Kate Ann Levin

Study design III: Cross-sectional studies

In this series, I previously gave an overview of the main types of study design and the techniques used to minimise biased results. Here, I describe cross-sectional studies, their uses, advantages and limitations.

Multilevel models with multivariate mixed response types

We build upon the existing literature to formulate a class of models for multivariate mixtures of Gaussian, ordered or unordered categorical responses and continuous distributions that are not Gaussian, each of which can be defined at any level of a multilevel data hierarchy. We describe a Markov chain Monte Carlo algorithm for fitting such models. We show how this unifies a number of disparate problems, including partially observed data and missing data in generalized linear modelling. The two-level model is considered in detail with worked examples of applications to a prediction problem and to multiple imputation for missing data. We conclude with a discussion outlining possible extensions and connections in the literature. Software for estimating the models is freely available.

Family structure, mother‐child communication, father‐child communication, and adolescent life satisfaction: A cross‐sectional multilevel analysis

Purpose – The purpose of this paper is to investigate the association between mother‐child and father‐child communication and childrens life satisfaction, and the moderating effect of communication with stepparents.Design/methodology/approach – Data from the 2006 Health Behaviour in School‐aged Children: WHO‐collaborative Study in Scotland (n=4,959) were analysed using multilevel linear regression analyses.Findings – There was an association between both mother‐child and father‐child communication and young peoples life satisfaction. Relationship with mother was particularly important, especially among girls. Among boys, not living in a traditional two‐parent family was a predictor of low life satisfaction, even when communication with one or more parents was easy. This effect was independent of economic disadvantage. The quality of the relationship with stepparents moderated these associations very slightly and in single father families only.Research limitations/implications – Strategies at the populat...

Urban–Rural Inequalities in Ischemic Heart Disease in Scotland, 1981–1999

OBJECTIVES We sought to describe the pattern and magnitude of urban-rural variation in ischemic heart disease (IHD) in Scotland and to examine the associations among IHD health indicators, level of rurality, and degree of socioeconomic deprivation. METHODS We used routine population and health data on the population aged 40-74 years between 1981 and 1999 and living in 826 small areas (average population=5600) in Scotland. Three IHD health indicators-mortality rates (deaths per 100,000 population), rates of continuous hospital stays (discharges per 100,000 population), and rates of mortality in the hospital or within 28 days of discharge (MH+) were analyzed with multilevel Poisson models. A 4-level rurality classification was used: urban areas, remote small towns, accessible rural areas, and remote rural areas. RESULTS Rates of mortality, continuous hospital stays, and MH+ increased with area socioeconomic deprivation. After adjustment for population age, gender, and deprivation, the relative risk of IHD mortality in remote rural areas was similar to that of urban areas in 1981; the relative risk of a continuous hospital stay was significantly lower (relative risk [RR] = 0.70; 95% confidence interval [CI] = 0.64, 0.76) and the relative risk of MH+ was higher (RR=1.18; 95% CI=1.04, 1.35) in remote rural areas. Mortality and MH+ declined for all ruralities over time. However, MH+ remains highest in remote rural areas and remote towns. CONCLUSIONS Low standardized ratios of IHD continuous hospital stays and mortality in remote rural areas mask health problems among rural populations. Although absolute and relative differences between urban and rural rates of MH+ have diminished, the relative risk of MH+ remains high in remote rural areas.

Psychometric Validation of the Revised Family Affluence Scale: a Latent Variable Approach

The aim was to develop and test a brief revised version of the family affluence scale. A total of 7120 students from Denmark, Greenland, Italy, Norway, Poland, Romania, Scotland and Slovakia reported on a list of 16 potential indicators of affluence. Responses were subject to item screening and test of dimensionality. Bifactor analysis revealed a strong general factor of affluence in all countries, but with additional specific factors in all countries. The specific factors mainly reflected overlapping item content. Item screening was conducted to eliminate items with low discrimination and local dependence, reducing the number of items from sixteen to six: Number of computers, number of cars, own bedroom, holidays abroad, dishwasher, and bathroom. The six-item version was estimated with Samejima’s graded response model, and tested for differential item functioning by country. Three of the six items were invariant across countries, thus anchoring the scale to a common metric across countries. The six-item scale correlated with parental reported income groups in six out of eight countries. Findings support a revision to six items in the family affluence scale.

Study design VII. Randomised controlled trials

Previously in this series, I have given an overview of the main types of study design and the techniques used to minimise the likelihood of obtaining biased results. In this article I describe more fully randomised controlled trials, their uses, advantages and limitations.

Study design IV: Cohort studies

Previously in this series I have given an overview of the main types of study design and the techniques used to minimise biased results. In this article I describe more fully cohort studies, their uses, advantages and limitations.

Study design V. Case–control studies

Previously in this series I have given an overview of the main types of study design and the techniques used to minimise biased results. In this article I describe more fully case control studies, their uses, advantages and limitations.

Trend analyses in the health behaviour in school-aged children study: methodological considerations and recommendations

BACKGROUND This article presents the scope and development of the Health Behaviour in School-aged Children (HBSC) study, reviews trend papers published on international HBSC data up to 2012 and discusses the efforts made to produce reliable trend analyses. METHODS The major goal of this article is to present the statistical procedures and analytical strategies for upholding high data quality, as well as reflections from the authors of this article on how to produce reliable trends based on an international study of the magnitude of the HBSC study. HBSC is an international cross-sectional study collecting data from adolescents aged 11-15 years, on a broad variety of health determinants and health behaviours. RESULTS A number of methodological challenges have stemmed from the growth of the HBSC-study, in particular given that the study has a focus on monitoring trends. Some of those challenges are considered. When analysing trends, researchers must be able to assess whether a change in prevalence is an expression of an actual change in the observed outcome, whether it is a result of methodological artefacts, or whether it is due to changes in the conceptualization of the outcome by the respondents. CONCLUSION The article present recommendations to take a number of the considerations into account. The considerations imply methodological challenges, which are core issues in undertaking trend analyses.

Study design I

This is the first of a series of articles that will describe the different types of study design; considerations when choosing a study design; and the advantages and disadvantages of each type of study. This first article explains the importance of choosing an appropriate design and the decisions to be made when doing so.