Dietary assessment instruments are susceptible to intervention-associated response set bias.

OBJECTIVE: Evaluations of trials of the effectiveness of dietary intervention programs may be compromised by response set biases, such as those attributable to social desirability. Participants who receive a behavioral intervention may bias their reports of diet to appear in compliance with intervention goals. This study examined whether responses to standard dietary assessment instruments could be affected by a brief dietary intervention. DESIGN: We assigned 192 undergraduate students randomly to (a) see a 17-minute videotape on the consequences of eating a high-fat diet or a placebo videotape on workplace management and (b) receive preintervention and post-intervention assessments or only postintervention assessment. Dietary assessments included 4 independent measures of fat intake. RESULTS: Among women, bias (intervention minus control) was -9.7 g fat (from a short food frequency questionnaire) and -0.6 high-fat foods (from a questionnaire about use of 23 foods in the previous day) (P < .05 for both). No results were significant among men or for 2 instruments that measured more qualitative aspects of fat-related dietary habits. APPLICATIONS: Even a modest dietary intervention can affect responses to dietary assessment instruments. Nutritionists should recognize that assessment of adherence to dietary change recommendations, when based on dietary self-report, can be overestimated as a result of response set biases.

Data analysis and sample size issues in evaluations of community-based health promotion and disease prevention programs: a mixed-model analysis of variance approach.

The growing interest in community-based approaches to health promotion and disease prevention (HP/DP) has been accompanied by a growing need to evaluate the effectiveness of such programs. Special issues that arise in these evaluation studies include (1) entire communities are assigned to intervention and control groups, (2) only a small number of communities can usually be studied, (3) the time course of changes in behavior and other outcomes is often of interest, and (4) surveys to measure such changes over time can be conducted with either repeated cross-sectional samples or with longitudinal samples. This paper shows how these issues can be addressed under a mixed-model analysis of variance approach. This approach serves to unify several ideas in the literature on evaluation of community studies, including use of time-series regression and the question of whether the individual or the community should be the unit of analysis. We also describe how the method can be used to estimate sample size requirements, statistical power, or minimum detectable program effect.

Poverty and HIV seropositivity: the poor are more likely to be infected.

We analyzed demographic and behavioral risk factors for HIV seropositivity using data from 3601 clients of the main HIV counseling and testing clinic for high-risk people in Seattle, Washington, USA. Clients with lower income were found to be more likely to be HIV seropositive, before and after controlling for other demographic and risk factors with logistic regression. This result supports the hypothesis that the impoverished are at increased risk for HIV infection due to the physical and social circumstances in which their poverty places them. These may include poor access to risk-reduction information and less support for implementation of risk-reduction strategies.

Evaluation of the use of braces to prevent injury to the knee in collegiate football players.

Data were collected in 1984 from seventy-one schools that were members of Division I of the National College Athletic Association and in 1985 from sixty-one of these schools to assess whether the use of so-called preventive braces for the knee was associated with a decrease in either the severity or the incidence (or both) of injuries to the knee in collegiate football players. Over-all, players who wore braces on the knees had significantly more injuries to the knee than players who did not. The severity of the injuries was no different in the two groups. Based on these findings, we cannot recommend the use of these braces in an attempt to prevent injury to collegiate football players.

The Seattle evaluation of computerized drug profiles: effects on prescribing practices and resource use.

Since 1979, all outpatient pharmacy transactions at the US Public Health Service Hospital in Seattle have been captured in a computer system which generates a profile of each patient's active and previously used drugs. We conducted a controlled trial in which patients were allocated to profile or no-profile groups while the computer continued to collect data on everyone. In all, 41,572 clinic visits made by 6,186 patients were studied. The incidence of preventable drug-drug interactions and redundancies was very low and was unaffected by profiles. For unclear reasons, prescription of two interacting drugs on the same visit was significantly more common for patients with profiles. The duration of drug-drug interaction episodes was significantly shorter for profile group patients, perhaps due to earlier detection of the error on subsequent visits. Profiles had no effect on prescribing volume or coordination of drug refill and visit schedules, but profile group patients made about 5 per cent fewer clinic visits than those in the no-profile group. In this setting, it appears that the prescribing of interacting or redundant drugs is more often due to inadequate provider knowledge than to inaccessible patient-specific drug data. Prevention of such errors would thus require a more active educational or monitoring program.

The Seattle evaluation of computerized drug profiles: effect on provider activities.

We studied whether furnishing care providers with computer-generated summaries of patients' current and past medications would reduce the time they spent on various drug-related tasks during patient visits. An observer used time-sampling methods to measure the amount of provider time spent on each of 10 activities during 166 clinic visits, some with profiles and some without them. Additional data were taken from the medical record on factors that might affect the time spent on various tasks. The results suggest that record reading time was reduced for first encounters between patients and providers in the medical clinics, where prescribing volume was highest; in other situations, no effect of profiles on record reading time was evident. Providers continued to document drug data in their progress notes, whether or not a profile was available, saving no time. Prescription writing required about one third less time when a profile was used than when a traditional prescription blank was used. Both uses of profiles for prescribing and the time saved per prescription also increased sharply with the number of drugs prescribed per visit.

Phased trial of a proven algorithm at a new primary care clinic.

A previous study showed that a clinical algorithm for respiratory illnesses, consisting of a checklist, a set of instructions (logic), and computer audit/feedback, could reduce costs significantly while maintaining a high quality of care. The results of this study show that the algorithm system, developed and validated at one primary care clinic, can be successfully imported to another primary care clinic. In the present study, the algorithm system significantly improved the completeness of the medical records, reduced the use of medical tests by 20 per cent-75 per cent, and reduced non-provider costs by 36 per cent per patient visit. This study also shows that all three components of the algorithm system appear to be necessary to achieve these improvements and maintain a high quality of medical care. These results suggest that a wider use of the algorithm system for minor acute medical problems is both feasible and useful in providing high-quality cost-effective care that is auditable.

Two studies of good clinical judgment.

There is a continuing controversy about the quality of unaided clinical judgment. This paper reports two studies that show that experienced medical providers made accurate probability assessments and applied those assessments to patient care in a manner consistent with principles of optimal decision making. In the first study, experienced clinicians and physician assistants accurately judged the relative frequency of three cough-related diagnoses in an outpatient population, suggesting that their encounter with several "unrelated" diagnostic problems does not interfere with their ability to judge accurately the frequency of any single diagnostic problem. In the second study, a group of clinicians assessed the probability that each patient seen in an outpatient clinic had pneumonia. Physicians were more likely to assign a pneumonia diagnosis and to order a radiograph for patients with a greater assessed probability of pneumonia (p less than 0.05). Most of the physicians appeared to use cutoff probabilities or "thresholds" above which they acted as though the patient had the disease and below which they acted as though the patient did not have the disease, consistent with rational decision-making principles. However, the threshold probabilities being used were quite different from physician to physician, implying that the physicians managed the patient population in a nonuniform manner. Thus it may be desirable to supplement "good" clinical judgment with decision aids to ensure standardized medical care.