In our present study a reduced odds of disability was associated with male sex and patients who had undergone PTCA. On the other hand we could show that increased odds of disability were associated with rising age, diabetes mellitus, hearing loss in both ears, CABG, heart failure and deficient nutritional status.
Our results confirmed associations of the demographic factors age and sex with disability after AMI demonstrated in previous studies. Studies including AMI-survivors found disability or loss of physical function was more frequent in women and older individuals irrespective of applied measure. Instruments utilized in past studies included the Disability Scale by Rosow and Breslau, the SF-12 and the HAQ-DI [18, 22, 23]. Studies examining associations between sex, age and ADL-disability in general elderly populations, confirm our results [8, 21]. Among these, results from a recent cross sectional study by Strobl et al. seem most comparable with our study, since it utilized data representative of our study region and age group from the KORA-Age cohort and applied the same instrument (HAQ-DI) to measure disability . It found increased odds of disability in women (OR: 2.49, 95% CI: 2.06-3.02) and increasing odds of disability with rising years of age (OR: 1.12, 95% CI: 1.11-1.14). These estimates resemble our findings. However, in contrast to our study Strobl et al. defined prevalent disability as a HAQ-DI score > 0 due to less frequent disability in the general population. Only 22.5% of the general elderly population scored higher than 0.49 (minor to severe disability) in the HAQ-DI. In our population of elderly AMI survivors, 52.9% scored higher than 0.49. Furthermore, proportions of women, who have higher probability of being disabled, are considerably higher in the general population (51.2%) than in AMI survivors (37.5%) .
In our study AMI treatments showed strong associations with ADL-disability. The increased odds of disability in patients with CABG may be related to the severity of the intervention itself, which initially affects the patients’ general health status negatively. As this study examined patients shortly after the AMI event, it cannot be excluded that CABG will prevent disability in the long term. In contrast, PTCA as a more gently reperfusion intervention was shown to be associated with a reduced odds of disability already at the time of hospital discharge. On the long term, however, Dodson et al. have not found any significant effect of PTCA on disability (measured by EuroQol-5D questionnaire) one year post-AMI in 2002 patients with AMI with a mean age of 59 years .
Heart failure emerged as the strongest cardiac indicator of prevalent ADL-disability. Previous research discovered a decline in ability to perform ADL among individuals with heart failure during the first year after AMI . The long-term prediction of disability assessed with patients from the Framingham disability study, identified heart insufficiency to be predictive of disability in women . However, when interpreting the above it must be considered that among the discussed studies none applied the HAQ-DI to assess disability.
Our finding that AMI patients with diabetes were more likely to be disabled is in line with results of studies on the older aged general population. For instance, Gregg et al. found that diabetes was associated with a 2- to 3-fold increased risk of physical disability, accessed by physical performance tests and self-rated ability to walk ¼ mile, climb 10 steps and do housework, in a sample of 6,588 U.S. adults aged 60 years or older .
The strong associations of hearing loss in both ears and nutritional status are noteworthy as they have not been taken into account in available studies examining disability within the population of AMI patients. In studies on the general elderly population, hearing impairment was found to be a significant predictor of disability [9, 47]. Strobl et al. investigated determinants of disability within the framework of KORA-Age and found sufficiently nourished individuals, as defined by the SCREEN malnutrition score, to have significantly reduced odds of any ADL-disability as assessed by the HAQ-DI (OR: 0.93, 95% CI: 0.92-1.56). However, as discussed above, since Strobl et al. defined prevalent disability as a HAQ-DI score > 0 due to much more infrequent disability in the general population, coefficients of the regression models are not directly comparable with our present study . Furthermore, a recent review of literature on the association between nutrition and mobility in elderly people suggested that low micronutrients correlate with mobility disability . While this study applies measures so different that comparability to our study remains questionable, the congruence of laboratory-based and questionnaire-based results remains noteworthy. These findings suggest a further examination of potential causal effects nutritional deficiency and hearing loss may have on disability in elderly AMI-survivors. Furthermore adjustment for these factors may be necessary to avoid confounding in future studies.
Sample characteristics and associations found in the bi-variate analysis are mostly confirmatory of observations made in past studies [21–23, 18, 13, 17]. A few non-significant findings of the bi-variate baseline analysis may well have clinical relevance. For instance, participants with STEMI showed slightly reduced odds of disability. One possible interpretation is the fact that patients with STEMI may be diagnosed earlier and this leads to earlier access to necessary clinical interventions which reduce further damage of the heart muscle and consequently may prevent disability.
Strengths and limitations
To our knowledge, this study is the first which explores factors associated with disability in patients with AMI aged 65 or older. A large number of different variables including clinical characteristics and cardiac treatment could be tested for their association with disability not considered in any of the related studies to date. As part of the randomized KORINNA-study, instruments were carefully selected and standardized assessments were performed. The covariate structure of the finalized regression model was confirmed to be a very stable model of the most dominant influence factors on the outcome in different selection procedures.
Findings relevant to future research on disability in AMI-survivors were made. Due to the large effect sizes it appears reasonable that adjustment for hearing loss and nutritional deficiency may be relevant to the estimation of other predictors in clinical research. Finally, our study raises the question whether differences in disability prevalence between elderly AMI-survivors and other populations are systematic or arbitrary. If disability differentials are caused by random differences in certain population characteristics such as hearing loss, adjustment is sufficient. If however, the structure in causal relationships between factors predictive of disability differs from other populations, the development of a disease-specific theoretical framework will be essential.
However, some methodological limitations should be considered. Since the analysis was conducted with data from a study not originally designed for this research question, sample size did not allow consideration of potentially relevant factors too infrequent (n < 40) for the execution of a stable multivariate analysis. Examples are the effects of social support and moderate to severe depressive symptoms on disability found in the bi-variate analyses. Furthermore, the exclusion of relevant characteristics with too low frequencies resulted in the unfeasibility of structural equation modeling. Thus, potentially relevant causal relationships between covariates could not be considered. In addition, effects may be overestimated. The ratio of predictors over outcome events in the full model is 1/11.2. Application of the rule of thumb proposed by Steyerberg and Harrell, suggests that with a ratio over 1/10 shrinkage while advisable is not absolutely necessary . Finally, due to the cross sectional design, actual causations between disability and covariates cannot be proven in this study.