure (risk difference, -3.4%; 95% confidence interval [CI], -6.7% to -0.2%; P = .039) and that there was a trend toward reduced risk of bacteriologic failure (risk difference, -1.7%; 95% CI, -5.4% to +2.1%; P = .38). Reclassification of intermediate outcomes as failures yielded similar results. There was significant heterogeneity among the trials, necessitating cautious interpretation of these outcomes. There were negligible differences in the risk of nephrotoxicity (risk difference, -0.6%; 95% CI, -2.4% to +1.1%; P = .46) and ototoxicity (risk difference, +0.3%; 95% CI, -1.2% to +1.8%; P = .71). We conclude that for many indications, extended-interval dosing of aminoglycosides appears to be as effective as conventional dosing, with similar rates of toxicity. The added convenience of extended-interval dosing makes it an attractive alternative to conventional dosing.