Background: Hypertension is a leading preventable risk factor of chronic disease and all-cause mortality. Housing is a fundamental social determinant of health. Yet, little is known about the impacts of liveable residential space and density on hypertension.

Methods and findings: This retrospective observational study (median follow-up of 2.2 years) leveraged the FAMILY Cohort, a large territory-wide cohort in Hong Kong, Special Administrative Region, People's Republic of China to quantify associations of objectively measured liveable space and residential density with blood pressure outcomes among adults aged ≥16 years. Blood pressure outcomes comprised diastolic blood pressure (DBP), systolic blood pressure (SBP), mean arterial pressure (MAP), and hypertension. Liveable space was measured as residential floor area, and density was assessed using the number of residential units per building block and neighborhood residential unit density within predefined catchments. Multivariable regression models examined associations of liveable floor area and residential density with prevalent and incident hypertension. We investigated effect modifications by age, sex, income, employment status, and housing type. Propensity score matching was further employed to match a subset of participants who moved to smaller residences at follow-up with equivalent controls who did not move, and generalized linear models examined the impact of moving to smaller residences upon blood pressure outcomes. Our fully adjusted models of prevalent hypertension outcomes comprised 30,439 participants at baseline, while 13,895 participants were available for incident models at follow-up. We found that each interquartile range (IQR) increment in liveable floor area was associated with lower DBP (beta [β] = -0.269 mm Hg, 95% confidence interval [CI]: -0.419 to -0.118, p < 0.001), SBP (β = -0.317 mm Hg, -0.551 to -0.084, p = 0.008), MAP (β = -0.285 mm Hg, -0.451 to -0.119 with p < 0.001), and prevalent hypertension (odds ratio [OR] = 0.955, 0.918 to 0.993, p = 0.022) at baseline. Each IQR increment in residential units per building block was associated with higher DBP (β = 0.477 mm Hg, 0.212 to 0.742, p = <0.001), SBP (β = 0.750 mm Hg, 0.322 to 1.177, p = <0.001), MAP (β = 0.568 mm Hg, 0.269 to 0.866, p < 0.001), and prevalent hypertension (OR = 1.091, 1.024 to 1.162, p = 0.007). Each IQR increase in neighborhood residential density within 0.5-mi street catchment was associated with lower DBP (β = -0.289 mm Hg, -0.441 to -0.137, p = <0.001), SBP (β = -0.411 mm Hg, -0.655 to -0.168, p < 0.001), MAP (β = -0.330 mm Hg, -0.501 to -0.159, p = <0.001), and lower prevalent hypertension (OR = 0.933, 0.899 to 0.969, p < 0.001). In the longitudinal analyses, each IQR increment in liveable floor area was associated with lower DBP (β = -0.237 mm Hg, -0.431 to -0.043, p = 0.016), MAP (β = -0.244 mm Hg, -0.444 to -0.043, p = 0.017), and incident hypertension (adjusted OR = 0.909, 0.836 to 0.988, p = 0.025). The inverse associations between larger liveable area and blood pressure outcomes were more pronounced among women and those residing in public housing. In the propensity-matched analysis, participants moving to residences of lower liveable floor area were associated with higher odds of incident hypertension in reference to those who did not move (OR = 1.623, 1.173 to 2.199, p = 0.002). The major limitations of the study are unmeasured residual confounding and loss to follow-up.

Conclusions: We disentangled the association of micro-, meso-, and macrolevel residential densities with hypertension and found that higher liveable floor area and neighborhood scale residential density were associated with lower odds of hypertension. These findings suggest adequate housing in the form of provisioning of sufficient liveable space and optimizing residential density at the building block, and neighborhood levels should be investigated as a potential population-wide preventive strategy for lowering hypertension and associated chronic diseases.