|dc.description.abstract||Depression is a substantial psychopathology encountered in the dialysis population yet its association with potentially modifiable psychological antecedents are not well known. Of these potential antecedents, individual’s perception of their condition are likely to play an important role in how they adjust to their illness (Leventhal, Brissette, & Leventhal, 2003). The Common Sense Model suggests that illness representations guide the self-regulation of illness (Leventhal, Meyer, & Nerenz, 1980; Leventhal, Nerenz, & Steele, 1984). The model posits that the interpretation of illness (illness perceptions) influence the response and procedures adopted in order to regulate the illness threat. The overarching aim of the work here is to examine whether illness perceptions predict depression and its trajectory in End-Stage Renal Disease (ESRD) patients, and to establish if depression and illness perceptions are associated with adverse clinical outcomes in these patients.
In order to achieve these aims it was first important to establish how best to assess depression and illness representations in the context of ESRD. A pilot study investigated whether the Beck Depression Inventory (BDI) and the Revised Illness Perception Questionnaire (IPQ-R) could be administered to haemodialysis patients (HD) while actively on dialysis. Patients completed the BDI and IPQ-R while on-dialysis and again at a time when off-dialysis (n=40). Level of agreement revealed no discernable difference between BDI and IPQ-R scores across the two conditions, although there was a slight bias with regards to scoring on somatic items of the BDI while on-dialysis. Given these data, on-dialysis assessments were employed in the studies reported. Furthermore the BDI was compared against a diagnostic standard for Major Depressive Disorder (MDD) in order to define an adjusted BDI cut-off score that would indicate potential depressive cases. The data revealed that a BDI≥16 had optimal sensitivity and specificity for MDD. This cut-off score was employed to define patients with “probable” depression.
The factor structure of the BDI was the focus in the following chapter. BDI data from two larger studies (reported later in the thesis) were pooled in order to conduct confirmatory factor analysis, testing several proposed structures of the BDI. The analysis revealed that two and three factor solutions had relatively poor fit to the data. A relatively novel bi-factor model proposed by Ward (2006) had the best fit. In this model there is a general depression factor that loaded onto all of the 21 BDI items, and two smaller orthogonal cognitive and somatic factors. These factors collectively explained 91% of the total variance in BDI-II total scores, suggesting that the BDI provides a good overall measure of global depressive symptoms.
The first study to examine the association between illness representations and depression was a cross-sectional study of established HD patients (n=215). Nearly 30% of the sample were depressed (BDI≥16), highlighting the extent of depressive symptoms in this patient group. Significant differences between depressed and non-depressed patients with regards to illness perceptions were evident. In logistic regression illness coherence, perceived consequences and treatment control perceptions predicted depression. Interestingly clinical variables including co-morbidity were unrelated to depression. This suggests that it is not disease severity or extra-renal co-morbidity per se that are vulnerabilities for depression, rather it is the interpretation of the disease that appears to be important.
The proceeding chapter extended this cross-sectional investigation by examining the trajectory of depression (i.e. change in depression) over the first year of dialysis therapy in relation to illness representations. An incident cohort of dialysis patients (n=160) were seen at a point soon after dialysis initiation and followed up 6 and 12 months thereafter. In particular, differences between patients who start dialysis via planned route (i.e. those with progressive renal failure who had been “worked-up” to dialysis) vs. those who started dialysis suddenly (unplanned starters) were sought. Unplanned starters were more depressed than the planned patients and held different illness perceptions. Structural equation modelling of the baseline data revealed that illness perceptions predicted depression, and that path to dialysis had an indirect effect on depression as mediated through illness perceptions. Over time, depression and illness perceptions appeared to remain relatively stable although there was some evidence of a non-linear decline in depression scores over the follow-up period. In addition, illness identity decreased over time, while illness coherence (understanding) increased. Clinical and demographic factors were not associated with the trajectory of depression as assessed using Latent Growth Models. However several illness perceptions were associated with a change in depression over time, suggesting that patient’s illness representations assist in the regulation (or under-regulation) of mood.
The first of two clinical oriented chapters examined the utility of illness representations in explaining fluid non-adherent behaviour. HD patients were categorised as either fluid adherent or non-adherent based upon Inter-dialytic Weight Gain (IDWG). Patients in the upper quartile of percent weight gain were defined as non-adherent (IDWG≥3.21% dry weight). The data revealed that non-adherent patients had lower timeline perceptions as compared to adherent patients. Logistic regression models were evaluated in order to identify predictors of fluid non-adherence. After several demographic and clinical variables had been controlled, lower consequence perceptions predicted non-adherence. This data points to the utility of understanding dialysis patient’s personal illness representations in relation to maladaptive health care behaviour.
Finally, the potential association between depression, illness representations and short term survival in incident dialysis patients was evaluated. Patients were followed up for a mean of 545 (±271) days in which there were 27 deaths (16.9%). Patients were censored if they were lost to follow-up, transplanted or recovered renal function. In Cox survival models after controlling for several co-variates including co-morbidity, depression significantly predicted mortality. Furthermore, survival models including illness perceptions revealed that treatment control perceptions were also predictive of mortality. These results suggest that depression and beliefs surrounding treatment control contribute to the survival of dialysis patients. Possible explanations regarding these associations are presented.
In conclusion the empirical investigations offered here support the thesis that illness perceptions predict depression in dialysis patients. Moreover there is evidence that illness representations are associated with maladaptive health behaviour (non-adherence) in dialysis patients. Depression and illness representations also predict short-term survival in incident patients after adjusting for important co-variates. Recent studies have shown that altering maladaptive illness perceptions via psychological intervention can have a positive influence upon outcomes (Petrie, Cameron, Ellis, Buick, & Weinman, 2002). Given the evidence presented in this thesis, testing interventions that target maladaptive illness representations in order to improve clinical and psychological outcomes seem highly relevant in this setting.||en_US