Dataset Information

A Two-Step Approach for Analysis of Nonignorable Missing Outcomes in Longitudinal Regression: an Application to Upstate KIDS Study.

ABSTRACT: BACKGROUND:Imperfect follow-up in longitudinal studies commonly leads to missing outcome data that can potentially bias the inference when the missingness is nonignorable; that is, the propensity of missingness depends on missing values in the data. In the Upstate KIDS Study, we seek to determine if the missingness of child development outcomes is nonignorable, and how a simple model assuming ignorable missingness would compare with more complicated models for a nonignorable mechanism. METHODS:To correct for nonignorable missingness, the shared random effects model (SREM) jointly models the outcome and the missing mechanism. However, the computational complexity and lack of software packages has limited its practical applications. This paper proposes a novel two-step approach to handle nonignorable missing outcomes in generalized linear mixed models. We first analyse the missing mechanism with a generalized linear mixed model and predict values of the random effects; then, the outcome model is fitted adjusting for the predicted random effects to account for heterogeneity in the missingness propensity. RESULTS:Extensive simulation studies suggest that the proposed method is a reliable approximation to SREM, with a much faster computation. The nonignorability of missing data in the Upstate KIDS Study is estimated to be mild to moderate, and the analyses using the two-step approach or SREM are similar to the model assuming ignorable missingness. CONCLUSIONS:The two-step approach is a computationally straightforward method that can be conducted as sensitivity analyses in longitudinal studies to examine violations to the ignorable missingness assumption and the implications relative to health outcomes.

SUBMITTER: Liu D

PROVIDER: S-EPMC5610633 | biostudies-literature | 2017 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Two-Step Approach for Analysis of Nonignorable Missing Outcomes in Longitudinal Regression: an Application to Upstate KIDS Study.

Liu Danping D Yeung Edwina H EH McLain Alexander C AC Xie Yunlong Y Buck Louis Germaine M GM Sundaram Rajeshwari R

Paediatric and perinatal epidemiology 20170802 5

<h4>Background</h4>Imperfect follow-up in longitudinal studies commonly leads to missing outcome data that can potentially bias the inference when the missingness is nonignorable; that is, the propensity of missingness depends on missing values in the data. In the Upstate KIDS Study, we seek to determine if the missingness of child development outcomes is nonignorable, and how a simple model assuming ignorable missingness would compare with more complicated models for a nonignorable mechanism.<h ...[more]

PMID: 28767145

Similar Datasets

Project description:An increasing percentage of births are conceived with assisted reproductive technology (ART) and other infertility treatment. Despite findings that such treatments may be associated with diminished gestation and birth size, scarce data exist regarding infertility treatments and children's development in the United States.To assess the use and type of infertility treatment in relation to children's development through age 36 months.Prospective cohort study (conducted 2008-2014) that sampled based on infertility treatment and plurality. Included in the study were infants born between 2008 and 2010 in New York state (excluding New York City) whose parents completed developmental screening instruments through 36 months of age. A total of 4824 mothers (97% of 4989) completed 1 or more developmental screening instruments for 5841 children, including 1830 conceived with infertility treatment and 2074 twins.Maternal self-report of any infertility treatment was further categorized into ART and ovulation induction/intrauterine insemination. Assisted reproductive technology use was previously validated by linkage with the Society for Assisted Reproductive Technology-Clinical Outcome Reporting System.Five developmental domains (fine motor, gross motor, communication, personal-social functioning, and problem-solving ability), as measured by the parental completion of the Ages and Stages Questionnaires at 4, 8, 12, 18, 24, 30, and 36 months of age. Generalized linear mixed modeling techniques estimated adjusted odds ratios (aORs) and 95% CIs for use and type of infertility treatment in relation to failing a developmental domain. Data were stratified by plurality and weighted for the sampling scheme.There were 1422 mothers (29.5%; mean [SD], age, 34.1 [5.2] years) who underwent infertility treatment. Infertility treatment was not associated with risk of their children failing any developmental domain (aOR, 1.33; 95% CI, 0.94-1.89). Assisted reproductive technology was associated with increased risk for failing any developmental domain but only when singletons and twins were evaluated together (aOR, 1.81; 95% CI, 1.21-2.72). Adjustment for birth weight further attenuated this estimate (aOR, 1.26; 95% CI, 0.82-1.93). After stratifying by plurality, type of treatment also was not significantly associated with failing any developmental domain for ovulation induction/intrauterine insemination (aOR, 1.00; 95% CI, 0.57-1.77 for singletons and aOR, 1.30; 95% CI, 0.76-2.21 for twins) or ART (aOR, 1.38; 95% CI, 0.78-2.43 for singletons and aOR, 1.58; 95% CI, 0.94-2.65 for twins).After considering plurality, children's development through age 3 years was similar irrespective of infertility treatment or specific type. To our knowledge, these findings are among the first to focus on non-ART treatments in the United States.

Dataset Information

A Two-Step Approach for Analysis of Nonignorable Missing Outcomes in Longitudinal Regression: an Application to Upstate KIDS Study.

Publications

A Two-Step Approach for Analysis of Nonignorable Missing Outcomes in Longitudinal Regression: an Application to Upstate KIDS Study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets