Structural equation model testing and the quality of natural killer cell activity measurements.

BACKGROUND: Browne et al. [Browne, MacCallum, Kim, Andersen, Glaser: When fit indices and residuals are incompatible. Psychol Methods 2002] employed a structural equation model of measurements of target cell lysing by natural killer cells as an example purportedly demonstrating that small but statistically significant ill model fit can be dismissed as "negligible from a practical point of view". METHODS: Reanalysis of the natural killer cell data reveals that the supposedly negligible ill fit obscured important, systematic, and substantial causal misspecifications. RESULTS: A clean-fitting structural equation model indicates that measurements employing higher natural-killer-cell to target-cell ratios are more strongly influenced by a progressively intrusive factor, whether or not the natural killer cell activity is activated by recombinant interferon gamma (rIFN gamma). The progressive influence may reflect independent rate limiting steps in cell recognition and attachment, spatial competition for cell attachment points, or the simultaneous lysings of single target cells by multiple natural killer cells. CONCLUSIONS: If the progressively influential factor is ultimately identified as a mere procedural impediment, the substantive conclusion will be that measurements of natural killer cell activity made at lower effector to target ratios are more valid. Alternatively, if the individual variations in the progressively influential factor are modifiable, this may presage a new therapeutic route to enhancing natural killer cell activity. The methodological conclusion is that, when using structural equation models, researchers should attend to significant model ill fit even if the degree of covariance ill fit is small, because small covariance residuals do not imply that the underlying model misspecifications are correspondingly small or inconsequential.

Electrodermal dissociation of chronic fatigue and depression: evidence for distinct physiological mechanisms.

Chronic fatigue syndrome (CFS) has an estimated prevalence between 0.5% and 3%, yet its diagnosis remains contentious. CFS is characterized by subjective symptoms that can be difficult to verify; moreover, depression is a commonly reported CFS complaint, whereas fatigue is a common symptom of depression. Our primary goal was dissociation of these disorders using psychophysiological methods. As previous research has implicated the autonomic nervous system in CFS, we conducted what we believe to be the first analysis of bilateral electrodermal and skin temperature responses of dextral females in a cross-modal orienting task, to investigate differences between these two patient groups and controls. A multivariate analysis of variance (MANOVA) examining three measures of electrodermal activity revealed prestimulus tonic skin conductance levels (SCLs) were markedly lower for the CFS group, with no difference between controls and depressives. Concurrent skin temperature levels were higher for the CFS group than the other two groups. These findings indicate that, despite overtly similar cognitive and symptom profiles, depression and CFS patients can be differentiated with psychophysiological measures. This study adds to the growing body of evidence demonstrating that CFS and depression have distinct neurobiological profiles, consistent with unique aetiologies.

Alcohol problems and interest in self-help: a population study of Alberta adults.

BACKGROUND: We quantified the prevalence of alcohol problems among Alberta adults and determined relationships between sociodemographic characteristics, problem drinking status, and interest in self-help materials to reduce alcohol use. METHODS: A computer-aided telephone interview was administered to a stratified random sample of 10,014 Albertans, 18 years of age or older (5,621 women and 4,393 men; M age = 43.3 years, SD = 16.0), with a response rate of 65.4%. Measures included: 1) current drinking status, 2) prior alcohol treatment, 3) problem drinking status (using the Alcohol Use Disorders Identification Test; AUDIT), and 4) interest in receiving free self-help materials to encourage safe drinking. Data were weighted to reflect age, sex, and regional Alberta population. RESULTS: Of the total sample, 19.3% abstained from drinking in the past year, 4.2% had received treatment for alcohol problems at some point in their lives, and 80.7% were current drinkers (i.e., consumed alcohol in the previous year). Some 15.2% (n = 1,193) of current drinkers were classified as having a drinking problem. Logistic regression analyses showed that problem drinkers had 3.5 times greater odds of being male and 2.3 times greater odds of being interested in self-help interventions, compared to other current drinkers. Being single, of younger age, and not being exposed to post-secondary education also significantly predicted problem drinking status. INTERPRETATION: Alcohol misuse is common among Alberta drinkers, but many of them are interested in receiving brief public health interventions designed to help them assert control over their behaviour.