Next steps for use of item response theory in the assessment of health outcomes

Objectives

We review the NCI/DIA conference, “Improving health outcomes assessment based on modern measurement theory and computerized adaptive testing,” and suggest next steps in use of item response theory (IRT) to assess health outcomes.

Background

In recent years the level of interest and use of IRT methods has increased dramatically among health outcomes researchers. The NCI/DIA conference on June 24–25, 2004, was one of the first systematic opportunities to examine many challenging issues in applying IRT to the health outcomes field.

Method

Based on the conference presentations, we identified five issues important to future applications of IRT to health outcomes.

Results

The five key issues are as follows: (1) collaboration between academia, government and industry; (2) common versus unique item banks; (3) educating and establishing standards for use and reporting of IRT; (4) demonstrating the value of IRT; and (5) continuing efforts to improve the user friendliness of IRT software.

Conclusions

Moving forward will require a collaborative effort between academia, government agencies, and industry to design and conduct IRT research. A common item bank developed with collaboration from investigators from multiple institutions could be very valuable to the field. The establishment of consensus standards for use and reporting of IRT results would help users and consumers of the methodology. Clear documentation of how IRT can lead to better patient-reported outcome measures and more accurate understanding of substantive issues is essential. Academia, government and industry should continue current work to enhance the user-friendliness of the IRT software.

     

The effects of an intensive lifestyle modification program on carotid artery intima-media thickness: a randomized trial

PURPOSE: This study evaluated the effect of the Dr. Dean Ornish Program for Reversing Heart Disease on cardiovascular disease as measured by the intima-media thickness of the common carotid artery and compared this effect to outcomes from patients participating in traditional cardiac rehabilitation. DESIGN: Randomized clinical trial. SETTING: Swedish American Health System. SUBJECTS: Ninety three patients with clinically confirmed coronary artery disease were randomly assigned to the intervention (n = 46) or traditional cardiac rehabilitation (n = 47). INTERVENTION: Dr. Dean Ornish Program for Reversing Heart Disease. MEASURES: Ultrasound of the carotid artery and other cardiovascular risk factors were measured at baseline, 6, and 12 months. ANALYSIS: Intent-to-treat analysis. RESULTS: There was no significant reduction in the carotid intima-media thickness of the carotid artery in the Ornish group or the cardiac rehabilitation group. Ornish Program participants had significantly improved dietary habits (p < .001), weight (p < .001), and body mass index (p < .001) as compared with the rehabilitation group. The decrease in the number of patients with angina from baseline to 12 months was 44% in Ornish and 12% in cardiac rehabilitation. CONCLUSIONS: The Ornish Program appears to causes improvements in cardiovascular risk factors but does not appear to change the atherosclerotic process as it affects the carotid artery.     

Emerging trends in health and productivity management

Many large U.S. employers have generally embraced a Health and Productivity Management (HPM) perspective to guide their multiple employee health management efforts. In looking ahead there are a number of emerging trends that are helping to shape these efforts. As health promotion professionals assess the implications of these trends on their respective role and function within the worksite, it may provide a very useful process for refining strategies for programming and professional development. The identified trends also have a variety of implications for health promotion vendors and the growth of the health management marketplace.     

Why don't some women return for cervical smears? A hermeneutic phenomenological investigation

To enhance understanding of how having a cervical smear can lead some women not to keep up-to-date with this test, a hermeneutic (interpretative) phenomenological study was undertaken. Participants were six purposively selected New Zealand women -- predominantly Māori -- at least 6 months overdue for a follow-up cervical screen in the previous 6 years. Each woman gave an in-depth interview. Transcribed and analyzed via a general inductive approach, the interviews suggested that the smears can violate women's positive aloneness with their bodies, and magnify aloneness as a negative state. Overdueness for the test avoids these effects. To minimize such effects, primary health care needs to acknowledge and address these issues, for example by providing an opportunity to connect the aloneness to trusted others.     

Confidence intervals for biomarker-based human immunodeficiency virus incidence estimates and differences using prevalent data

Prevalent biologic specimens can be used to estimate human immunodeficiency virus (HIV) incidence using a two-stage immunologic testing algorithm that hinges on the average time, T, between testing HIV-positive on highly sensitive enzyme immunoassays and testing HIV-positive on less sensitive enzyme immunoassays. Common approaches to confidence interval (CI) estimation for this incidence measure have included 1) ignoring the random error in T or 2) employing a Bonferroni adjustment of the box method. The authors present alternative Monte Carlo-based CIs for this incidence measure, as well as CIs for the biomarker-based incidence difference; standard approaches to CIs are typically appropriate for the incidence ratio. Using American Red Cross blood donor data as an example, the authors found that ignoring the random error in T provides a 95% CI for incidence as much as 0.26 times the width of the Monte Carlo CI, while the Bonferroni-box method provides a 95% CI as much as 1.57 times the width of the Monte Carlo CI. Further research is needed to understand under what circumstances the proposed Monte Carlo methods fail to provide valid CIs. The Monte Carlo-based CI may be preferable to competing methods because of the ease of extension to the incidence difference or to exploration of departures from assumptions.     

Family Size,Birth Order,and Parental Age Among Male Paraphilics and Sex Offenders

A sample of 1823 male paraphilics, sex offenders, and non-sex offender controls were compared on family size, birth order, and parents’ ages at the time of the probands’ births. Sample data were also compared to population data from Statistics Canada. The men in all groups were from larger than average Canadian families and they tended to be later born. Paraphilics and sex offenders had even larger families than offender controls. Their parents tended to be older at their birth with 34.2% of mothers and 51.3% of fathers over 30 years of age, but there were no statistically significant subgroup differences. There were also significantly more multiparous teenage mothers than expected and more paraphilics’ fathers who were younger than the mothers, both factors associated in the literature with increased risk of perinatal complications and abnormalities. The confounding influences of parental age, birth order, and family size were examined and indicated the need for large samples and multivariate analysis in evaluating the role of family variables associated with paraphilics and sex offenders.