Should adjustment for covariates be used in prevalence estimations?

Background

Adjustment for covariates (also called auxiliary variables in survey sampling literature) is commonly applied in health surveys to reduce the variances of the prevalence estimators. In theory, adjusted prevalence estimators are more accurate when variance components are known. In practice, variance components needed to achieve the adjustment are unknown and their sample estimators are used instead. The uncertainty introduced by estimating variance components may overshadow the reduction in the variance of the prevalence estimators due to adjustment. We present empirical guidelines indicating when adjusted prevalence estimators should be considered, using gender adjusted and unadjusted smoking prevalence as an illustration.

Methods

We compare the accuracy of adjusted and unadjusted prevalence estimators via simulation. We simulate simple random samples from hypothetical populations with the proportion of males ranging from 30% to 70%, the smoking prevalence ranging from 15% to 35%, and the ratio of male to female smoking prevalence ranging from 1 to 4. The ranges of gender proportions and smoking prevalences reflect the conditions in 1999–2003 Behavioral Risk Factors Surveillance System (BRFSS) data for Massachusetts. From each population, 10,000 samples are selected and the ratios of the variance of the adjusted prevalence estimators to the variance of the unadjusted (crude) ones are computed and plotted against the proportion of males by population prevalence, as well as by population and sample sizes. The prevalence ratio thresholds, above which adjusted prevalence estimators have smaller variances, are determined graphically.

Results

In many practical settings, gender adjustment results in less accuracy. Whether or not there is better accuracy with adjustment depends on sample sizes, gender proportions and ratios between male and female prevalences. In populations with equal number of males and females and smoking prevalence of 20%, the adjusted prevalence estimators are more accurate when the ratios of male to female prevalences are above 2.4, 1.8, 1.6, 1.4 and 1.3 for sample sizes of 25, 50, 100, 150 and 200, respectively.

Conclusion

Adjustment for covariates will not result in more accurate prevalence estimator when ratio of male to female prevalences is close to one, sample size is small and risk factor prevalence is low. For example, when reporting smoking prevalence based on simple random sampling, gender adjustment is recommended only when sample size is greater than 200.     

Case-mix adjustment of the National CAHPS benchmarking data 1.0: a violation of model assumptions?

OBJECTIVE: To compare models for the case-mix adjustment of consumer reports and ratings of health care. DATA SOURCES: The study used the Consumer Assessment of Health Plans (CAHPS) survey 1.0 National CAHPS Benchmarking Database data from 54 commercial and 31 Medicaid health plans from across the United States: 19,541 adults (age > or = 18 years) in commercial plans and 8,813 adults in Medicaid plans responded regarding their own health care, and 9,871 Medicaid adults responded regarding the health care of their minor children. STUDY DESIGN: Four case-mix models (no adjustment; self-rated health and age; health, age, and education; and health, age, education, and plan interactions) were compared on 21 ratings and reports regarding health care for three populations (adults in commercial plans, adults in Medicaid plans, and children in Medicaid plans). The magnitude of case-mix adjustments, the effects of adjustments on plan rankings, and the homogeneity of these effects across plans were examined. DATA EXTRACTION: All ratings and reports were linearly transformed to a possible range of 0 to 100 for comparability. PRINCIPAL FINDINGS: Case-mix adjusters, especially self-rated health, have substantial effects, but these effects vary substantially from plan to plan, a violation of standard case-mix assumptions. CONCLUSION: Case-mix adjustment of CAHPS data needs to be re-examined, perhaps by using demographically stratified reporting or by developing better measures of response bias.     

The treatment of overdose--time for a change?

To determine emergency management of overdose relating to gastrointestinal decontamination procedures in Scottish Accident & Emergency (A & E) Departments. A postal questionnaire was sent to the 28 main A & E Departments in Scotland. There was a 75% response rate. Nineteen departments(90%) continue to perform gastric lavage, with the majority carrying out this procedure more than one hour post-ingestion of commonly presenting overdoses. Sixteen departments (76%) regularly administer activated charcoal and four (19%) use syrup of ipecacuanha. Twenty departments (95%) had access to Toxbase. The majority of respondents (86%) feel there is a need for standardisation of treatment in acute overdose. Despite the availability of guidelines provided by the UK National Poisons Information Service and the Position Statements, there is no consensus in the actual management of acute overdose among Scottish A&E departments.     

Should a single centre for the assay of biochemical markers of nutritional status be mandatory in multicentric trials?

Clinical trials to identify patients at risk and to assess new therapeutic agents in the nutritional field are often single-site. The principal advantage of mounting a multicentric trial is that patient accrual is much quicker. Albumin, transthyretin, and C-reactive protein are frequently used biochemical markers of nutritional and inflammation status. However, the different techniques, reagents, and calibrators used to measure these markers introduce wide variations in values among laboratories. This study was carried out as part of a prospective multicentric study in chronic respiratory disease patients to evaluate variability and comparability of results among laboratories for these biochemical markers, and to determine whether centralization is necessary. Thirty enrolled laboratories provided their own range of reference values for those proteins a nd were then requested to process two control samples C1 and C2 blind. The results showed a broad dispersion of values for albumin and transthyretin. In 7% of laboratories, results of albumin for C2 (mean, all techniques: 39.1+/-3 g/l) were <35 g/l, the threshold value indicating a potential risk of malnutrition. When only laboratories using immunonephelemetry were considered, the results were satisfactory (CV<10% for all proteins). Given the possible incorrect classification of patients at risk, measurement should be made per site only if all participants use an immunonephelometric method. Otherwise, a centralizing assay of these biological markers should be considered.     

Should structural interventions be evaluated using RCTs? The case of HIV prevention

Structural interventions addressing macro-social determinants of risk have been suggested as potentially important adjuncts to biomedical and behavioural interventions for the prevention of HIV and other diseases. A few interventions of this type have been evaluated using randomised controlled trials (RCTs), the most rigorous design to evaluate the effects of biomedical and behavioural interventions. The appropriateness of applying RCTs to structural interventions is however debated. This paper considers whether issues of ethics, feasibility and utility preclude the use of RCTs in evaluations of structural interventions for HIV prevention. We conclude there is nothing particular to this category of interventions prohibiting use of RCTs. However, we suggest that RCTs may prove unacceptable, unfeasible or not useful in certain circumstances, such as where an intervention brings important benefits other than HIV prevention (such as increased income); where leaders of clusters do not allow decisions about macro-social policies to be determined randomly; where the unit of social organization addressed by an intervention is so large that recruitment of adequate numbers of clusters is impossible; and where the period required to trial interventions extends beyond practical decision-making time-scales. In such cases, alternative evaluative designs must be assessed for their ability to provide evidence of intervention effectiveness.     

The De-Scent of Sexuality: Should We Smell a Rat?

Archives of Sexual Behavior -