Semiparametric count data regression for self-reported mental health

‘‘For how many days during the past 30 days was your mental health not good?” The responses to this question measure self-reported mental health and can be linked to important covariates in the National Health and Nutrition Examination Survey (NHANES). However, these count variables present major distributional challenges: The data are overdispersed, zero-inflated, bounded by 30, and heaped in 5- and 7-day increments. To address these challenges—which are especially common for health questionnaire data—we design a semiparametric estimation and inference framework for count data regression. The data-generating process is defined by simultaneously transforming and rounding (star ) a latent Gaussian regression model. The transformation is estimated nonparametrically and the rounding operator ensures the correct support for the discrete and bounded data. Maximum likelihood estimators are computed using an expectation-maximization (EM) algorithm that is compatible with any continuous data model estimable by least squares. star regression includes asymptotic hypothesis testing and confidence intervals, variable selection via information criteria, and customized diagnostics. Simulation studies validate the utility of this framework. Using star regression, we identify key factors associated with self-reported mental health and demonstrate substantial improvements in goodness-of-fit compared to existing count data regression models.     

Adjusting the focus: A public health ethics approach to data research

This paper contends that a research ethics approach to the regulation of health data research is unhelpful in the era of population‐level research and big data because it results in a primary focus on consent (meta‐, broad, dynamic and/or specific consent). Two recent guidelines – the 2016 WMA Declaration of Taipei on ethical considerations regarding health databases and biobanks and the revised CIOMS International ethical guidelines for health‐related research involving humans – both focus on the growing reliance on health data for research. But as research ethics documents, they remain (to varying degrees) focused on consent and individual control of data use. Many current and future uses of health data make individual consent impractical, if not impossible. Many of the risks of secondary data use apply to communities and stakeholders rather than individual data subjects. Shifting from a research ethics perspective to a public health lens brings a different set of issues into view: how are the benefits and burdens of data use distributed, how can data research empower communities, who has legitimate decision‐making capacity? I propose that a public health ethics framework – based on public benefit, proportionality, equity, trust and accountability – provides more appropriate tools for assessing the ethical uses of health data. The main advantage of a public health approach for data research is that it is more likely to foster debate about power, justice and equity and to highlight the complexity of deciding when data use is in the public interest.     

国家级生物大数据中心展望

大数据时代下,科学大数据已经成为科技创新和社会经济发展的新动力。我国是生物数据生产大国,生命大数据是人口健康和国家安全的重要战略资源。面对我国生物数据因存储零散、缺乏系统监管而大量丢失和流失,以及严重依赖国际生物组学大数据中心的局面,亟需从国家层面建设我国自己的生命大数据保存和管理体系。本文以美国NCBI为例介绍了国际生物大数据中心的发展历程及现状,阐明我国建立国家级生物大数据中心的重要性、迫切性、当前历史机遇和发展前景。中国科学院北京基因组研究所生命与健康大数据中心为此做了大量努力,并在数据存储、汇交和转化应用上取得了阶段性成果,以期推进我国生物大数据中心的建设,提高生命科学研究的国际竞争力和影响力。     

Genetic parameters for producer-recorded health data in Canadian Holstein cattle

Health traits are of paramount importance for economic dairy production. Improvement in liability to diseases has been made with better management practices, but genetic aspects of health traits have received less attention. Dairy producers in Canada have been recording eight health traits (mastitis (MAST), lameness (LAME), cystic ovarian disease (COD), left displaced abomasum (LDA), ketosis (KET), metritis (MET), milk fever (MF) and retained placenta (RP)) since April 2007. Genetic analyses of these traits were carried out in this study for the Holstein breed. Edits on herd distributions of recorded diseases were applied to the data to ensure a sufficient quality of recording. Traits were analysed either individually (MAST, LAME, COD) or were grouped according to biological similarities (LDA and KET, and MET, MF and RP) and analysed with multiple-trait models. Data included 46 104 cases of any of the above diseases. Incidence ranged from 2.3% for MF to 9.7% for MAST. MET and KET also had an incidence below 4.0%. Variance components were estimated using four different sire threshold models. The differences between models resulted from the inclusion of days at risk (DAR) and a cow effect, in addition to herd, parity and sire effects. Models were compared using mean squared error statistic. Mean squared error favoured, in general, the sire and cow within sire model with regression on DAR included. Heritabilities on the liability scale were between 0.02 (MET) and 0.21 (LDA). There was a moderate, positive genetic correlation between LDA and KET (0.58), and between MET and RP (0.79).     

The Brazilian Family Health Strategy and adult health: Evidence from individual and local data for metropolitan areas

Previous studies have found that the expansion of primary health care in Brazil following the country-wide family health strategy (ESF), one of the largest primary care programs in the world, has improved health outcomes. However, these studies have relied either on aggregate data or on limited individual data, with no fine-grained information available concerning household participation in the ESF or local supply of ESF services, which represent crucial aspects for analytical and policy purposes. This study analyzes the relationship between the ESF and health outcomes for the adult population in metropolitan areas in Brazil. We investigate this relationship through two linked dimensions of the ESF: the program’s local supply of health teams and ESF household registration. In contrast with previous studies focusing on comparisons between certain definitions of "treated" versus "nontreated" populations, our results indicate that the local density of health teams is important to the observed effects of the ESF on adult health. We also find evidence consistent with the presence of positive primary health care spillovers to people not registered with the ESF. However, current ESF coverage levels in metropolitan areas have limited ability to address prevailing health inequalities. Our analysis suggests that the local intensity of ESF coverage should be a key consideration for evaluations and policy efforts related to future ESF expansion.     

Improving the feeding of captive felines through application of field data

The tactics employed by large felines in food-getting in the wild are contrasted with the conditions which prevail in captivity. While zoos cannot recreate a completely natural existence, there are points at which nature's ways can guide management in improving the quality of life for zoo animals. Equating an adequate diet with good nutrition leads to dietary substitutions which ignore non-nutritive requirements. Among the issues discussed are substitute activities, oral health in relation to food texture, and the psychological aspects of feeding.     

Prenatal economic shocks and birth outcomes in UK cohort data

We consider the effects of major prenatal economic shocks experienced by mothers on two indicators of newborn-infant health, birth weight and head circumference, using detailed microdata from the UK ALSPAC survey. Controlling for physiological and socioeconomic factors, an economic shock in the first 18 weeks of gestation lowers birth weight by 40–70 g and head circumference by 2–3 mm. We find evidence of transmission via poorer maternal health due to absolute material deprivation and tobacco and alcohol consumption, but not for the endocrinological effects of increased psychosocial anxiety. The fragile-male hypothesis holds for birth weight but not for head circumference, as predicted by recent theories on gender differences in prenatal development.     

Integer‐valued functional data analysis for measles forecasting

Measles presents a unique and imminent challenge for epidemiologists and public health officials: the disease is highly contagious, yet vaccination rates are declining precipitously in many localities. Consequently, the risk of a measles outbreak continues to rise. To improve preparedness, we study historical measles data both prevaccine and postvaccine, and design new methodology to forecast measles counts with uncertainty quantification. We propose to model the disease counts as an integer‐valued functional time series: measles counts are a function of time‐of‐year and time‐ordered by year. The counts are modeled using a negative‐binomial distribution conditional on a real‐valued latent process, which accounts for the overdispersion observed in the data. The latent process is decomposed using an unknown basis expansion, which is learned from the data, with dynamic basis coefficients. The resulting framework provides enhanced capability to model complex seasonality, which varies dynamically from year‐to‐year, and offers improved multimonth‐ahead point forecasts and substantially tighter forecast intervals (with correct coverage) compared to existing forecasting models. Importantly, the fully Bayesian approach provides well‐calibrated and precise uncertainty quantification for epi‐relevant features, such as the future value and time of the peak measles count in a given year. An R package is available online.     

Neural networks for clustered and longitudinal data using mixed effects models

Although most statistical methods for the analysis of longitudinal data have focused on retrospective models of association, new advances in mobile health data have presented opportunities for predicting future health status by leveraging an individual's behavioral history alongside data from similar patients. Methods that incorporate both individual-level and sample-level effects are critical to using these data to its full predictive capacity. Neural networks are powerful tools for prediction, but many assume input observations are independent even when they are clustered or correlated in some way, such as in longitudinal data. Generalized linear mixed models (GLMM) provide a flexible framework for modeling longitudinal data but have poor predictive power particularly when the data are highly nonlinear. We propose a generalized neural network mixed model that replaces the linear fixed effect in a GLMM with the output of a feed-forward neural network. The model simultaneously accounts for the correlation structure and complex nonlinear relationship between input variables and outcomes, and it utilizes the predictive power of neural networks. We apply this approach to predict depression and anxiety levels of schizophrenic patients using longitudinal data collected from passive smartphone sensor data.     

Assessing internet survey data collection methods with ethnic nurse shift workers

An increasing number of ethnic minorities are expected to enter the United States workforce based on projected demographic changes. This includes American Indian/Alaskan Native (AI/AN) nurses. Sociocultural influences on sleep disturbances, sleepiness, and other aspects related to shift-work tolerance are of unrecognized importance. More minority nurses are needed to provide culturally congruent care; however, AI/AN nurses represent less than 1% of nurses located throughout the American workforce. This article aims to verify the feasibility of Internet data collection (Web-based survey) methods and instrument stability as the first part of a two-phase study comparing individual differences and shift-work-related sleep disturbances between AI/AN and White non-Hispanic (WNH) nurses. In the first phase, an Internet survey was used to reach a cross-section of AI/AN and WNH nurses. The on-line survey was composed of accepted shift-work-related instruments. Items estimating sleep disturbances, sociocultural choices, time awareness, polychronicity, morningness/eveningness, ethnic identity, and demographic questions were asked. The survey was linked to a series of Web pages describing the study purpose, inclusion and exclusion criteria, consent form, Web survey, and the second phase of the study in which subjects were invited to participate in actigraphy measurements. The survey was pilot-tested for error codes, item confusion, length, and completion time. Forced-answer questions were added asking ethnicity, age group, license type, state where licensed, and legal name on nursing license before accessing the survey. Data were saved periodically, cued by the word “continue.” The database was located on a secure server and password protected. Nurses were recruited using published articles and printed advertisements, hospital e-mail systems, national nursing organization Web sites (minoritynurse.com; NANAINA.org), nursing Web site discussion groups, snow-balling, and word of mouth. The site was accessed 656 times with the Internet survey being completed by 138 WNH and 56 AI/AN nurses meeting the inclusion criteria. Except for the polychronicity measure (PAI3), instruments measuring time awareness, chronotype, and situational sleepiness achieved acceptable reliability coefficients with Internet data collection. Using pull-down menus would improve questions asking specific times. Internet data collection with different ethnic groups is possible; however, accessing the target population may be difficult. Despite extensive recruitment efforts, few AI/AN nurses participated. Computer literacy and failing to relate to the study's purpose may have limited the interest of the AI/AN nurses. It is possible to recruit nurse shift workers and collect individual difference and sleep disturbance data through the Internet; however, the researcher must remain vigilant throughout the process.     

Practice facilitation for improving cardiovascular care: secondary evaluation of a stepped wedge cluster randomized controlled trial using population-based administrative data

BackgroundPractice facilitation (PF), a multifaceted approach in which facilitators (external health care professionals) help family physicians to improve their adoption of best practices, has been highly successful. Improved Delivery of Cardiovascular Care (IDOCC) was an innovative PF trial designed to improve evidence-based care for people who have, or are at risk of, cardiovascular disease (CVD). The intervention was found to be ineffective as assessed by a patient-level composite score based on chart reviews from a subsample of patients (N = 5292). Here, we used population-based administrative data to examine IDOCC’s effect on CVD-related hospitalizations.MethodsIDOCC used a pragmatic, stepped wedge cluster randomized controlled design involving primary care providers recruited across Eastern Ontario, Canada. IDOCC’s effect on CVD-related hospitalizations was assessed in the 2 years of active intervention and post-intervention years. Marginal and mixed-effects regression analyses were used to account for the study design and to control for patient, physician, and practice characteristics. Secondary and subgroup analyses investigated robustness.ResultsOur sample included 262,996 patient/year observations representing 54,085 unique patients who had, or were at risk of, CVD, from 70 practices. There was a strong decreasing secular trend in CVD-related hospitalizations but no statistically significant effect of IDOCC. Relative to patients in the control condition, patients in the intervention condition were estimated to have 4 % lower odds of CVD-related hospitalizations (adjOR = 0.96, 99 % CI 0.83 to 1.11). The nonsignificant result persisted across robustness analyses.ConclusionsClinical outcomes from administrative databases were examined to form a more complete picture of the (in)effectiveness of a large-scale quality improvement intervention. IDOCC did not have a significant effect on CVD hospitalizations, suggesting that the results from the primary composite adherence score analysis were neither due to choice of outcome nor relatively short follow-up period.

Trial registration

ClinicalTrials.gov {"type":"clinical-trial","attrs":{"text":"NCT00574808","term_id":"NCT00574808"}}NCT00574808, registered on 14 December 2007.

Electronic supplementary material

The online version of this article (doi:10.1186/s13063-016-1547-2) contains supplementary material, which is available to authorized users.     

Inference for kappas for longitudinal study data: applications to sexual health research

Summary .   Analysis of instrument reliability and rater agreement is used in a wide range of behavioral, medical, psychosocial, and health-care-related research to assess psychometric properties of instruments, consensus in disease diagnoses, fidelity of psychosocial intervention, and accuracy of proxy outcomes. For categorical outcomes, Cohen's kappa is the most widely used index of agreement and reliability. In many modern-day applications, data are often clustered, making inference difficult to perform using existing methods. In addition, as longitudinal study designs become increasingly popular, missing data have become a serious issue, and the lack of methods to systematically address this problem has hampered the progress of research in the aforementioned fields. In this article, we develop a novel approach based on a new class of kappa estimates to tackle the complexities involved in addressing missing data and other related issues arising from a general multirater and longitudinal data setting. The approach is illustrated with real data in sexual health research.     

Technical note: comparison of the Maresh reference data with the who international standard for normal growth in healthy children

The Maresh reference data on stature and long bone lengths in a sample of healthy middle-class children from Denver, Colorado [Maresh: Am J Dis Child 66 (1943) 227-257; Maresh: Am J Dis Child 89 (1955) 725-742; Maresh: Human growth and development (1970) p 155-200], have been used extensively by biological anthropologists to estimate juvenile age and body size using skeletal elements and to assess growth in skeletal series from different ethnic populations or archaeological cultural groups. How well these data reflect the potentially diverse growth patterns of healthy human populations from different geographic areas is unknown. Similarly, the efficacy of using the Maresh reference data to estimate stunting prevalence in prehistoric populations is unknown. This report presents the results from a comparison of the Maresh data on supine length and standing height to the World Health Organization (WHO) international child growth standard. The WHO growth standard is meant to depict typical human growth under optimal conditions and can be used to assess children worldwide, regardless of ethnicity and socioeconomic status. The results from this comparison indicate that although the Maresh reference data generally conform to the WHO standard, reflecting a normal human growth pattern, and therefore serve as a suitable reference for comparative studies of growth patterns, these reference data are not suitable for estimating stunting prevalence.     

Latent pattern mixture models for informative intermittent missing data in longitudinal studies

A frequently encountered problem in longitudinal studies is data that are missing due to missed visits or dropouts. In the statistical literature, interest has primarily focused on monotone missing data (dropout) with much less work on intermittent missing data in which a subject may return after one or more missed visits. Intermittent missing data have broader applicability that can include the frequent situation in which subjects do not have common sets of visit times or they visit at nonprescheduled times. In this article, we propose a latent pattern mixture model (LPMM), where the mixture patterns are formed from latent classes that link the longitudinal response and the missingness process. This allows us to handle arbitrary patterns of missing data embodied by subjects' visit process, and avoids the need to specify the mixture patterns a priori. One assumption of our model is that the missingness process is assumed to be conditionally independent of the longitudinal outcomes given the latent classes. We propose a noniterative approach to assess this key assumption. The LPMM is illustrated with a data set from a health service research study in which homeless people with mental illness were randomized to three different service packages and measures of homelessness were recorded at multiple time points. Our model suggests the presence of four latent classes linking subject visit patterns to homeless outcomes.     

Ethical principles for promoting health research data sharing with sub‐Saharan Africa

A powerful feature of global health research is data‐sharing with regions which bear the heaviest burden of disease. It offers novel opportunities for aggregating data to address critical global health challenges in ways higher than relying on individual studies. Yet there exist important stratifiers of the capacity to share data, particularly across the Global North‐South divide. Systemic challenges that characterize sub‐Saharan Africa and disadvantage the region's scientific productivity threaten the burgeoning data‐sharing culture too. Like all endeavors requiring equal commitments under unequal circumstances, a strong ethical impetus is needed to help reduce inequities and imbalances to encourage adherence. This article discusses mandatory data‐sharing in relation to peculiar challenges faced by sub‐Saharan African scientists to suggest ethical principles for rethinking and reframing solutions. We propose six principles which mirror guidelines from the Institute of Medicine and encapsulate principles from the Emanuel Framework, Nairobi Data Sharing Principles, and the COHRED guidelines.     

LCI data and data quality

Life cycle inventory (LCI) is becoming an established environmental management tool that quantifies all resource usage and waste generation associated with providing specific goods or services to society. LCIs are increasingly used by industry as well as policy makers to provide a holistic ‘macro’ overview of the environmental profile of a good or service. This information, effectively combined with relevant information obtained from other environmental management tools, is very useful in guiding strategic environmental decision making. LCIs are very data intensive. There is a risk that they imply a level of accuracy that does not exist. This is especially true today, because the availability of accurate LCI data is limited. Also, it is not easy for LCI users, decision-makers and other interested parties to differentiate between ‘good quality’ and ‘poor quality’ LCI data. Several data quality requirements for ‘good’ LCI data can be defined only in relation to the specific study in which they are used. In this paper we show how and why the use of a common LCI database for some of the more commonly used LCI data, together with increased documentation and harmonisation of the data quality features of all LCI data, is key to the further development of LCI as a useful and pragmatic environmental management tool. Initiatives already underway to make this happen are also described.