This paper suggests incorporating the new family of ‘healthy lifespan inequality’ indicators into the list of measures that are regularly reported to monitor population health. To the extent that one is interested in measuring (1) how long individuals are expected to live, (2) how much of that time is spent in good health, and (3) how heterogeneous the distributions of lifespans can be, it is only natural to be also interested in the heterogeneity of the distribution of healthy lifespans. Applying the traditional LE, HE, LI together with the new HLI measures across education groups and sexes in contemporary Spain, several interesting patterns can be identified. In line with previous studies [12,13,14], we observe that education differences in HE are substantial and greatly exceed differences in LE, a well-known pattern also observed across countries (i.e. international differences in LE being smaller than international differences in HE ). The female advantage in LE disappears when considering HE indicators, overall and across education groups. The observed social patterning in LI coincides with recent studies suggesting that socio-economically advantaged groups tend to exhibit lower levels of lifespan inequality, and that the latter is systematically higher among men [4, 15,16,17,18,19,20].
The observed levels of LI dwarf when compared to their HLI counterparts: the new indicators suggest that the variability in the ages at which daily activity limitations start can be substantially larger than the variability in the ages at which individuals die (increases by a factor of 1.5). Our findings indicate that the levels of HLI are slightly higher for women – a result that coheres with the mortality advantage and morbidity disadvantage of women vis-à-vis men . Importantly, HLI increases with decreasing educational attainment, both for women and for men. Recently, it has been suggested that because low-SES individuals tend to live shorter lives and face greater uncertainty in the age at which they will die than high-SES individuals, they are exposed to a ‘double burden of inequality’ [4, 18]. In light of our findings, where low-educated individuals are additionally expected to have shorter healthy lifespans and face greater uncertainty in the age at which they will start experiencing physical limitations, one could argue they are indeed exposed to a ‘quadruple burden of inequality’.
To our knowledge, this is the first study proposing measures to assess the extent to which the years spent in good health are (un)equally distributed across individuals. Previous analyses have investigated differences in health expectancies across subnational populations (e.g. comparing HE across women and men, socio-economic groups and/or racial groups [12,13,14]). While somewhat related, the two approaches are fundamentally different: the latter compares levels of HE across a closed list of pre-specified groups, and the former investigates variability in healthy lifespan distributions across individuals. These approaches echo a two-sided debate of the early 2000s, when the World Health Organization (2000) recommended going beyond group-based mean comparisons [22, 23] and incorporate individual-based data in the analysis of health inequalities [24, 25]. Since the turn of the century, the concept of inter-individual inequality has gained traction, thus favoring the spread of a bourgeoning literature on lifespan inequality. Yet, these contributions are based, implicitly or explicitly, on the assumption that ‘longer lives are normatively preferable to shorter ones’ – a gross simplification that greatly facilitates the construction of the corresponding LI measures but that might be hardly tenable in many circumstances. Confronted with the choice between ‘a prolonged yet unhealthy life’ and ‘a shorter but fully healthy life’, it is not clear that the former would be universally chosen in favor of the latter [26, 27]. The inclusion of any year of life irrespective of the health conditions in which that year is lived into standard lifespan inequality measures can muddy the waters in regard to the interpretation of their values. These conceptual problems are sidestepped by our HLI measures, which only take into consideration the variability in healthy lifespans.
In this paper we have focused our attention on the distribution of healthy lifespans, but it could be equally reasonable to look at the distribution of unhealthy lifespans. Both approaches are interesting in their own right, and it is not a priori clear what the relationship between the two approaches could be. One could easily imagine hypothetical scenarios with low levels of healthy lifespan inequality (i.e., everyone enjoying the same number of healthy years) but high levels of “unhealthy lifespan inequality” (i.e., large differences in the number of years individuals spend in bad health) or vice versa. These extremely interesting questions will be investigated in future research.
The new indices hold promise to be an important complement to traditional LE, HE and LI measures, which, on their own, do not explain the whole story and might lead to the elaboration of unfair or misinformed policies. Inter alia, HLI indicators can be crucial for the design of equitable pension schemes and retirement policies that are sensitive to the underlying heterogeneity in the population, and for the public provision of medical care (especially at advanced ages). From a public health policy perspective, larger HLI might be indicative of a worsening state of affairs across or within socially relevant groups – a cause of legitimate ethical concern, especially when social patterning in health is attributable to preventable causes. Finally, studying healthy lifespan variation can enrich the longstanding ‘compression/expansion of morbidity’ debate, which aims at understanding whether prevalence of morbid conditions accord with or diverge from trends in mortality [28, 29]. Since its inception more than 30 years ago, the contrasting hypotheses in this debate have been mostly tested by comparing LE with HE indicators (i.e., inspecting trends in average years of life vis-à-vis trends in average years in good health [1, 2, 7]). Yet, the original formulation of the compression of morbidity hypothesis —which was stated in terms like ‘compression into a shorter span between the age of disability onset and death’, or ‘rectangularization of the morbidity curve’ – naturally lends itself into an inspection of HLI trends to test its validity. There are compelling reasons to believe that going beyond such ‘averages comparison’ by taking into consideration the entire distribution of ages at which death and diseases occur can throw considerable light into a long-lasting debate with crucial implications for understanding the development of human health and the performance of health systems.
This study has several limitations. First, our method to estimate healthy lifespan distributions is based on simplistic and somewhat unrealistic assumptions. Following our approach (in which the individuals of a fictitious cohort are subject to the current mortality and morbidity conditions throughout their lifetimes), we are implicitly assuming that there is no possibility of recovery from disease/disability, and that the risk of mortality is the same for healthy and unhealthy individuals. These are exactly the same limitations assailing the Sullivan method  that is commonly used to estimate HE indicators. Despite these shortcomings, the method has succeeded in becoming the workhorse of ‘health expectancy studies’ owing to its simplicity and applicability in a wide range of geographical, temporal and conceptual settings . Some studies have shown that, under mild regularity conditions, Sullivan’s method is generally acceptable for monitoring long-term trends in HE . In future research, it would be desirable to apply other methods to estimate healthy lifespan distributions more realistically. Some of these methods rely on longitudinal datasets that allow tracking individuals’ health trajectories over time and calculate transition probabilities across several health states (e.g. multistate life table or Markov chain techniques ). Unfortunately, the relative scarcity of longitudinal datasets across space and time limits the empirical applicability of such sophisticated methods.
Second, our approach to measure less-than-good health could be criticized on grounds of arbitrariness. ‘Health’ is a multidimensional and fuzzy concept whose measurement can be operationalized using other possible health outcomes. Additionally, there are several techniques to measure how such outcomes contribute to healthy/unhealthy life-years, using either dichotomous, ordinal or continuous scales [1, 2]. These well-known challenges have already been encountered by previous attempts to measure HE and compare its values across countries/over time  so they are not exclusive to the measurement of HLI indicators alone. To address them, it is important to use consistent definitions when making comparisons and, if feasible, use different health indicators to check the robustness of results. In our setting, the use of alternative conceptualizations of the concept of ‘disability’ besides the standard GALI indicator (i.e. self-perceived health or the ability to perform certain physical tasks) does not alter the substantive findings of the paper (see Additional File 1: Tables S3 and S4).
Third, our analyses are restricted to the ages ranging between 35 and 85 because of data limitation constraints. In all likelihood, the differences between women and men and across education groups would be altered if one considered all the ages above 35. Spain is among the world’s most longevous countries and many deaths and morbid conditions occur above age 85. Despite the limitations of our dataset, a clear education gradient emerges and the values of traditional LE, HE and LI indicators go in the expected direction. Thus, the quality of the dataset should be good enough to illustrate the usefulness of the new HLI indicators proposed in the paper.