All Items 4

Analysis of partially observed clustered data using generalized estimating equations and multiple imputation

Horton, Nicholas J. (Department of Mathematics and Statistics, Amherst College)

Clustered data arise in many settings, particularly within the social and biomedical sciences. For example, multiple-source reports are commonly collected in child and adolescent psychiatric epidemiologic studies where researchers use various informants (for instance, parents and adolescents) to provide a holistic view of a subject’s symptoms. Fitzmaurice et al. (1995, American Journal of Epidemiology 142: 1194–1203) have described estimation of multiple-source models using a standard generalized estimating equation (GEE) framework. However, these studies often have missing data because additional stages of consent and assent are required. The usual GEE is unbiased when data are missing completely at random in the context of Little and Rubin (2002, Statistical Analysis with Missing Data [Wiley]). This is a strong assumption that may not be tenable. Other options, such as the weighted GEE, are computationally challenging when missingness is nonmonotone. Multiple imputation is an attractive method to fit incomplete data models while requiring only the less restrictive missing-at-random assumption. Previously, estimation of partially observed clustered data was computationally challenging. However, recent developments in Stata have facilitated using them in practice. We demonstrate how to use multiple imputation in conjunction with a GEE to investigate the prevalence of eating disorder symptoms in adolescents as reported by parents and adolescents and to determine the factors associated with concordance and prevalence. The methods are motivated by the Avon Longitudinal Study of Parents and their Children, a cohort study that enrolled more than 14,000 pregnant mothers in 1991–92 and has followed the health and development of their children at regular intervals. While point estimates for the missing-at-random model were fairly similar to those for the GEE under missing completely at random, the missing-at-random model had smaller standard errors and required less stringent assumptions regarding missingness.

Analysis of partially observed clustered data using generalized estimating equations and multiple imputation
Prevalence of purging at age 16 and associations with negative outcomes among girls in three community-based cohorts

Horton, Nicholas J. (Department of Mathematics and Statistics, Amherst College)

Background--The comorbidity of purging behaviours, such as vomiting, inappropriate use of laxatives, diuretics or slimming medications, has been examined in literature. However, most studies do not include adolescents, individuals who purge in the absence of binge eating, or those purging at subclinical frequency. This study examines the prevalence of purging among 16-year-old girls across three countries and their association with substance use and psychological comorbidity. Methods--Data were obtained by questionnaire in 3 population-based cohorts (Avon Longitudinal Study of Parents and Children (ALSPAC), United Kingdom, n = 1,608; Growing Up Today Study (GUTS), USA, n = 3,504; North Finland Birth Cohort (NFBC85/86), Finland, n = 2,306). Multivariate logistic regressions were employed to estimate associations between purging and outcomes. Four models were fit adjusting for binge eating and potential confounders of these associations. Results--In ALSPAC, 9.7% of girls reported purging in the 12-months prior to assessment, 7.3% in GUTS, and 3.5% in NFBC. In all 3 cohorts, purging was associated with adverse outcomes such as binge drinking (ALSPAC: odds ratio (OR) = 2.0, 95% confidence interval (CI) = 1.4–2.9; GUTS: OR = 2.5, 95% CI = 1.5–4.0; NFBC: OR = 1.7, 95% CI = 1.0–2.8), drug use (ALSPAC: OR = 2.9, 95% CI = 1.8–4.7; GUTS: OR = 4.5, 95% CI = 2.8–7.3; NFBC: OR = 4.1, 95% CI = 2.6–6.6), depressive symptoms in ALSPAC (OR = 2.2, 95% CI = 1.5–3.1) and GUTS(OR = 3.7, 95% CI = 2.2–6.3), and several psychopathology measures including clinical anxiety/depression in NFBC (OR = 11.2, 95% CI = 3.9, 31.7). Conclusion--Results show a higher prevalence of purging behaviours among girls in the United Kingdom compared to those in the United States and Finland. Our findings support evidence highlighting that purging in adolescence is associated with negative outcomes, independent of its frequency and binge eating.

Prevalence of purging at age 16 and associations with negative outcomes among girls in three community-based cohorts
Web-based alcohol screening and brief intervention for university students: A randomized trial

Horton, Nicholas J. (Department of Mathematics and Statistics, Amherst College)

Importance--Unhealthy alcohol use is a leading contributor to the global burden of disease, particularly among young people. Systematic reviews suggest efficacy of web-based alcohol screening and brief intervention and call for effectiveness trials in settings where it could be sustainably delivered. Objective--To evaluate a national web-based alcohol screening and brief intervention program. Main Outcomes and Measures--A fully automated 5-month follow-up assessment was conducted that measured 6 primary outcomes: consumption per typical occasion, drinking frequency, volume of alcohol consumed, an academic problems score, and whether participants exceeded medical guidelines for acute harm (binge drinking) and chronic harm (heavy drinking). A Bonferroni-corrected significance threshold of .0083 was used to account for the 6 comparisons and a sensitivity analysis was used to assess possible attrition bias. Conclusions and Relevance--A national web-based alcohol screening and brief intervention program produced no significant reductions in the frequency or overall volume of drinking or academic problems. There remains a possibility of a small reduction in the amount of alcohol consumed per typical drinking occasion.

Web-based alcohol screening and brief intervention for university students: A randomized trial
Assessing eating disorder symptoms in adolescence: Is there a role for multiple informants?

Horton, Nicholas J. (Department of Mathematics and Statistics, Amherst College)

Objectives--Epidemiologic studies of adolescent psychiatric disorders often collect information from adolescents and parents, yet most eating disorder epidemiologic studies rely only on adolescent report. Methods--We studied the eating disorder symptom reports provided by 7,968 adolescents from the Avon Longitudinal Study of Parents and Children (ALSPAC), and their parents, who were sent questionnaires at participants’ ages 14 and 16 years. Both adolescents and parents were asked questions about the adolescent's eating disorder symptoms, including binge eating, vomiting, laxative use, fasting, and thinness. We assessed the concordance of parent and adolescent report cross-sectionally using kappa coefficients, and further looked at how the symptom reports were predictive of adolescent body mass and composition measured at a clinical assessment at 17.5 years. Generalized estimating equations were used to model the symptom reports’ associations with risk factors and clinical outcomes. Results--Parents and adolescents were largely discordant on symptom reports cross-sectionally (kappas0.3), with the parent generally less likely to report bulimic symptoms than the adolescent but more likely to report thinness. Female adolescents were more likely to report bulimic symptoms than males (e.g., 2-4 times more likely to report binge eating), while prevalence estimates according to parent reports of female vs. male adolescents were similar. Both parent and adolescent symptom reports at ages 14 and 16 years were predictive of age-17.5 body mass and composition measures; parentally-reported binge eating was more strongly predictive of higher body mass and composition.

Assessing eating disorder symptoms in adolescence: Is there a role for multiple informants?