Hong Kong’s record longevity hides a birthplace gap in later-life survival

Hong Kong routinely posts some of the highest life expectancy figures on the planet, remaining among the longest in high-income settings in recent decades and turning its 7.4 million residents into a demographic marvel. But a new analysis of official mortality and census records suggests that this celebrated longevity is not shared evenly among the people who live there. Adults born in Hong Kong, the study finds, have consistently enjoyed more expected remaining years of life than residents born across the border in Mainland China, Macao, or Taiwan, and the difference is concentrated almost entirely in later-life deaths from pneumonia and cancers. The gap widened for years, then narrowed sharply in 2021, in a shift that reflects both improvement in one population and deterioration in the other.

The study, published in The Lancet Regional Health Western Pacific, was conducted by a research team led by Chen together with Bishai and Canudas-Romo. The team constructed sex-specific, single-year-of-age period life tables for four benchmark census years, 2006, 2011, 2016, and 2021, using death-registration microdata and resident population denominators from Hong Kong’s Census and Statistics Department. Their headline measure was adult remaining life expectancy at age 15, written e15, which represents the expected number of further years of life for someone who has already survived to exact age 15 under the mortality conditions observed in a given calendar year. Because it excludes deaths in childhood and adolescence, e15 cannot be compared directly with life expectancy at birth, but it isolates the adult mortality experience that shapes how long residents live once they reach working age.

The numbers tell a striking story. Among women, the Hong Kong-born advantage in e15 stood at 2.6 years in 2006, with Hong Kong-born women expecting 72.2 further years against 69.6 years for MMT-born women. The gap widened to 3.3 years in 2011, eased to 3.1 years in 2016, and then contracted abruptly to just 1.3 years in 2021, when the figures were 72.8 and 71.5 years respectively. Men followed the same arc with larger swings: the male gap grew from 3.4 years in 2006 to 4.4 years in 2011 and 4.8 years in 2016, before falling to 2.2 years in 2021, with Hong Kong-born men expecting 68.7 further years compared with 66.5 years for MMT-born men. Women outlived men in both birthplace groups throughout.

To work out where in the age structure these differences arose, the researchers applied Arriaga’s additive decomposition method, which breaks a difference in life expectancy into the contributions of individual ages and, jointly, individual causes of death. The result was unambiguous: the gap was an affair of later life. Among women, ages 85 and above were the largest and most consistent contributor, accounting for between 0.74 and 1.53 years, or roughly 46 to 57 percent of the female gap in any given year, with ages 65 to 84 adding a further 36 to 43 percent. Among men, ages 65 to 84 dominated, contributing 1.19 to 2.17 years, or about 45 to 54 percent of the male gap, followed by the 85-plus range. Younger age groups mattered comparatively little in either sex.

The cause decomposition sharpened the picture further. Pneumonia and cancers made the largest positive contributions to the Hong Kong-born advantage. Pneumonia alone contributed 0.43 to 0.62 years among women and 0.55 to 0.95 years among men across the study years, while other cancers added 0.16 to 0.46 years for women and 0.27 to 0.71 years for men. Lung cancer was a consistent contributor, particularly among men at 0.43 to 0.64 years, and liver cancer added 0.22 to 0.41 years among men. In 2021 the gap became remarkably concentrated: pneumonia accounted for roughly 34 percent of the difference in both sexes, and pneumonia plus all cancer categories together explained 0.84 years, or 65 percent of the 1.3-year female gap, and 1.91 years, or 88 percent of the 2.2-year male gap. The authors note that historical tobacco exposure is one plausible contributor to the lung cancer findings, since male smoking prevalence was about 26 percent in Hong Kong in the early 2000s against roughly 57 percent among men in Mainland China in 2002, but they stress that smoking was not measured in the study population and the comparison supports only a hypothesis.

Within the MMT-born population itself, the study uncovered a second, more puzzling layer of stratification by length of stay. Recent arrivals with fewer than 10 years in Hong Kong showed higher e15 than longer-settled residents throughout 2011 to 2021 among women, although the difference narrowed from 5.0 years in 2011 and 5.6 years in 2016 to 2.6 years in 2021. Men showed the same pattern in 2011 and 2016, with gaps of 6.4 and 6.5 years, but the direction reversed in 2021, when e15 among short-stay men was 3.1 years lower than among long-stay men. That 2021 female contrast proved fragile: when the researchers repeated the comparison with a seven-year cutoff, the female difference was statistically compatible with no difference at all, while the male reversal persisted. The authors warn that length of stay may capture differences in age at migration, cohort composition, and socioeconomic selection rather than any effect of residence duration itself.

Interpreting these patterns requires care, and the study is emphatically descriptive rather than causal. One candidate mechanism is the so-called salmon bias, in which migrants return to their place of origin near the end of life and their deaths escape Hong Kong’s registration system, potentially inflating the apparent longevity of the mobile population. Cross-boundary living is substantial: official estimates indicate 537,200 Hong Kong residents were usually staying in Guangdong in mid-2024, about 7.1 percent of the population, including 99,600 aged 65 or above. The authors suggest that the disappearance of the older-age, short-stay male advantage in 2021, when boundary controls and quarantine policies reduced cross-boundary movement to extremely low levels, is compatible with the possibility that reduced end-of-life relocation revealed deaths that might otherwise have gone unregistered, although they cannot test this directly and treat it as a measurement priority rather than an explanation.

The sharp narrowing of the birthplace gap between 2016 and 2021 reflected two simultaneous movements rather than uniform progress. e15 declined by roughly one year among Hong Kong-born women and men while increasing by 0.77 years among MMT-born women and 1.58 years among MMT-born men, with the principal contributions arising at older ages. Population composition may have shifted: Hong Kong experienced substantial outflow around 2021 through the comparatively highly educated British National Overseas route, while selective talent-oriented admission pathways expanded, with places allocated under the Quality Migrant Admission Scheme rising from 411 in 2017 to 2,004 in 2021. Pandemic disruptions are another candidate, but the authors caution against reading the rising prominence of pneumonia as a direct COVID-19 effect, since directly attributed COVID-19 deaths were coded outside the pneumonia category and accounted for only 125 of 50,653 deaths in 2020 and 70 of 51,536 in 2021. Indirect effects from disrupted diagnosis, reduced screening, and interrupted chronic-disease care remain possible but unquantified.

The technical scaffolding behind the estimates is considerable. Grouped population denominators, available only in five-year age bands up to an open 85-plus category, were disaggregated to single ages using the penalized composite link model, and uncertainty was propagated through 5,000-replicate Poisson parametric bootstraps that resampled deaths in every age-cause cell. Sensitivity analyses using alternative starting ages, restricted age ranges, and abridged life tables built directly from the original grouped denominators confirmed the main birthplace findings, with the largest difference from the benchmark estimates just 0.121 years. The authors acknowledge that the 85-plus contributions are model-dependent, that underlying-cause certification is most uncertain at advanced ages because of multimorbidity, and that visa categories and individual migration histories were unavailable, preventing origin-specific life tables even though Mainland China-born residents accounted for 98.5 percent of MMT-born deaths. What the study establishes is that Hong Kong’s aggregate longevity conceals structured differences in later-life mortality by birthplace and settlement history, concentrated in pneumonia and cancers. The findings argue for migration-sensitive mortality surveillance and equity-focused evaluation of pneumonia prevention, cancer control, and continuity of care across birthplace groups, especially in cross-boundary healthcare settings, and for linked individual data on migration, socioeconomic position, and residence history to finally separate health selection from cumulative disadvantage and measurement mismatch.

Subject of Research: Birthplace differences in adult remaining life expectancy in Hong Kong, decomposed by age and cause of death

Article Title: Adult remaining life expectancy among Hong Kong-born and Mainland China/Macao/Taiwan-born residents in Hong Kong

Article References: Chen, X., Canudas-Romo, V., & Bishai, D. M. (2026). Adult remaining life expectancy among Hong Kong-born and Mainland China/Macao/Taiwan-born residents in Hong Kong. The Lancet Regional Health – Western Pacific, 75, Article 101998. https://doi.org/10.1016/j.lanwpc.2026.101998

Image Credits: AI Generated

DOI: 10.1016/j.lanwpc.2026.101998

Keywords: life expectancy, Hong Kong, migration, mortality, pneumonia, cancer, demography, salmon bias, length of stay, Arriaga decomposition, healthy ageing, Adult