Matthew Hankins

The use and reporting of cluster analysis in health psychology: A review

PURPOSE Cluster analysis is a collection of relatively simple descriptive statistical techniques with potential value in health psychology, addressing both theoretical and practical problems. There are many methods of cluster analysis from which to choose, with no clear guidelines to aid researchers. In the absence of guidelines it is likely that methods already reported by published researchers will be adopted, and so clear reporting of statistical methodology, while always important, is particularly crucial with cluster analysis. The aim of this review is to describe and evaluate the reporting of cluster analysis in health psychology publications. METHODS Electronic searches of 18 health psychology journals identified 59 articles using cluster analysis published between 1984 and 2002. Articles were submitted to systematic evaluation against published criteria for the reporting of cluster analysis. RESULTS Just 27% of the papers reviewed met all five criteria, although 61% met at least four. Details of the similarity measure and the computer program used were most frequently omitted. Furthermore, while researchers usually reported the procedures employed to determine the number of clusters and to validate the clusters, these procedures were often lacking in rigour, and were reported in insufficient detail for replication. CONCLUSIONS The reporting of cluster analysis was found to be generally unsatisfactory, with many studies failing to provide enough information to allow replication or the evaluation of the quality of the research. Clear guidelines for conducting and reporting cluster analyses in health psychology are needed.

Psychological impact of human papillomavirus testing in women with borderline or mildly dyskaryotic cervical smear test results: cross sectional questionnaire study

Abstract Objective To describe the psychological impact on women of being tested for human papillomavirus (HPV) when smear test results are borderline or mildly dyskaryotic. Design Cross sectional questionnaire study. Setting Two centres participating in an English pilot study of HPV testing in women with borderline or mildly dyskaryotic smear test results. Participants Women receiving borderline or mildly dyskaryotic smear test results tested for HPV and found to be HPV positive (n = 536) or HPV negative (n = 331); and women not tested for HPV with borderline or mildly dyskaryotic smear results (n = 143) or normal smear results (n = 366). Main outcome measures State anxiety, distress, and concern about test result, assessed within four weeks of receipt of results. Results Women with borderline or mildly dyskaryotic smear results who were HPV positive were more anxious, distressed, and concerned than the other three groups. Three variables independently predicted anxiety in HPV positive women: younger age (β = −0.11, P = 0.03), higher perceived risk of cervical cancer (β = 0.17, P < 0.001), and reporting that they did not understand the meaning of test results (β = 0.17, P = 0.001). Testing HPV negative was not reassuring: among women with abnormal smear test results, those who were HPV negative were no less anxious than those who were not tested for HPV. Conclusions Informing women more effectively about the meaning of borderline or mildly dyskaryotic smear test results and HPV status, in particular about the absolute risks of cervical cancer and the prevalence of HPV infection, may avoid some anxiety for those who are HPV positive while achieving some reassurance for those who test HPV negative.

Psychological stress and wound healing in humans: A systematic review and meta-analysis ☆

OBJECTIVE The current review aims to synthesize existing knowledge about the relationship between psychological stress and wound healing. METHODS A systematic search strategy was conducted using electronic databases to search for published articles up to the end of October 2007. The reference lists of retrieved articles were inspected for further studies and citation searches were conducted. In addition, a meta-analysis of a subset of studies was conducted to provide a quantitative estimation of the influence of stress on wound healing. RESULTS Twenty-two papers met the inclusion criteria of the systematic review and a subsample of 11 was included in a meta-analysis. The studies assessed the impact of stress on the healing of a variety of wound types in different contexts, including acute and chronic clinical wounds, experimentally created punch biopsy and blister wounds, and minor damage to the skin caused by tape stripping. Seventeen studies in the systematic review reported that stress was associated with impaired healing or dysregulation of a biomarker related to wound healing. The relationship between stress and wound healing estimated by the meta-analysis was r=-0.42 (95% CI=-0.51 to -0.32) (P<.01). CONCLUSION Attention now needs to be directed towards investigating potential moderators of the relationship, mediating mechanisms underpinning the association, as well as the demonstration of a causal link by the development of experimental interventions in healthy populations.

Cognitive impairment and 7-year mortality in dialysis patients.

BACKGROUND Although dementia has predicted mortality in large dialysis cohorts, little is known about the relationship between less pronounced cognitive deficits and mortality in patients with end-stage renal disease. This study assessed whether cognitive impairment without dementia was an independent predictor of 7-year survival in dialysis patients after controlling for other risk factors. STUDY DESIGN Prospective single-cohort study. SETTING & PARTICIPANTS 145 prevalent dialysis patients from 2 units in London, UK, were followed up for 64.3 ± 27.4 months and censored at the time of change to a different treatment. PREDICTORS Cognitive impairment, defined as performance 1 standard deviation less than normative values on 2 or more cognitive tests within a neurocognitive battery assessing attention/concentration, memory, and psychomotor function domains. Depression, quality-of-life, and clinical measures also were obtained. OUTCOMES & MEASUREMENTS All-cause mortality was the primary outcome. Cox proportional hazard models were used to assess the contribution of demographics and clinical and psychological measures and cognitive impairment to mortality. RESULTS 98 (67.6%) patients were cognitively impaired at baseline. At follow-up, 56 (38.6%) patients had died, 29 of cardiac causes. Unadjusted Kaplan-Meier analysis showed higher mortality in cognitively impaired patients, in whom 7-year survival was 49% versus 83.2% in those with no cognitive impairment (P < 0.001). Mortality risk associated with cognitive impairment remained significant in adjusted analysis controlling for sociodemographic, clinical, and psychological factors (adjusted HR, 2.53; 95% CI, 1.03-6.22; P = 0.04). LIMITATIONS Small sample size and number of events. CONCLUSIONS Cognitive impairment is an independent predictor of mortality in dialysis patients. Although the implications of early recognition and treatment of cognitive impairment for clinical outcomes are unclear, these results suggest that patient management protocols should attempt to ensure prevention of cognitive decline in addition to managing coexisting medical conditions.

Statistical guidelines for studies of the theory of reasoned action and the theory of planned behaviour

Abstract We identify potential problems in the statistical analysis of social cognition model data, with special emphasis on the theories of reasoned action (TRA) and planned behaviour (TPB). Some statistical guidelines are presented for empirical studies of the TRA and the TPB based upon multiple linear regression and structural equation modelling (SEM). If the model is tested using multiple regression, the assumptions of this technique must be considered and variables transformed if necessary. Adjusted R2 (not R2) should be used as a measure of explained variance and semipartial correlations are useful in assessing each components unique contribution to explained variance. R2 is not an indicator of model adequacy and residuals should be examined. Expectancy-value variables that are the product of expectancy and value measures represent the interaction term in a multiple regression and should not be used. SEM approaches make explicit the assumptions of unidimensionality of constructs in the TRA/TPB, assumptions that might usefully be challenged by competing models with multidimensional constructs. Finally, statistical power and sample size should be considered for both approaches. Inattention to any of these aspects of analysis threatens the validity of TRA/TPB research.

Time orientation and health-related behaviour: Measurement in general population samples

Research on health behaviour and time orientation has been hindered by a lack of consensus about appropriate measurement. Study 1 assessed the reliability of the Consideration of Future Consequences Scale (CFC) and the Zimbardo Time Perspective Inventory (ZTPI) in a general population sample (n = 300). Although more reliable, the CFC was less readable. Study 2 assessed the validity of a shortened ZTPI, measuring future and present orientation, and the full CFC. The measures had good discrimination to distinguish interpersonal differences. Construct validity of present, but not future, orientation as measured by the ZTPI, was evidenced by its mediation of the association between socioeconomic status and expectations of participating in diabetes screening. The CFC mediated this relationship more weakly. Further investigation of present orientation in understanding health-related behaviour is warranted.

The psychological impact of human papillomavirus testing in women with borderline or mildly dyskaryotic cervical smear test results: 6-month follow-up

State anxiety (S-STAI-6), distress (GHQ-12), concern and quality of life (EuroQoL-EQ-5D) 6 months after human papillomavirus (HPV) testing in women with borderline or mildly dyskaryotic smear test results were assessed based on a prospective questionnaire study, with 6-month follow-up after the smear test result. Two centres participated in an English pilot study of HPV testing. Participants included two groups of women receiving abnormal smear test results: (tested for HPV and found to be (a) HPV positive (n=369) or (b) HPV negative (n=252)) and two groups not tested for HPV (those receiving (c) abnormal smear test results (n=102) or (d) normal smear test results (n=288)). There were no differences in anxiety, distress or health-related quality of life between the four study groups at 6 months. Levels of concern about the smear test result remained elevated in all groups receiving an abnormal smear test result, and were highest in the group untested for HPV. Predictors of concern across all groups receiving an abnormal smear test were perceived risk of developing cancer, being HPV positive or untested for HPV, sexual health worries and the smear being a womans first smear test. The raised anxiety and distress observed in women immediately after being informed of an abnormal smear test result and that they are HPV positive was no longer evident at 6 months. Concern about the smear test result was however still raised in these women and those who tested negative for HPV, and particularly among those who did not undergo HPV testing.

The Transplant Effects Questionnaire (TxEQ): The development of a questionnaire for assessing the multidimensional outcome of organ transplantation — example of end stage renal disease (ESRD)

OBJECTIVES: To develop a questionnaire to assess the responses of transplant recipients to the receipt of an organ, including their self-care behaviour. DESIGN: Following a literature review, open-ended interviews and a focus group, a transplant questionnaire was developed. Two studies (Study 1: N = 231, Study 2: N = 105) were conducted to evaluate its psychometric properties. METHODS: A pool of 51 items was derived from themes identified in published studies and from interviews and a focus group discussion with renal transplant recipients. These were constructed into a questionnaire and were then administered to two renal transplant out-patients populations. Item responses of study sample 1 were subjected to principal components analysis (PCA) using varimax rotation to examine the structure of responses. In order to investigate the stability of the factor structure found in Study 1, item responses of the second sample were subjected to confirmatory factor analysis (CFA) using structural equation modelling. RESULTS: PCA indicated six factors that accounted for 64.2% of the variance. With extraneous items omitted, the final questionnaire derived from Study 1 has 24 items clustered around five conceptual coherent factors: worry about transplant (22.1%), guilt regarding donor (11.9%), disclosure (9.58%), medication adherence (8.73%), and responsibility (6.63%). CFA on the final 24-item version of the TxEQ revealed that the resulting model was a good fit for the Study 2 data (RMSEA = 0.08, pclose =.005). CONCLUSIONS: The TxEQ has potential application as a measure in the area of transplantation research. CFA demonstrated that the factor structure of the TxEQ is consistent across different renal transplant out-patients populations. Further research is currently in progress to assess other groups of transplant recipients and to examine its relationship to other measures.

Acute neuropsychological changes in hemodialysis and peritoneal dialysis patients

This study examined the impact of different dialysis treatments on the neuropsychological (NP) functioning of 145 end-stage renal disease patients. Hemodialysis (HD) and peritoneal dialysis (PD) patients were administered an NP test battery and measures of mood on 2 consecutive days (pre- and 24 hr postdialysis). Biochemistry was assessed at each session. Results indicated significant improvements in NP functioning (attention, concentration, verbal and visual memory, and psychomotor speed) in HD patients 24 hr postdialysis. No such fluctuations were found in PD patients. Although biochemical changes were found in the HD patients at the same time points, these were not consistently related to the NP changes.

The expectancy-value muddle in the theory of planned behaviour - and some proposed solutions

The authors of the Theories of Reasoned Action and Planned Behaviour recommended a method for statistically analysing the relationships between beliefs and the Attitude, Subjective Norm, and Perceived Behavioural Control constructs. This method has been used in the overwhelming majority of studies using these theories. However, there is a growing awareness that this method yields statistically uninterpretable results (Evans, 1991). Despite this, the use of this method is continuing, as is uninformed interpretation of this problematic research literature. This is probably due to the lack of a simple account of where the problem lies, and the large number of alternatives available. This paper therefore summarizes the problem as simply as possible, gives consideration to the conclusions that can be validly drawn from studies that contain this problem, and critically reviews the many alternatives that have been proposed to address this problem. Different techniques are identified as being suitable, according to the purpose of the specific research project.