Ordinal dose-response modeling approach for the phthalate syndrome

BACKGROUND

The phthalate syndrome (PS) is a collection of related male reproductive developmental effects, ranging in severity, that have been observed in rats after gestational exposure to developmentally-toxic phthalates. For statistical purposes, the PS is defined as a single endpoint and one dose-response analysis is conducted, rather than conducting multiple analyses on each individual endpoint.

OBJECTIVE

To improve dose-response modeling approaches for the PS and other syndromes of effects by accounting for differing severity levels among the endpoints.

METHODS

Ordinal dose-response modeling was performed on PS data from a published study of diisobutyl phthalate (DIBP) gestational exposure to male Sprague-Dawley rats. To incorporate PS endpoint severity, the endpoints were categorized into ordinal levels based on the expected impact of male developmental endpoint's on fertility. Then, a benchmark dose was estimated for each ordinal level. A bootstrap procedure was used to account for the nested nature of the data, and a sensitivity analysis was performed to assess the bootstrap results. A comparison of the estimates between the ordinal and the dichotomous model was performed.

RESULTS

The ordinal version of the log-logistic model applied to the data categorized by PS endpoint severity level provided benchmark dose estimates that were closer to each other in value and had lower variability than the traditional dichotomous application. The sensitivity analysis confirmed the validity of the bootstrap results.

CONCLUSION

The ordinal dose-response modeling method accounts for severity differences among dichotomous PS endpoints, can be expanded in the future to include more severity levels, and can be used in both single and cumulative phthalate risk assessments.

Robustness of Parameter Estimation to Assumptions of Normality in the Multidimensional Graded Response Model

A central assumption that is implicit in estimating item parameters in item response theory (IRT) models is the normality of the latent trait distribution, whereas a similar assumption made in categorical confirmatory factor analysis (CCFA) models is the multivariate normality of the latent response variables. Violation of the normality assumption can lead to biased parameter estimates. Although previous studies have focused primarily on unidimensional IRT models, this study extended the literature by considering a multidimensional IRT model for polytomous responses, namely the multidimensional graded response model. Moreover, this study is one of few studies that specifically compared the performance of full-information maximum likelihood (FIML) estimation versus robust weighted least squares (WLS) estimation when the normality assumption is violated. The research also manipulated the number of nonnormal latent trait dimensions. Results showed that FIML consistently outperformed WLS when there were one or multiple skewed latent trait distributions. More interestingly, the bias of the discrimination parameters was non-ignorable only when the corresponding factor was skewed. Having other skewed factors did not further exacerbate the bias, whereas biases of boundary parameters increased as more nonnormal factors were added. The item parameter standard errors recovered well with both estimation algorithms regardless of the number of nonnormal dimensions.