A latent class model to multiply impute missing treatment indicators in observational studies when inferences of the treatment effect are made using propensity score matching

Analysts often estimate treatment effects in observational studies using propensity score matching techniques. When there are missing covariate values, analysts can multiply impute the missing data to create m completed data sets. Analysts can then estimate propensity scores on each of the completed data sets, and use these to estimate treatment effects. However, there has been relatively little attention on developing imputation models to deal with the additional problem of missing treatment indicators, perhaps due to the consequences of generating implausible imputations. However, simply ignoring the missing treatment values, akin to a complete case analysis, could also lead to problems when estimating treatment effects. We propose a latent class model to multiply impute missing treatment indicators. We illustrate its performance through simulations and with data taken from a study on determinants of children's cognitive development. This approach is seen to obtain treatment effect estimates closer to the true treatment effect than when employing conventional imputation procedures as well as compared to a complete case analysis.     

Genotype Imputation Reference Panel Selection Using Maximal Phylogenetic Diversity

The recent dramatic cost reduction of next-generation sequencing technology enables investigators to assess most variants in the human genome to identify risk variants for complex diseases. However, sequencing large samples remains very expensive. For a study sample with existing genotype data, such as array data from genome-wide association studies, a cost-effective approach is to sequence a subset of the study sample and then to impute the rest of the study sample, using the sequenced subset as a reference panel. The use of such an internal reference panel identifies population-specific variants and avoids the problem of a substantial mismatch in ancestry background between the study population and the reference population. To efficiently select an internal panel, we introduce an idea of phylogenetic diversity from mathematical phylogenetics and comparative genomics. We propose the “most diverse reference panel”, defined as the subset with the maximal “phylogenetic diversity”, thereby incorporating individuals that span a diverse range of genotypes within the sample. Using data both from simulations and from the 1000 Genomes Project, we show that the most diverse reference panel can substantially improve the imputation accuracy compared to randomly selected reference panels, especially for the imputation of rare variants. The improvement in imputation accuracy holds across different marker densities, reference panel sizes, and lengths for the imputed segments. We thus propose a novel strategy for planning sequencing studies on samples with existing genotype data.     

The changing incidence of human papillomavirus-associated oropharyngeal cancer using multiple imputation from 2000 to 2010 at a Comprehensive Cancer Centre

Introduction Human papillomavirus (HPV) is a risk and prognostic factor for oropharyngeal cancer (OPC). Determining whether the incidence of HPV-associated OPC is rising informs health policy. Methods HPV status was ascribed using p16 immunohistochemistry in 683/1474 OPC patients identified from the Princess Margaret Hospital's Cancer Registry (from 2000 to 2010). Missing p16 data was estimated using multiple (n = 100) imputation (MI) and validated using an independent OPC cohort (n = 214). Non-OPC head and neck squamous cell carcinoma (HNSCC) (n = 3262) were also used for time-trend comparison. Regression was used to compare HNSCC subsets and time-trends. The c-index was used to measure the predictive ability of MI. Results The incidence of OPC rose from 23.3% of all HNSCC in 2000 to 31.2% in 2010 (p = 0.002). In the subset of OPC tested for p16, there was no change in p16 positivity over time (p = 0.9). However, p16 testing became more frequent over time (p < 0.0001), but was nonetheless biased, favouring never-smokers [OR 1.87 (95% CI 1.29–2.70)] and tumors of the tonsil [OR 2.30 (1.52–3.47)] or base-of-tongue [OR 1.72 (1.10–2.70)]. These same factors were also associated with p16-positivity [ORs 3.22 (1.27–8.16), 7.26 (3.50–15.1), 5.83 (2.70–12.7), respectively]. Following MI and normalization, the proportion of OPC that was p16-associated rose from 39.8% in 2000 to 65.0% in 2010, p = 0.002, fully explaining the rise in OPC in our patient population. Conclusion The rise in HNSCC referrals seen from 2000 to 2010 at our institution was driven primarily by p16-associated OPC. MI was necessary to derive reliable conclusions when cases with missing data are considerable.     

Selection models and pattern-mixture models for incomplete data with covariates

Most models for incomplete data are formulated within the selection model framework. This paper studies similarities and differences of modeling incomplete data within both selection and pattern-mixture settings. The focus is on missing at random mechanisms and on categorical data. Point and interval estimation is discussed. A comparison of both approaches is done on side effects in a psychiatric study.     

A nonparametric procedure for the two-factor mixed model with missing data

We develop a nonparametric imputation technique to test for the treatment effects in a nonparametric two-factor mixed model with incomplete data. Within each block, an arbitrary covariance structure of the repeated measurements is assumed without the explicit parametrization of the joint multivariate distribution. The number of repeated measurements is uniformly bounded whereas the number of blocks tends to infinity. The essential idea of the nonparametric imputation is to replace the unknown indicator functions of pairwise comparisons by the corresponding empirical distribution functions. The proposed nonparametric imputation method holds valid under the missing completely at random (MCAR) mechanism. We apply the nonparametric imputation on Brunner and Dette's method for the nonparametric two-factor mixed model and this extension results in a weighted partial rank transform statistic. Asymptotic relative efficiency of the nonparametric imputation method with the complete data versus the incomplete data is derived to quantify the efficiency loss due to the missing data. Monte Carlo simulation studies are conducted to demonstrate the validity and power of the proposed method in comparison with other existing methods. A migraine severity score data set is analyzed to demonstrate the application of the proposed method in the analysis of missing data.