JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

When the results of a reference (or gold standard) test are missing or not error-free, the accuracy of diagnostic tests is often assessed through latent class models with two latent classes, representing diseased or nondiseased status. Such models, however, require that conditional on the true disease status, the tests are statistically independent, an assumption often violated in practice. Consequently, the model generally fits the data poorly. In this paper, we develop a general latent class model with random effects to model the conditional dependence among multiple diagnostic tests (or readers). We also develop a graphical method for checking whether or not the conditional dependence is of concern and for identifying the pattern of the correlation. Using the random-effects model and the graphical method, a simple adequate model that is easy to interpret can be obtained. The methods are illustrated with three examples from the biometric literature. The proposed methodology is also applicable when the true disease status is indeed known and conditional dependence could well be present.

A B S T R A C T Growth hormone (GH) release was studied in adults of normal stature, ages 21-86 yr. The subjects were 85-115% of ideal body weight, between the 5th and 95th percentiles in height, and free ofactive or progressive disease. 9 to 12 individuals in each decade from third to ninth were evaluated. The following criteria of GH status were measured: serum GH concentration, analyzed by radioimmunoassay at halfhour intervals for 4 h after onset of sleep, and at 1-h intervals from 8 a.m. to 4 p.m. in 52 subjects; daily retention of N, P, and K in response to 0.168 U human (h)GH/kg body wt314/day in 18 subjects; and plasma somatomedin C (SmC) level before and during exogenous hGH treatment in 18 subjects.All 10 individuals, 20-29 yr old, released substantial amounts of endogenous GH during both day and night (average peak serum GH obtained during day and night was 7.3 and 20.3 ng/ml, respectively); average plasma SmC was 1.43 U/ml (95% tolerance limits, 0.64-2.22 U/ ml). There was no significant effect of exogenous hGH on elemental balances or on plasma SmC. In contrast, 6 of 12 individuals 60-79 yr old showed the following evidences of impaired GH release: peak waking and sleeping seruim GH < 4 nglml; plasma SmC < 0.38 U/ ml; a significant retention in N, P, and K; and a significant rise in plasma SmC, in response to exogenous hGH.Plasma SmC, serum GH during sleep, serum GH during the day, retentions of N, P, and K in response to exogenous hGH, and rise in plasma SmC in response to hGH were all intercorrelated (P < 0.05). Plasma SmC < 0.38 U/ml corresponded to peak nocturnal serum GH < 4 ng/ml. The prevalence of plasma SmC < 0.38 U/ml

TC did not reduce the RR of falling in transitionally frail, older adults, but the direction of effect observed in this study, together with positive findings seen previously in more-robust older adults, suggests that TC may be clinically important and should be evaluated further in this high-risk population.

Endotoxin concentrations were measured in the portal, hepatic and peripheral venous blood of two groups of patients with cirrhosis using a limulus-based chromogenic assay. The high sensitivity of chromogenic detection allowed measurement of endotoxin as low as 10 to 15 pg per ml, an order of magnitude greater than previously possible by gelation studies. Group 1 consisted of 56 patients with cirrhosis undergoing angiographic evaluation. In this group, there was wide variability in hepatic venous concentration [73 +/- 110 pg per ml (mean +/- S.D.)] and peripheral venous concentration [31 +/- 58 pg per ml]. However, paired t test showed peripheral venous concentration was significantly (p less than 0.001) lower than hepatic venous concentration. Neither hepatic or peripheral venous endotoxin levels correlated significantly with a variety of clinical, biochemical or radiological parameters. Group 2 consisted of 21 patients with cirrhosis undergoing shunt surgery. Endotoxin levels again showed a wide range, with portal venous concentration (142 +/- 167 pg per ml) and simultaneous peripheral venous concentration (82 +/- 150 pg per ml). Paired t test in this group showed a significant (p less than 0.001) portal to peripheral venous gradient. This study showed the feasibility of measuring endotoxin in plasma to low concentrations by a chromogenic assay technique. It supports the concept of relatively high levels of endotoxin in the portal circulation. In the presence of liver disease, systemic endotoxemia occurs, which is augmented by stressful situations.