Factors associated with reaching 90 years of age: a study of men born in 1913 in Gothenburg, Sweden


I.R. van Manen, Department of General Practice, Erasmus MC – University Medical Centre Rotterdam, Rotterdam, The Netherlands.
(fax: 010-704-4766; e-mail: ilse.v.manen@gmail.com).

With interest, we read the prospective study by Wilhelmsen et al. [1] about health factors associated with reaching old age. The conclusion of the study is that there are several modifiable lifestyle factors associated with reaching age 90 years in men. We think it is a unique prospective cohort study with a long follow-up and accurate described methods and results. However, we have some questions remaining.

First, the statistical results are well described and the authors declare that these results provide a support for public health policies. However, the area under the ROC curve (which predicts the probability to determine who will reach age 90 years) never is >0.604. Because of this, in addition with the cohort effect (already mentioned by the authors), it seems to us that the clinical relevance of these outcomes is less than claimed by the authors.

Second, it is known that when a large number of variables are investigated, there is a considerable risk of chance findings, the multiple testing effect. The study investigated many variables; however, in the discussion, this risk of chance findings is not considered.

Third, in the results, it is described that all variables that were statistically significant in the univariable regression analysis were included in a multivariable analyses. The authors mentioned that there is a trend of increasing obesity. Obesity is a well-known risk factor for numerous health conditions, cardiovascular diseases amongst others. We wonder why the analysis was not adjusted for BMI. How do the authors think this may influence the results?

Finally, we are also interested in the epidemiological numbers of the study population. How many men where actually born in 1913? Are there thoughts or hypothesis about those who died before the start of the study?

We are very interested in the authors’ thoughts on these comments and questions.

Conflict of interest statement

No conflict of interest was declared.