How HALE Turns Health Loss into an Expectancy
The subject of this chapter is the WHO measure Healthy Life Expectancy, or HALE, and the specific problem is how a population can convert survival and health loss into one expected duration. The useful result is not a memorized formula. It is the ability to see what each term contributes, why HALE is a weighted summary rather than a count of diagnosis-free years, and where the data behind the calculation can make comparisons fragile.
From life-table years to healthy years
HALE starts with a question about remaining time: at a given age, how many years can a person expect to live in “full health,” after accounting for years lived in less than full health because of disease or injury? The word “expect” signals a population average, not a promise about an individual life. The word “full” signals a health definition that must be operationalized. HALE does not simply read a healthy-year total from a medical record; it constructs one by combining a life table with information about functional-health loss.
A life table supplies the mortality side of the calculation. It describes how many people remain alive at each age and how many years the life-table population lives across age intervals. Let lₓ represent the number of survivors at age x, and let Lₓ represent the total years lived by that life-table population between age x and x+5. The exact notation is less important than the roles: lₓ identifies the people whose remaining duration is being summarized, while Lₓ measures the time that the life-table population actually contributes across the next interval.
The health side enters through Dₓ, the equivalent lost healthy-year fraction for that interval. It expresses how much of the years lived in the interval is treated as lost from full health because of the health conditions and functional losses included in the measure. The WHO metadata defines the interval’s years lived without disability as:
YWDₓ = Lₓ × (1 − Dₓ)
If the lost-health fraction is small, most of the life-table years remain in the healthy-year total. If it is larger, fewer equivalent healthy years are credited for the same amount of time lived. This is the key difference between a weighted health-loss measure and a binary disease-free measure: time can contribute partially rather than being classified only as entirely healthy or entirely unhealthy.
The calculation then adds those adjusted years across the remaining age intervals. In simplified form:
HALEₓ = (sum of YWDᵢ from age x to the final life-table interval) ÷ lₓ
The numerator is the total number of equivalent healthy years lived by the life-table population after health loss has been applied. The denominator converts that population total into an average for a person who has reached age x. If the calculation is made at birth, the starting survivor population is the life-table population at birth. If it is made at age 60, it concerns survivors at 60. That is why an age label is part of the result rather than an optional annotation.
The denominator also prevents a subtle mistake: the numerator is not the number of healthy years one particular person has accumulated. It is the healthy-time contribution of an entire life-table population. Dividing by the survivors at the starting age turns that aggregate into an average remaining duration. If the starting age changes, both the people represented and the remaining intervals change. A value at age 60 is therefore not obtained simply by subtracting the value at birth from another total; it is a new conditional summary for a different survivor group.
Likewise, the loss fraction applies to years lived in an age interval, not to a single diagnosis in isolation. The health information may describe the prevalence or functional consequences of several conditions, and the method converts that information into equivalent lost healthy time for the interval. The result retains the fact that health affects time, but it does not retain a complete personal chronology of which condition appeared first, how it interacted with another, or whether a limitation later improved. Those details may matter for care or causal research even when they are intentionally compressed in a population indicator.
The sequence can be visualized as a chain:
Several consequences follow from this structure.
First, HALE is not life expectancy with a few diagnosed diseases subtracted. Life expectancy is built from survival; HALE retains that survival foundation but adjusts the years lived according to a measure of functional-health loss. A population can have the same mortality pattern as another and still receive a different HALE if the distribution or valuation of nonfatal health loss differs. Conversely, a change in mortality can affect the available years even if the health-loss information stays the same.
Second, HALE is not a direct individual prognosis. The life-table population is a statistical construction, and the health-loss fraction summarizes information across people and health states. A person’s actual future may include a pattern of recovery, chronic illness, or functional change that the aggregate does not represent in detail. HALE is valuable for describing population health; it should not be read as an appointment with a fixed number of personal healthy years.
Third, the measure is broad by design. The WHO metadata describes HALE as a summary across all causes combined. That breadth helps answer a population-level question about overall health rather than the narrower question of how one disease affects life. It also creates compression. Once different diseases, injuries, and functional losses are represented in one weighted total, the final number cannot reveal every pathway that produced it. A change in HALE is a signal that the combined summary changed; it is not, by itself, a diagnosis of which cause changed or why.
That compression can be useful when a decision concerns the overall burden of health loss. It can also conceal offsetting movements. One group of conditions may become less disabling while another becomes more common, leaving the combined value nearly unchanged. The reverse can happen as well: a small change in the summary may reflect important changes in several underlying states that move in opposite directions. HALE is a dashboard-level measure, not a causal decomposition. To explain a change, analysts need the component health data and the mortality context rather than the summary alone.
The weighting is therefore not a cosmetic adjustment. It is the mechanism that gives HALE its meaning. Without Dₓ, the calculation would reduce to ordinary life-table years divided by survivors: a life expectancy. Without the life table, health-loss prevalence would not become an expected remaining duration. HALE exists in the interaction between the two.
What enters the calculation—and what can distort it
The formula looks compact because it compresses several evidence streams. The WHO metadata identifies sufficiently complete vital-registration systems that record deaths as preferred data for estimating all-cause death rates. It also identifies national health examination surveys as a source for the prevalence of diseases, injuries, and disabilities. Sample registration systems and verbal autopsy are listed as other possible sources. These inputs do different jobs: mortality data shape the survival framework, while health information supplies the loss applied to years lived.
This division matters when interpreting a result. Mortality statistics may be strong while nonfatal health information is incomplete, or survey information may be extensive while death registration is less reliable. A single published HALE value can hide that asymmetry. The output is one number, but its evidential foundation has at least two major parts, each with its own coverage and measurement error.
The two streams also operate on different observational rhythms. Death registration usually records an event that has occurred, whereas a survey estimates health states from the people and questions included at a particular time. Turning those inputs into a common age-specific summary requires assumptions about how the observed health distribution relates to years lived. The calculation can be consistent without the underlying observations being equally rich in every age group or location.
The WHO metadata also allows HALE to be disaggregated by sex, location—including urban or rural areas and major regions—and socioeconomic characteristics such as education or wealth quintile. Disaggregation makes the summary more useful because an overall average can conceal differences among groups. But it also increases the need to inspect denominators and data quality. A subgroup estimate is not just a smaller version of a national estimate. It depends on whether the mortality and health-loss observations are sufficiently complete and comparable for that subgroup. A difference between two groups can be real, measurement-related, or a mixture of both; the indicator alone cannot decide which explanation is correct.
The first major limitation is missing reliable mortality and morbidity data, especially in low-income countries. If deaths are incompletely recorded, the life-table side can misstate survival. If disease, injury, or disability prevalence is poorly measured, the health-loss fraction can be wrong or unevenly estimated. The two weaknesses can interact: an apparently comparable HALE value may reflect different levels of observation rather than a clean difference in population health.
The second limitation concerns self-reported information. Health examination surveys may include reported symptoms, conditions, or functional limitations, but the WHO metadata warns about the comparability of self-reported data and the measurement of health-state preferences used for such reporting. People in different settings may interpret the same question differently, have different expectations about activity, or face different cultural and institutional contexts. That does not make self-report useless. It means that a response must be understood as an observation produced by a specific instrument and population, not as a transparent reading of an underlying biological quantity.
The third issue is valuation. To combine many kinds of health loss into one fraction, the method needs a way to represent differences in severity or functional consequence. The more a calculation depends on converting diverse states into equivalent healthy years, the more important it becomes to know how those equivalences were established. A weighted summary may preserve information that a yes-or-no threshold would discard, but the weights themselves are part of the measurement design. They should be treated as assumptions with a purpose, not as facts supplied by nature.
This is where the contrast with the French clinical study becomes useful. The clinical study defines a disease-free state through severe events recorded in hospital trajectories. HALE instead applies a health-loss fraction to life-table time across causes. The clinical measure can follow a clearly dated event in a patient record; HALE can express degrees of loss across a population summary. Neither observation rule is a universal substitute for the other. They answer different questions about the same broad concern: how much of life is lived in a defined state of health.
The difference also clarifies what “full health” does and does not mean in HALE. It is the endpoint of the measure’s health-loss accounting, not a guarantee that every credited year is free of any diagnosis or discomfort. A year can contribute to the healthy-year total even when the population includes health conditions, if the method represents the associated loss as a fraction rather than excluding the entire year. That is why calling HALE “years without disease” would be inaccurate. The method is designed to summarize functional health loss across all causes, not to certify the absence of illness.
A number can consequently be precise in arithmetic and uncertain in interpretation. The calculation may follow the stated formula exactly while the inputs remain affected by incomplete registration, non-comparable survey responses, or choices about health-state valuation. Precision answers whether the computation was carried out consistently. It does not answer whether the health state was observed equally well across populations or whether two estimates were built from equivalent evidence.
This distinction is especially important for trends. If HALE rises over time, the change may reflect longer survival, less functional-health loss, improved measurement, or some combination. The WHO metadata alone does not identify the direction or size of every possible bias. A responsible interpretation therefore reports the estimate with its age, population, data sources, and health-loss definition, rather than treating the number as a self-explanatory score.
HALE’s strength is its ability to put mortality and nonfatal health loss into one broad population measure. Its limitation is inseparable from that strength: combining many dimensions makes the result easier to summarize and harder to unpack. Once the construction is visible, the comparison with other healthspan measures becomes more honest. The question is not whether HALE is “the” healthy-life number. It is what a weighted, all-cause summary can show that a survey of activity limitation or a hospital record of severe events may show differently.
The next case makes that alternative concrete. A national hospital-discharge database can create a detailed clinical history for millions of people, but it also forces a new question: when the data come from hospitals, whose health is actually represented?