The individual over time: time series applications in health care research

This paper presents a summary and a brief theoretical introduction to time series ARIMA modeling of single subject data. Time series, a statistical technique that may be appropriate when data are measured repeatedly and at nearly equal intervals of time, has potential research applications in the study of chronic diseases such as diabetes, hypertension, and herpes simplex. Both intervention models and multivariate models are covered, with examples illustrating the utility of time series techniques in chronic disease research. Time series modeling of a subject with diabetes before and after being placed on a regimen of chlorpropamide is used to demonstrate the potential of intervention analysis. Multivariate time series techniques are illustrated by modeling the relationship between exercise and blood glucose, and by modelling the relationship between psychosocial distress and lymphocyte subsets of the cellular immune system.     

Income and health: the time dimension

It is widely recognised that poverty is associated with poor health even in advanced industrial societies. But most existing studies of the relationship between the availability of financial resources and health status fail to distinguish between the transient and permanent impact of poverty on health. Many studies also fail to address the possibility of reverse causation; poor health causes low income. This paper aims to address these issues by moving beyond the static perspective provided by cross-sectional analyses and focusing on the dynamic nature of people's experiences of income and health. The specific objective is to investigate the relationship between income and health for adult participants in the British Household Panel Survey from 1991 to 1996/97. The paper pays particular attention to: the problem of health selection; the role of long-term income; and, the effect of income dynamics on health. The results confirm the general findings from the small number of longitudinal studies available in the international literature: long-term income is more important for health than current income; income levels are more significant than income change; persistent poverty is more harmful for health than occasional episodes; and, income reductions appear to have a greater effect on health than income increases. After controlling for initial health status the association between income and health is attenuated but not eliminated. This suggests that there is a causal relationship between low income and poor health.     

Dietary restraint and self-esteem as predictors of weight gain over an 8-year time period

The study aimed to assess dietary restraint and self-esteem as predictors of weight change over a time period of 8 years. Questionnaires assessing self-reported weight, body dissatisfaction, dietary restraint, and self-esteem were completed by 77 young adults (19 men and 58 women) on two occasions separated by 8 years. On average, participants gained approximately 6 kg over the 8 years. Although neither dietary restraint nor self-esteem predicted weight change on its own, their interaction did. Furthermore, the relationship between restraint and weight change was best described as curvilinear. It was concluded that dietary restraint is predictive of subsequent weight gain, but in a more complex way than previously assumed.     

Income related inequalities in mental health in Great Britain: analysing the causes of health inequality over time

Using regression techniques this paper estimates the level of income related health inequality in GB in 1992 and 1998. Inequality is decomposed to investigate which socio-demographic factors are important contributors to health differences. The paper includes a range of measured and subjective income variables to control for absolute income. A relative deprivation measure is included to test the impact of income inequality on health inequality. It is found that subjective financial status is a major determinant of ill-health and makes a major contribution to income related inequalities in health. Relative deprivation is an important contributor for women but not for men.     

Pathways from deprivation to health differed between individual and neighborhood-based indices

OBJECTIVE: To explore the role of behavioral and psychosocial factors in explaining the social gradient in self-rated health as defined either by an individual or a neighborhood deprivation index. STUDY DESIGN AND SETTING: Data were from the baseline survey of the UK Flexible Sigmoidoscopy trial. Recruitment through general practices was stratified to generate a socioeconomically diverse sample (N=5,253, aged 55-64). Assessments included an individual and neighborhood deprivation index, each of which were categorized in four levels; three behavioral and three psychosocial factors; and self-rated health. RESULTS: Neighborhood deprivation was more strongly related to behavioral than to psychosocial factors, whereas individual deprivation was strongly related to both. The social gradient in poor self-rated health (odds in most compared to least deprived group) was 6.5 for individual and 4.2 for the neighborhood deprivation index. Behavioral and psychosocial variables explained, respectively, 7% and 11% of the individual deprivation gradient and 11% and 4% of the neighborhood gradient. The psychosocial pathway did not significantly mediate the neighborhood deprivation effect on self-rated health. CONCLUSION: Intermediary pathways of the social gradient in self-rated health differed between individual and neighborhood-based deprivation indices, suggesting at least partly independent influences on poor health of individual and neighborhood-level deprivation.     

Comparing subjective and objective neighbourhood deprivation and their association with health over time among older adults in England

The importance of neighbourhood on individual health is widely documented. Less is known about the relative role of objective and subjective reports of neighbourhood conditions, how their effect on health changes as people age, and whether they moderate each other’s impact on health. This study uses the English Longitudinal Study of Ageing (ELSA) to determine whether older adults report worse self-rated health as they age, and whether this differs between objective and subjective measures of neighbourhood. ELSA data contain 53,988 person-years across six waves collected biannually between 2002 and 03 and 2012 and 13. Objective neighbourhood conditions are measured by the 2004 Index of Multiple Deprivation, and subjective neighbourhood conditions are captured by a summative neighbourhood dissatisfaction score. We find both objective and subjective neighbourhood composite scores independently predict poor health. There is no change over time in the probability of reporting poor health by baseline objective or subjective neighbourhood scores, suggesting neighbourhood effects do not compound as older adults age. There is no moderating effect of area dissatisfaction on the relationship between neighbourhood deprivation and health. The findings provide little support for causal neighbourhood effects operating in later life and indicate different causal pathways through which objective and subjective neighbourhood deprivation impact on health.     

The QALY model and individual preferences for health states and health profiles over time: a systematic review of the literature.

The numbers of quality-adjusted life years (QALYs) gained are increasingly being used to represent the gains in individual utility from treatment. This requires that the value of a health improvement to an individual is a simple product of gains in quality of life and length of life. The article reports on a systematic review of the literature on 2 issues: whether the value of a state is affected by how long the state lasts, and by states that come before or after it. It was found that individual preferences over health are influenced by the duration of health states and their sequence. However, although there is much variation across individual respondents, the assumptions tend to hold much better when valuations are aggregated across respondents, which is encouraging for economic evaluations that rely on using average (mean or median) values.     

Partnership history and mental health over time

STUDY OBJECTIVE: To describe the mental health of men and women with differing histories of partnership transitions. DESIGN: Data from nine waves of the British Household Panel Survey, a stratified general population sample, were used to calculate age standardised ratios and 95% confidence intervals for mean General Health Questionnaire scores for groups with different partnership transition histories. PARTICIPANTS: 2127 men and 2303 women aged under 65 who provided full interviews at every survey wave. MAIN RESULTS: Enduring first partnerships were associated with good mental health. Partnership splits were associated with poorer mental health, although the reformation of partnerships partially reversed this. Cohabiting was more beneficial to men's mental health, whereas marriage was more beneficial to women's mental health. The more recently a partnership split had occurred the greater the negative outcome for mental health. Women seemed more adversely affected by multiple partnership transitions and to take longer to recover from partnership splits than men. Single women had good mental health relative to other women but the same was not true for single men relative to other male partnership groups. CONCLUSIONS: Partnership was protective of mental health. Mental health was worse immediately after partnership splits, and the negative outcomes for health were longer lasting in women. Future work should consider other factors that may mediate, confound, or jointly determine the relation between partnership change and health.     

Masculinity and perceived normative health behaviors as predictors of men's health behaviors

This study examined the unique contributions of masculinity and men's perceptions of the normativeness of men's and women's health behaviors in predicting men's self-reported health behaviors. One hundred and forty men aged 18-78 were recruited from 27 unmoderated and moderated Internet listservs of potential interest to men. They completed measures on-line assessing masculinity, their perceptions of normative health behaviors for men and women, and 8 health behaviors (i.e., alcohol abuse, seatbelt use, tobacco use, physical fighting, use of social support, exercise, dietary habits, and receipt of annual medical check-ups). Findings suggest that masculinity and the perceived normativeness of other men's health behaviors significantly predicted participants' own health behaviors beyond that accounted for by socio-demographic variables (e.g., education, income). Perceptions of the normativeness of women's health behaviors were unrelated to participants' health behaviors. The findings support previous research which has found that traditional masculine gender socialization and social norms models encourage men to put their health at risk, and suggest directions for health promotion efforts when working with men.     

Social/health maintenance organization and fee-for-service health outcomes over time

Evaluating the performance of long-term care (LTC) demonstrations requires longitudinal assessment of multiple outcomes where selective mortality and disenrollment, if not accounted for, can give the appearance of reduced (or enhanced) efficacy. We assessed outcomes in social/health maintenance organizations (S/HMOs) and Medicare fee-for-service (FFS) care using a multivariate model to estimate active life expectancy (ALE). S/HMO enrollees and samples of FFS clients in four sites were analyzed and outcome differences assessed for a 3-year period. Results provide insights into S/HMO performance under different conditions and, more generally, into evaluating LTC demonstrations without randomized client and control groups.     

Multilevel analysis of individual and contextual factors as predictors of return to work

OBJECTIVE: The objective of this study was to examine if individual and contextual levels of work environment factors predict return to work (RTW). METHODS: Baseline data from 52 workplaces was linked to a national absence register. Four hundred twenty-eight persons with more than 2 weeks of sickness absence during a 2-year period were identified. Follow up was 1 year to examine three RTW outcomes. Multilevel logistic and Poisson regression models were used. RESULTS: At the individual level, significant associations were found between one psychosocial and four physical factors and RTW within 4 weeks. Two physical factors predicted RTW within 1 year. Two psychosocial and two physical factors significantly prolonged duration of sickness absence. No significant contextual level risk factors were found. CONCLUSION: At the individual level, both the psychosocial and physical work environment factors are important independent predictors of RTW.     

Social capital and change in psychological health over time

The positive association between social capital and general health outcomes has been extensively researched over the past decade; however, studies investigating social capital and psychological health show less consistent results. Despite this, policy-makers worldwide still employ elements of social capital to promote and improve psychological health. This United Kingdom study investigates the association between changes in psychological health over time and three different individual-level proxies of social capital, measures of socio-economic status, social support and the confounders age and gender. All data are derived from the British Household Panel Survey data, with the same individuals (N = 7994) providing responses from 2000–2007.     

Change in health risks and work productivity over time

We sought to examine the relationship between changes in health risks and changes in work productivity. Pre- and postanalysis was conducted on 500 subjects who participated in a wellness program at a large national employer. Change in health risks was analyzed using McNemar chi-square tests, and change in mean productivity was analyzed using paired t tests. A repeated measures regression model examined whether a change in productivity was associated with a change in health risks, controlling for age and gender. Individuals who reduced one health risk improved their presenteeism by 9% and reduced absenteeism by 2%, controlling for baseline risk level, age, gender, and interaction of baseline risk and risk change. In conclusion, reductions in health risks are associated with positive changes in work productivity. Self-reported work productivity may have utility in the evaluation of health promotion programs.     

Initial knowledge and attitudes as predictors of intervention effects: the individual management plan

This paper addresses questions concerning the effects of patient education strategies to increase adherence to antihypertensive medical regimens. It endeavors to fill gaps in our knowledge about intervention effects that may link research findings with health care applications. Data from a community intervention study of 432 hypertensives are used to address two broad questions relating research on patient education to health care practice: Do interventions have different effects on individuals who differ with respect to initial attitudes or knowledge, and to what extent do interventions produce changes in the factors thought to mediate their effects on behavior? This investigation tests a client-oriented approach to the delivery of interventions, the Individual Management Plan (IMP). The IMP approach is based on the hypothesis that interventions will be differentially effective for individuals with varying levels of initial attitudes and knowledge thought to mediate the effects of the interventions on behavior. Analysis of variance, correlational analyses, and analysis of covariance models are used to test various aspects of this hypothesis. Analyses are performed for adherence to medication regimens, dietary sodium restrictions, and weight-loss recommendations. The results of analyses offer only minimal support for the IMP hypothesis. The state of the art and constraining factors in the study are examined to guide future applications of the IMP approach.     

Factors affecting change over time in self-reported health

BACKGROUND: Self-reported health status has become a conventional measure of health status at the population level. Further, the literature supports its use as a valid indicator of morbidity and mortality. However, relatively little attention has been paid to how self-reported health status changes over time or the factors affecting change. This paper explores the factors affecting health status change over time using data from a neighbourhood health survey. METHODS: Two rounds (2001; 2003) of health survey data (n=671) were collected across 4 distinct neighbourhoods in Hamilton, Ontario. Logistic regression analysis is used to predict change in self-reported health status between the two time periods as well as determinants of change using a range of compositional, contextual and collective characteristics of individuals as potential explanatory variables. RESULTS: Results reveal that approximately one third of participants experienced a change in health status between the two survey years. Interestingly, the key factors affecting change in health status are compositional characteristics of individuals (e.g., smoking, health care use) as opposed to contextual (e.g., neighbourhood of residence) or collective (e.g., marital status). Contrary to published literature, the current study does not reveal any significant links between a change in health status and either gender or age. CONCLUSION: These results inform our understanding of both the stability of health ratings over time and the determinants of health status change. Further research should be undertaken to enhance this understanding; in particular, studies with larger sample sizes, longer time frames and more sensitive indicators of composition, context and collective are needed.     

Change over time of mortality predictors after HAART initiation in a Senegalese cohort

Background In 1998, Senegal was among the first sub-Saharan African countries to launch a Highly active anti-retroviral therapy (HAART) access program. Initial studies have demonstrated the feasibility and efficacy of this initiative. Analyses showed a peak of mortality short after starting HAART warranting an investigation of early and late mortality predictors. Methods 404 HIV-1-infected Senegalese adult patients were enrolled and data censored as of September 2005. Predictor effects on mortality were first examined over the whole follow-up period (median 46 months) using a Cox model and Shoenfeld residuals. Then, changes of these effects were examined separately over the early and late treatment periods; i.e., less and more than 6-month follow-up. Results During the early period, baseline body mass index and baseline total lymphocyte count were significant predictors of mortality (Hazard Ratios 0.82 [0.72–0.93] and 0.80 [0.69–0.92] per 200 cell/mm³, respectively) while baseline viral load was not significantly associated with mortality. During the late period, viro-immunological markers (baseline CD4-cell count and 6-month viral load) had the highest impact. In addition, the viral load at 6-month was a significant predictor (HR = 1.42 [1.20–1.66]). Conclusion In this cohort, impaired clinical status could explain the high early mortality rate while viro-immunological markers were rather predictors of late mortality.     

New tests of QALYs when health varies over time

This paper performs new tests of the QALY model when health varies over time. Our tests do not involve confounding assumptions and are robust to violations of expected utility. The results support the use of QALYs at the aggregate level, i.e. in economic evaluations of health care. At the individual level, there is less support for QALYs. The individual data are, however, largely consistent with a more general QALY-type model that remains tractable for applications.     

Tests of utility independence when health varies over time

In the conventional quality adjusted life year (QALY) model, people's preferences are assumed to satisfy utility independence. When health varies over time, utility independence implies that the value attached to a health state is independent of the health state that arise before or after it. Two separate studies were conducted involving a total of 155 respondents. In study one, we conducted five tests of utility independence using a standard gamble question. Three of the tests of utility independence were repeated in study two after randomisation was introduced in order to take account of possible ordering effects. Utility independence holds in the majority of cases examined here and so our work generally supports the use of utility independence to derive more tractable models.