Subjective health is likely to confound the association between will-to-live and survival.

Subjective health is likely to confound the association between will-to-live and survival SIR—We enjoyed reading the inspiring study on will-to-live and survival in older home-dwelling people [1]. The association between will-to-live and survival in these patients looks striking. However, and although the association remained statistically significant after multivariate adjustments , will-to-live may be a risk marker rather than a risk factor. It is indeed plausible that 'objective health' determines subjective health [2], which in turn determines will-to-live. It is also very likely that objective health largely determines survival. The association between will-to-live and survival could thus be explained by the underlying objective health. Objective health cannot be directly measured and accounted for in multivariate analysis. The fact that diagnoses , the Charlson comorbidity score and subjective health were dependent on each other supports the hypothesis that they all relate to objective health. We understand that only one of them can be included in the multivariable model to avoid multicollinearity, but the Charlson score may not have been the best choice. The variation of subjective health according to will-to-live observed in the study was larger and more statistically significant than the variation of the Charlson score (Table I) [1]. Subjective health is, therefore, likely to be a more powerful confounding factor than the Charlson score. Moreover, previous studies have shown that subjective health predicts mortality [2]. We, therefore, hypothesize that and are keen to know whether the association between will-to-live and survival decreases and perhaps even vanishes if the multivariable model is adjusted for subjective health rather than for the Charlson score. Pitkala KH. Will-to-live and survival in a 10-year follow-up among older people. Age Ageing 2012; 41: 789–94. 2. Jylha M. What is self-rated health and why does it predict mortality? Towards a unified conceptual model. Subjective health, will-to-live and survival SIR—We read with great interest Prof. Steichen's et al. comments [1] on our study regarding will to live (WTL) and survival [2]. The author hypothesised that our concept WTL might be a more powerful confounding factor than the Charlson comorbidity index, and he suggested a multi-variate model adjusted for subjective health instead of the Charlson index. Therefore, we reanalysed our data with Cox regression analysis adjusted for age, gender, education, subjective health, smoking and Mini-Mental State Examination score. We omitted depression, which may be a dependent variable with subjective health. In this analysis, the predictive value of WTL diminished …