Variable selection for distribution-free models for longitudinal zero-inflated count responses

A Distribution-Free Model for Longitudinal Metagenomic Count Data

Longitudinal metagenomics has been widely studied in the recent decade to provide valuable insight for understanding microbial dynamics. The correlation within each subject can be observed across repeated measurements. However, previous methods that assume independent correlation may suffer from incorrect inferences. In addition, methods that do account for intra-sample correlation may not be applicable for count data. We proposed a distribution-free approach, namely CorrZIDF, which extends the current method to model correlated zero-inflated metagenomic count data, offering a powerful and accurate solution for detecting significance features. This method can handle different working correlation structures without specifying each margin distribution of the count data. Through simulation studies, we have shown the robustness of CorrZIDF when selecting a working correlation structure for repeated measures studies to enhance the efficiency of estimation. We also compared four methods using two real datasets, and the new proposed method identified more unique features that were reported previously on the relevant research.

High prevalence and adverse health effects of loneliness in community-dwelling adults across the lifespan: role of wisdom as a protective factor

OBJECTIVES

This study of loneliness across adult lifespan examined its associations with sociodemographics, mental health (positive and negative psychological states and traits), subjective cognitive complaints, and physical functioning.

DESIGN

Analysis of cross-sectional data.

PARTICIPANTS

340 community-dwelling adults in San Diego, California, mean age 62 (SD = 18) years, range 27-101 years, who participated in three community-based studies.

MEASUREMENTS

Loneliness measures included UCLA Loneliness Scale Version 3 (UCLA-3), 4-item Patient-Reported Outcomes Measurement Information System (PROMIS) Social Isolation Scale, and a single-item measure from the Center for Epidemiologic Studies Depression (CESD) scale. Other measures included the San Diego Wisdom Scale (SD-WISE) and Medical Outcomes Survey- Short form 36.

RESULTS

Seventy-six percent of subjects had moderate-high levels of loneliness on UCLA-3, using standardized cut-points. Loneliness was correlated with worse mental health and inversely with positive psychological states/traits. Even moderate severity of loneliness was associated with worse mental and physical functioning. Loneliness severity and age had a complex relationship, with increased loneliness in the late-20s, mid-50s, and late-80s. There were no sex differences in loneliness prevalence, severity, and age relationships. The best-fit multiple regression model accounted for 45% of the variance in UCLA-3 scores, and three factors emerged with small-medium effect sizes: wisdom, living alone and mental well-being.

CONCLUSIONS

The alarmingly high prevalence of loneliness and its association with worse health-related measures underscore major challenges for society. The non-linear age-loneliness severity relationship deserves further study. The strong negative association of wisdom with loneliness highlights the potentially critical role of wisdom as a target for psychosocial/behavioral interventions to reduce loneliness. Building a wiser society may help us develop a more connected, less lonely, and happier society.

A novel biomarker of cardiometabolic pathology in schizophrenia?

BACKGROUND

Persons with schizophrenia and schizoaffective disorder (PwS) have high rates of cardiometabolic pathology that contributes to premature mortality. Adiponectin is a metabolic hormone affecting insulin sensitivity and inflammation, and is active in the brain. High-molecular weight (HMW) adiponectin is considered a more sensitive marker of metabolic dysfunction than total adiponectin, but has been poorly studied in schizophrenia.

METHODS

This was a cross-sectional study of 100 PwS, age range 26-68 years (46 women), and 93 age- and sex-comparable non-psychiatric comparison (NC) subjects. Assessments included measures of psychopathology, physical health, cognitive function, and circulating biomarkers of metabolic dysfunction (HMW adiponectin, lipids, insulin resistance) and inflammation (high-sensitivity C-reactive protein or hs-CRP, Tumor Necrosis Factor-α, Interleukin-6, and Interleukin-10).

RESULTS

HMW adiponectin levels were lower in PwS compared to NCs. Lower HMW adiponectin levels were associated with higher body mass index (BMI), higher Framingham risk for coronary heart disease, higher number of metabolic syndrome criteria, greater insulin resistance, lower HDL cholesterol, and higher hs-CRP in both groups. Only in PwS, lower HMW adiponectin correlated with younger age. In the best-fit regression models of HMW adiponectin, lower levels were associated with lower HDL cholesterol and minority race/ethnicity in both groups; but with younger age, non-smoking, higher insulin resistance, and a diagnosis of schizoaffective disorder only among PwS, and with male sex, better cognitive functioning, and higher hs-CRP levels in NCs only.

DISCUSSION

HMW adiponectin may be a promising biomarker of cardiometabolic health, especially among PwS. Adiponectin is a potential target for lifestyle and pharmacological interventions. Research on the possible role of HMW adiponectin in modifying cardiometabolic pathology in schizophrenia is needed.

Study of Independent Living Residents of a Continuing Care Senior Housing Community: Sociodemographic and Clinical Associations of Cognitive, Physical, and Mental Health

OBJECTIVE

To examine associations of sociodemographic and clinical factors with cognitive, physical, and mental health among independent living older adults in a continuing care senior housing community (CCSHC).

METHODS

This was a cross-sectional study at the independent living sector of a CCSHC in San Diego County, California. Participants included English-speaking adults aged 65-95 years, of which two-thirds were women. Of the 112 subjects recruited, 104 completed basic study assessments. The authors computed composite measures of cognitive, physical, and mental health. The authors also assessed relevant clinical correlates including psychosocial factors such as resilience, loneliness, wisdom, and social support.

RESULTS

The CCSHC residents were similar to a randomly selected community-based sample of older adults on most standardized clinical measures. In the CCSHC, physical health correlated with both cognitive function and mental health, but there was no significant correlation between cognitive and mental health. Cognitive function was significantly associated with physical mobility, satisfaction with life, and wisdom, whereas physical health was associated with age, self-rated physical functioning, mental well-being, and resilience. Mental health was significantly associated with income, optimism, self-compassion, loneliness, and sleep disturbances.

CONCLUSION

Different psychosocial factors are significantly associated with cognitive, physical, and mental health. Longitudinal studies of diverse samples of older adults are necessary to determine risk factors and protective factors for specific domains of health. With rapidly growing numbers of older adults who require healthcare as well as supportive housing, CCSHCs will become increasingly important sites for studying and promoting the health of older adults.

Comparison of schizophrenia outpatients in residential care facilities with those living with someone: Study of mental and physical health, cognitive functioning, and biomarkers of aging

This paper aims to compare mental and physical health, cognitive functioning, and selected biomarkers of aging reflecting metabolic pathology and inflammation, in outpatients with schizophrenia from two residential settings: residential care facilities (RCFs) and living with someone in a house/apartment. This cross-sectional study examined community-dwelling adults with schizophrenia either in RCFs (N = 100) or in a house/apartment with someone (N = 76), recruited for two NIH-funded studies in San Diego. Assessments included measures of mental/physical health, cognitive function, and metabolic (glycosylated hemoglobin, cholesterol) and inflammatory (C-Reactive Protein, Tumor Necrosis Factor-alpha, Interleukin-6) biomarkers of aging. General logistic models were used to analyze factors associated with residential status. RCF residents had several indicators of worse prognosis (never being married, higher daily antipsychotic dosages, increased comorbidities and higher Framingham risk for coronary heart disease) than individuals living with someone. However, RCF residents had better mental well-being and lower BMI, as well as comparable biomarkers of aging as those living with someone. While the cross-sectional nature of the study does not allow us to infer causality, it is possible that the supportive environment of RCFs may have a positive impact on mental and physical health of persons with schizophrenia. Longitudinal follow-up studies are needed to test this hypothesis.

A semiparametric marginalized zero-inflated model for analyzing healthcare utilization panel data with missingness

Zero-inflated count outcomes arise quite often in research and practice. Parametric models such as the zero-inflated Poisson and zero-inflated negative binomial are widely used to model such responses. However, interpretations of those models focus on the at-risk subpopulation of a two-component population mixture and fail to provide direct inference about marginal effects for the overall population. Recently, new approaches have been proposed to facilitate such marginal inferences for count responses with excess zeros. However, they are likelihood based and impose strong assumptions on data distributions. In this paper, we propose a new distribution-free, or semiparametric, alternative to provide robust inference for marginal effects when population mixtures are defined by zero-inflated count outcomes. The proposed method also applies to longitudinal studies with missing data following the general missing at random mechanism. The proposed approach is illustrated with both simulated and real study data.

Clinical Correlates of Insulin Resistance in Chronic Schizophrenia: Relationship to Negative Symptoms

Higher prevalence of physical comorbidity and premature mortality in persons with schizophrenia (PwS) results primarily from heightened cardiovascular and metabolic risks. The literature suggests that insulin resistance precedes the development of obesity, smoking, and use of antipsychotic medications, although these likely play a compounding role later in the course of the disorder. It is thus important to discover the clinical characteristics of PwS with high insulin resistance, as these individuals may represent an etiopathologically distinct subgroup with a distinct course and treatment needs. We conducted a cross-sectional study and compared insulin resistance between 145 PwS and 140 nonpsychiatric comparison (NC) participants, similar in age, sex, and race distribution. In addition, we examined correlates of insulin resistance in PwS. As expected, PwS had higher levels of insulin resistance [Homeostatic Model Assessment of Insulin Resistance (HOMA-IR)] and body mass index (BMI) compared to the NC participants. HOMA-IR in the PwS was most associated with negative symptoms, BMI, and non-White race/ethnicity. The mechanistic relationships between insulin resistance and negative symptoms in schizophrenia patients warrant further investigation, and future studies should examine outcomes of PwS with this cluster of physical and mental symptoms and determine how management of insulin resistance might improve health of these individuals.

Distribution-free models for latent mixed population responses in a longitudinal setting with missing data

Many biomedical and psychosocial studies involve population mixtures, which consist of multiple latent subpopulations. Because group membership cannot be observed, standard methods do not apply when differential treatment effects need to be studied across subgroups. We consider a two-group mixture in which membership of latent subgroups is determined by structural zeroes of a zero-inflated count variable and propose a new approach to model treatment differences between latent subgroups in a longitudinal setting. It has also been incorporated with the inverse probability weighted method to address data missingness. As the approach builds on the distribution-free functional response models, it requires no parametric distribution model and thereby provides a robust inference. We illustrate the approach with both real and simulated data.

Identification of burden hotspots and risk factors for cholera in India: An observational study

Background

Even though cholera has existed for centuries and many parts of the country have sporadic, endemic and epidemic cholera, it is still an under-recognized health problem in India. A Cholera Expert Group in the country was established to gather evidence and to prepare a road map for control of cholera in India. This paper identifies cholera burden hotspots and factors associated with an increased risk of the disease.

Methodology/Principle findings

We acquired district level data on cholera case reports of 2010–2015 from the Integrated Disease Surveillance Program. Socioeconomic characteristics and coverage of water and sanitation was obtained from the 2011 census. Spatial analysis was performed to identify cholera hotspots, and a zero-inflated Poisson regression was employed to identify the factors associated with cholera and predicted case count in the district. 27,615 cholera cases were reported during the 6-year period. Twenty-four of 36 states of India reported cholera during these years, and 13 states were classified as endemic. Of 641 districts, 78 districts in 15 states were identified as “hotspots” based on the reported cases. On the other hand, 111 districts in nine states were identified as “hotspots” from model-based predicted number of cases. The risk for cholera in a district was negatively associated with the coverage of literate persons, households using treated water source and owning mobile telephone, and positively associated with the coverage of poor sanitation and drainage conditions and urbanization level in the district.

Conclusions/Significance

The study reaffirms that cholera continues to occur throughout a large part of India and identifies the burden hotspots and risk factors. Policymakers may use the findings of the article to develop a roadmap for prevention and control of cholera in India.