Perron et al. Respond

We appreciate the interest of Steensma et al. in the theoretical limitations of the period life table-life expectancy (PLT-LE) and their group’s initiative to understand these limitations when the PLT is applied to an HIV-positive subpopulation. Nevertheless, we would like to make a few clarifications regarding their reaction to our commentary.¹

First, we do not recommend discarding the PLT-LE altogether as their letter suggests. We did not question the use of the PLT to summarize general populations’ mortality experience, but rather questioned the recent trend toward using the PLT to summarize diseased subpopulations’ mortality experience.

Second, there are indeed many published methodological articles aiming to disentangle the forces that explain why the PLT-LE might differ from the cohort LE. To our knowledge, however, these publications are limited to using the PLT-LE to summarize the general population mortality experience. This includes the chapter by Guillot² to which Steensma et al. refer. The focus of our commentary is different. It is about what happens when such a method is applied to a diseased subpopulation instead of the general population: when the PLT is applied to a subgroup defined by a personal characteristic (a medical condition, for instance) that can be acquired at any age, affect mortality differently over time, and, in some cases, come and go over time for a certain individual. The cohort influence bias, heterogeneity bias, and, eventually, tempo effect bias described by Guillot will be of concern, just as in general populations’ PLT-LE. But other important issues arise. Guillot does not address them, and neither do any other existing article on period versus cohort LE.

Finally, it is specifically because the PLT-LE metric is of “vital importance for providing timely public health information for action,” as emphasized by Steensma et al., that it is also vital to recognize its limitations and potentially misleading interpretations. We reiterate that using the PLT-LE to summarize subpopulation mortality experience requires a clear statement on the real cohort to whom the measure is intended to apply and clear evidence that age at first occurrence and time since first occurrence have a trivial effect on the mortality of afflicted individuals. Without these, the PLT-LE of diseased subpopulations provides only abstract numbers that should not be considered accurate public health information for action.