Self-rated health profiles in midlife adults with chronic illness

Nursing Review of Quantitative Research

NURSING

REVIEW OF QUANTITATIVE RESEARCH REPORT

The purpose of this research is to review the work entitled: "Profiles of Self-Rated Health in Midlife Adults with Chronic Illness" which is a study conducted by Finnegan, Marion and Cox (2004)

Cognitive appraisal of one's health or SRH 'self-rated health' appraisal has been established in earlier studies to be indicative mortality, morbidity and health service utilization in midlife adults with chronic illnesses. The study of Finnegan, Marion and Cox (2004) seeks to understand the differences that exist in midlife adults with chronic illness in relation to cognitive appraisal of their own health and specifically as related to risk of further illnesses over the lifetime of these patients.

PURPOSE OF STUDY

The objective of the study conducted by Finnegan, Marion and Cox (2004) is sated to be 'two-fold' specifically for the purpose of uncovering "classes of chronically ill midlife adults who shared unique profiles of characteristics that predicted SRH over time" and secondly for the purpose of revealing "the predictive factors of SRH for each class over time." (2004)

III. QUESTIONS OF THE RESEARCH

How do the associations of variables such as type and number of chronic illnesses, symptoms, functional ability and lifestyle factors correlate with SRH in midlife adults with chronic illnesses who may be at risk for further illnesses?

What factors in the profiles or patterns of characteristics of the midlife adult with a chronic illness predict SRH?

IV. HYPOTHESIS

The hypothesis in this research is that different characteristics in the profile of the midlife adult with a chronic illness are predictors of that individual's SRH.

V. STUDY VARIABLES

Study variables include the multiple variables of type and number of chronic illnesses, symptoms, functional ability and lifestyle factors. Static background variables are stated to include demographic characteristics, social influence, previous health care experience and environmental resources. Independent variables are stated to be demographic characteristics (gender, race and ethnicity) secondly social influence (married or unpartnered) and third were environmental variables. Co morbidity, smoking, work limitation due to a health problem that limited paid work as well as being overweight.

VI. CONCEPTUAL MODEL OR THEORETICAL FRAMEWORK

The conceptual framework of the study are stated to have included "selected components of the Interaction Model of Client Health Behavior issues" which assisted in guiding the research to include variables that were "derived from the literature that would likely predict cognitive appraisal or SRH." (Finnegan, Marion and Cox, 2004) The Interaction Model of Client Health Behavior is stated to include three elements that interact which are those of: (1) client singularity; (2) client-provider interaction; and (3) health outcomes. For establishing individual or group static background variables and dynamic interpersonal variables the client singularity element was utilized. The intervention within this framework is stated to be viable because it is much better match to "the uniqueness of the individual or group." (Ibid) Within this theoretical framework SRH served as a measure of "the dynamic variable cognitive appraisal which has represented knowledge, beliefs, and attitudes toward a given health outcome..." (Ibid)

IX. SAMPLE AND SETTING

The sampling in this study was one of 23,600 individuals in 7,600 households in the state of Florida. The major limitations to this study are stated to be related to 'measurement error and endogeneity'. Stated is that endogeneity "may be confounding some of the relationships between the time-varying predictor variables and SRH."

It is stated that this explanation of reverse causality is also "plausible for depressed mood, being overweight, current smoking, more than one health condition, work limitation, and living with a partner." (Finnegan, Marion and Cox, 2004)

The sample includes 6,335 respondents born during the period from 1931 to 1941 who had at least one of the chronic illnesses of hypertension, diabetes, cancer, chronic lung disease, heart problems, stroke, psychiatric problems and arthritis." (Finnegan, Marion and Cox, (2004) There were four time-varying predictors that were found to be class dependent in the study which were those of: (1) the effects of work limitation; (2) depressed mood; (3) living with a partner; and (4) being overweight which differed significantly across the three classes in r3lation to both significant fixed effects and random effects.

X. IDENTIFICATION AND CONTROL OF EXTRANEOUS VARIABLES (WHAT ARE THE EXTRANEOUS VARIABLES IN THIS STUDY?

Finnegan, Marion & Cox report that extraneous variables in this study were not allowed well enough for in the controls of the study and that requirements of future research exist that incorporate the use of instrumental variables in order to strengthen the case for antecedent relationships between the predictors and SRH" and that in order to "establish causality definitely." (2004)

XI. STUDY INSTRUMENT

The report findings in the study of Finnegan, Marion & Cox states that the study was conducted through use of the instrument of a baseline interview that were 'in-home' and 'face-to-face' with follow-up telephone interviews conducted each second years of the study between the year of 1994 and 2000.

XII. DATA COLLECTION

Collection of the data was accomplished with files that the RAND Corporation developed from public version of HRS data through collection of information in interviews in the home of the participants and every second year by telephone interviews.

XIII. DATA ANALYSIS

Data analysis was accomplished through use of Intercooled Stata V.8. Latent class (LC) regression through use of Latent Gold 3.0 as the analytic method for achieving the aims of simultaneous classification and prediction over time (Vermont & Magidson, 2003) The LC regression model is one inclusive of "nonparametric random effects therefore the traditional regression assumption is relaxed in relation to that which the same model holds for all cases. In the LC regression model there are "no assumptions made about the distributional form of the random effects." There are existing "separate sets of regression coefficients for each individual coming from a multivariate normal distribution in parametric effects regression. (Finnegan, Marion and Cox, (2004; paraphrased)

In the LC regression models two levels of independent variables existing which are those of (1) time-constant covariates; and (2) time-varying predictor variables. The time-constant covariates influence the latent variable while the time-varying predictor variables influence the independent variable. In estimation of the model the start was with a 'one-class' model with progressive adding of classes and comparison of models through use of the "Bayesian Information criterion (BIC) which compared the "expected cell frequency counts with actual cell frequency counts found in the sample data." (Ibid) Stated is that improvements in model fit "were indicated by a decreasing BIC with each successive model as classes were added and completion of the model estimation was at the point the BIC showed a rise. A descriptive analysis was conducted as well which states that "overall there was a trend toward poorer levels of SRH with each successive data collection wave." (Ibid) Stated is that the unbalanced data was handled by Latent GOLD which estimated models through use of information that was available for "each of the response patterns." Stated is that each "case with complete baseline data contributed to the model solution. (Vermunt & Magidson, 2003) The final model selected was on the bases of theory as well as interpretation statically. It is stated that the classes within the 'three-class' model "would reflect distinct, but overlapping profiles of SRH, which could be labeled as positive health (PH) average health (AH) and negative health (NH). The PH profile reported excellent and very good health while the majority of those in the NH profile reported fair or poor health. Reports from the AH profile were rated very good, good and fair, all middle range reports.

XIV. STRENGTHS/LIMITATIONS

Strengths in this study are that use of existing data saves time, costs and resources and the ability to "study larger, more representative samples and include more variables; and opportunities to examine data from a different theoretical perspective" as well as making provisions of support for primary studies in the future. (Moriarity, et al. 1999 as cited by Finnegan, Marion & Cox, 2004) Limitations of the study are stated to be are the measurement error and endogeneity. There was limited scope in measuring some of the variables of the study due to the "availability of appropriate indicators" (Ibid) Furthermore endogeneity is stated confound relationships between the time-varying predictor variables and SRH.

Nursing: Review of Quantitative Research