Abstract
Background
Radon, a naturally occurring radioactive gas, is ubiquitous in the environment. Little is known about radon’s impact on cardiovascular disease (CVD). This study aims to evaluate the association between residential radon exposure and incident myocardial infarction (MI) and stroke.
Methods
This cohort study included participants with electronic health records (EHR) and residential address information from the All of Us Research Program between May 31, 2017, and October 1, 2023. Residential radon exposures were estimated from a 2024 high-resolution model incorporating over six million measurements across the USA. The main outcome was MI and stroke diagnosis or condition, obtained from EHR. Stratified Cox proportional hazards regression models were used to estimate the hazard ratios (HRs) and 95% CIs for the risk of incident MI and stroke for log-2 transformed radon exposure and quartiles of radon exposure, adjusting for sociodemographic, behavioral, and clinical covariates. Sex- and smoker-stratified analyses were conducted. Penalized spline models were used to estimate the nonlinear association.
Results
A total of 304,050 participants were included. The mean (SD) age was 58.5 (12.9) years old. Among them, 59.2% were female, 18.2% non-Hispanic Black, and 59.0% non-Hispanic White. The median radon exposure was 1.14 pCi/L (interquartile range: 0.90–1.71 pCi/L). Over 950,895 person-years, 1334 MI and 1869 stroke cases were identified. Per doubling of radon exposure was associated with increased risks for incident MI (HR = 1.46, 95% CI 1.04–2.08) and stroke (HR = 1.60, 1.02–2.49). Participants in the third and fourth quartiles had a significantly higher risk for MI and stroke compared to the lowest quartile. Specifically, the fourth quartile was associated with HR = 2.20 (95% CI 1.18–4.10) for MI and HR = 2.38 (95% CI 1.16–4.89) for stroke. Associations persisted across sexes and current smoking status. Nonlinear analyses revealed steep risk increased starting at 1 pCi/L, plateauing at 1.5 pCi/L.
Conclusions
In this cohort study of 304,050 participants, residential radon exposure was significantly associated with elevated risks for incident MI and stroke. Our results suggest that risks may emerge at levels well below this action level. These findings call for further research incorporating household-level exposure measures and lifetime residential histories to confirm the association.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12916-025-04547-4.
Keywords: Radon, Myocardial infarction, Stroke, All of Us
Background
Radon, a naturally occurring radioactive gas with widespread exposure in the general population, is a well-established environmental carcinogen linked to lung cancer [1, 2], accounting for nearly 21,000 lung cancer deaths in the USA [3]. Recent studies also associated radon exposure to breast cancer [4] and stomach cancer [5, 6]. In addition to radon’s carcinogenicity, evidence from toxicologic research suggested that radon exposure also contributes to cardiovascular diseases (CVD), as ionizing radiation arising from the natural radon decay is known to induce oxidative stress [7], endothelial dysfunction [8, 9], and systemic inflammation [10, 11], all of which are processes central to atherosclerosis and thrombogenesis [12, 13].
However, the relationship between radon exposure and CVD risk is under-investigated in the population research. Occupational studies on miners and uranium processing workers with regular radon exposure reported non-significant but increased associations between radon exposure and CVD risk or CVD-related mortality [14–17]. The annual radon exposure level in the occupational cohorts was estimated to be approximately 10 times higher than that in the general population [18]. Nonetheless, the overall risk estimates in these occupational studies may be biased downward due to the healthy worker effect when compared to the general population, though the internal dose–response relationship may still be informative. To date, only a few studies have examined the association between radon exposure and CVD risk in the general population. One study reported no significant association of residential radon exposure with increased CVD mortality in the USA [19], however exclusively focusing on mortality might underestimate CVD burden attributed to radon exposure [20]. Another study observed a significant association between residential radon exposure and stroke prevalence in South Korea when the exposure level was over 2.7 pCi/L [21], but this study was cross-sectional, and the temporal relationship cannot be established. Three prospective studies were recently conducted to evaluate radon exposure and incident stroke in the USA. Among them, one study observed an increased risk for stroke in relation to radon exposure only among never-smokers [22]. Two recent studies using data from Women’s Health Initiative (WHI) concluded that residential radon exposure increased stroke susceptibility as well as incident stroke risk [23, 24]. All three prospective studies reported increased stroke in areas with radon exposure ≥ 2 pCi/L.
Because radon is ubiquitous in the environment and there is no known safe level of radon exposure [25], radon exposure poses huge public health challenges. The current US Environmental Protection Agency (EPA) radon action level of 4 pCi/L is primarily based on evidence from lung cancer studies, while the prior studies reported increased stroke risk below this action level [21–24]. However, it remains unclear whether this action level is adequate to protect against other health risks, particularly CVD, the leading cause of mortality in the USA. The sparse evidence regarding the relationship between radon exposure and CVD risk warrants more rigorous investigations in the general population. Notably, the prior studies in the USA predominantly relied on radon estimates that were published over three decades ago and missed finer, within-unit variations [26]. For instance, the three studies that focused on incident stroke risk used the US EPA radon zones that were published in 1993 and only contained three categories for radon exposure, precluding dose–response analysis of radon’s impact on CVD [22–24]. Moreover, the generalizability of these prior studies was also limited, as only women were included in the two WHI studies [23, 24] and incident CVD other than stroke were never investigated.
To better understand the association between radon exposure and CVD risk, we analyzed the electronic health record (EHR) data of over 300,000 participants in the All of Us Research Program [27]. Our analysis examined the link between residential radon exposure and incident CVD. Specifically, we considered two CVD emergencies in this study: myocardial infarction (MI) and stroke, because of their substantial contribution to CVD mortality. We used the high-resolution radon maps at the zip code level across the USA published in 2024 to assess residential radon exposure among the All of Us participants.
Methods
The all of us research program
Our study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. The All of Us Research Program is an ongoing prospective cohort that currently includes over 849,000 adults living in the USA and its territories initiated in 2017. The goals, recruitment methods and sites, and scientific rationale for All of Us have been described previously [27]. All of Us data include participants’ responses to a series of questionnaires, physical measurements collected by study staff at enrollment, and information from participants’ EHR. All data are made available to researchers via the Researcher Workbench upon approval from the All of Us advisory committee.
This study used the All of Us controlled tier dataset V8 released on February 3, 2025, which includes participants enrolled between May 31, 2017, and October 1, 2023. The median follow-up time was 3.8 (interquartile range: 1.4–4.6) years. We restricted our study population to those with valid EHR and residential data. We restricted our analysis to participants aged 35 or above because of the low risk of MI or stroke among the young population [28]. The flowchart for participant selection can be found in Fig. S1 in Additional File 1. The study is approved and overseen by the All of Us Institute Review Board. Informed consent was waived because of the use of deidentified archival data.
EHR-derived diagnoses
EHR-derived diagnoses were determined using Systematized Nomenclature of Medicine — Clinical Terms codes and mapped to Observational Health and Medicines Outcomes Partnership concept ID by the All of Us Data and Research Center. The full list of concepts used to determine MI and stroke can be found in Methods S1 and S2 in Additional File 1.
We identified primary diagnoses or conditions of MI and stroke after enrollment (i.e., incident MI and stroke) from EHR. For participants with incident MI or stroke, the follow-up time was calculated as the difference between enrollment and the initial diagnosis. For participants without an outcome of interest, the follow-up time was calculated as the difference between enrollment and October 1, 2023 (the cutoff date for dataset V8), or death, whichever came earlier. We also retrieved atherosclerotic cardiovascular disease (ASCVD) history and hypertension status from EHR. Hypertension was defined as two or more hypertension diagnoses or descriptions and at least one hypertension medication prescription in EHR [29].
Radon exposure
Data on average radon concentration between 2001 and 2021 at the 5-digit zip code level across the USA was retrieved from a national model published in 2024 [26]. Specifically, this model used over six million radon measurements conducted by independent laboratories as part of mandatory property inspection to predict nationwide radon levels. These measurements, representative of the general population’s radon exposure, serve as an ideal proxy for long-term exposure levels [26, 30]. A random forest model was built to predict radon concentrations at the zip code level using these measurements with nearly 200 geological, meteorological, architectural, and socioeconomic covariates. We used radon concentrations at the screening floor from this model. More details about the model and radon exposure can be found elsewhere [26]. The radon exposure data are publicly available in the following link: https://osf.io/3shc4/.
The All of Us V8 dataset only contains individuals’ data on 3-digit residential zip code. We therefore calculated population-weighted radon exposure levels at screening floor using radon concentration and population data at the 5-digit zip code level. Our exposure estimates reflect radon concentration at participants’ residence at enrollment.
Covariates
All of Us participants answered the Basics and Lifestyle questionnaires when they were enrolled. The Basics questionnaire elicits demographic information including age, race/ethnicity, education, marital status, household income, and geography. The Lifestyle questionnaire collects data on the use of tobacco, alcohol, and other drugs [31]. Based on these survey questionnaires, we retrieved data on age at enrollment, gender (male, female, other), self-reported race/ethnicity (Hispanic, non-Hispanic Black, non-Hispanic White, other), household income (under $35 K, $35–75 K, $75–150 K, and above $150 K), current smoking status at baseline (no, yes; former smokers were classified as “no”), and health insurance status (no, yes). Body mass index (BMI) was calculated using the height and weight measured at enrollment and grouped into underweight (< 18.5), normal (18.5–25), overweight [25–30], and obese (> 30) categories. The “other” category for race/ethnicity included respondents who identified their race/ethnicity as American Indian, Asian, more than one race/ethnicity, and those not listed above. We also included the quartiles of the deprivation index at the residential address in the model. The deprivation index is a composite score based on six different socioeconomic variables at the community level [32]. The deprivation index was normalized to a range from 0 to 1, with a higher index indicating more deprivation.
Statistical analysis
We used stratified Cox proportional hazards models to estimate the hazard ratio (HR) and 95% confidence interval (CI) for MI and stroke in relation to residential radon exposure. The time scale of the Cox proportional hazards model was time on study (i.e., follow-up time since enrollment). The radon exposure was treated as a continuous variable using log-2 transformed values and as a categorical variable using quartiles, respectively. For log-2 transformed values, the results should be interpreted as HR per doubling increase in radon exposure. The stratified terms included sex assigned at birth, race and ethnicity, and age at enrollment (10-year intervals). The models were additionally adjusted for household income, current smoking status, alcohol drinking status, hypertension status, ASCVD history, and the deprivation index at residential address. To adjust for residual autocorrelation within the geographic units (i.e., 3-digit zip code), we used the generalized estimating equation to calculate a statistically robust CI. Model fit was compared with partial likelihood ratio tests between Cox models with and without radon exposure variables to understand if the radon exposure variable meaningfully improves our models for CVD. Because the partial likelihood ratio test was used, they were unaffected by robust standard errors.
We also stratified the study population by sex and current smoking status and ran the regression models in the stratified populations. To estimate the nonlinear exposure–response relationship, we fitted stratified Cox proportional hazard models with penalized splines for radon exposure with three evenly distributed knots (at the 25th, 50th, and 75th percentiles). The 25th percentile was selected as the reference in nonlinear relationship analysis. To mitigate the influence of extreme values, we demonstrated the nonlinear relationship within the 5th to the 95th percentiles of range of radon exposure in this study population. We presented results for the spline models along with those for the log-2 transformed radon exposure value in the linear model to evaluate the robustness of our results. Missing values were addressed using multiple imputation based on the random forest imputation algorithm [33]. We imputed five complete datasets and pooled estimates from these datasets according to Rubin’s rule. All models followed these criteria.
We conducted several sensitivity analyses to examine the robustness of our results: (1) Additionally adjusted for fine particulate matter (PM2.5) level during the follow-up because radon is considered a part of air pollution, (2) including All of Us participants of all ages, (3) excluding individuals living at the current address for less than 3 years, and (4) unadjusted for hypertension and ASCVD history. The statistical analysis was performed using the survival package in R, version 4.4.2 (R Foundation for Statistical Computing). All codes for analysis are available on the All of Us Researcher Workbench.
Results
Study population
After excluding participants younger than 35 and without EHR and residential address data (Fig. S1 in Additional File 1), the study included 304,050 participants across the USA and 950,895 person-years through October 1, 2023 (Table 1 and Fig. 1). The mean (SD) age was 58.5 (12.9) years old. Among them, 35.1% were older than 65 years, 59.2% female, 39.7% male, 15.5% Hispanic, 18.2% non-Hispanic Black, and 59.0% non-Hispanic White. Notably, under 25% of participants reported living in the current address for less than 3 years (Fig. S2 in Additional File 1). In the total population, the average radon exposure level was 1.36 pCi/L (Fig. 2; Table S1 in Additional File 1), with an interquartile range from 0.90 to 1.71 pCi/L, while the EPA reports a national average level of 1.3 pCi/L and recommends caution and consideration for radon levels above 2 pCi/L [25]. In comparison, the incident MI and stroke patients were exposed to a higher radon level, with an average level of 1.51 pCi/L for MI and 1.50 pCi/L for stroke.
Table 1.
Distribution of selected characteristics in the overall study population and stratified populations in the All of Us Research Program
| Characteristics | Total (n = 304,050) | Radon exposure | |||
|---|---|---|---|---|---|
| 1st quartile (n = 78,116) |
2nd quartile (n = 75,854) |
3rd quartile (n = 75,795) |
4th quartile (n = 74,285) |
||
| Age at enrollment | |||||
| Mean (SD) | 58.5 (12.9) | 57.6 (12.6) | 57.6 (13.1) | 58.3 (12.7) | 60.4 (13.1) |
| 35–49 | 82,920 (27.3) | 22,360 (28.6) | 22,903 (30.1) | 20,491 (27.1) | 17,166 (23.2) |
| 50–64 | 114,280 (37.6) | 31,461 (40.2) | 28,294 (37.2) | 29,076 (38.4) | 25,449 (34.3) |
| > 64 | 106,850 (35.1) | 24,414 (31.2) | 24,809 (32.6) | 26,153 (34.5) | 31,474 (42.5) |
| Sex at birth | |||||
| Female | 179,369 (59.0) | 46,045 (58.9) | 44,446 (58.6) | 44,605 (58.8) | 44,273 (59.6) |
| Male | 120,269 (39.6) | 31,011 (39.7) | 30,459 (40.2) | 29,935 (39.5) | 28,864 (38.9) |
| Other | 330 (0.1) | 99 (0.1) | 76 (0.1) | 95 (0.1) | 60 (0.1) |
| Missing | 4082 (1.3) | 961 (1.2) | 873 (1.2) | 1160 (1.5) | 1088 (1.5) |
| Race and ethnicity | |||||
| Hispanic | 46,953 (15.4) | 17,596 (22.5) | 20,494 (27.0) | 6182 (8.2) | 2681 (3.6) |
| Non-Hispanic Black | 55,291 (18.2) | 22,059 (28.2) | 11,028 (14.5) | 17,005 (22.4) | 5199 (7.0) |
| Non-Hispanic White | 178,931 (58.8) | 31,033 (39.7) | 38,785 (51.1) | 47,213 (62.3) | 61,900 (83.3) |
| Other | 15,908 (5.2) | 5843 (7.5) | 4005 (5.3) | 3528 (4.7) | 2532 (3.4) |
| Missing | 6967 (2.3) | 1585 (2.0) | 1542 (2.0) | 1867 (2.5) | 1973 (2.7) |
| Household income | |||||
| < 35,000 | 86,895 (28.6) | 25,787 (33.0) | 22,605 (29.8) | 20,814 (27.5) | 17,689 (23.8) |
| 35,000– < 50,000 | 22,655 (7.5) | 4237 (5.4) | 5870 (7.7) | 5493 (7.2) | 7055 (9.5) |
| 50,000– < 75,000 | 32,714 (10.8) | 5827 (7.5) | 7762 (10.2) | 8450 (11.1) | 10,675 (14.4) |
| 75,000– < 150,000 | 61,066 (20.1) | 11,702 (15.0) | 13,189 (17.4) | 16,734 (22.1) | 19,441 (26.2) |
| ≥ 150,000 | 39,829 (13.1) | 10,710 (13.7) | 7533 (9.9) | 11,447 (15.1) | 10,139 (13.6) |
| Missing | 60,891 (20.0) | 19,853 (25.4) | 18,895 (24.9) | 12,857 (17.0) | 9286 (12.5) |
| Education | |||||
| Less than high school | 27,133 (8.9) | 11,844 (15.2) | 7855 (10.4) | 5349 (7.1) | 2085 (2.8) |
| High school or equivalent | 52,445 (17.2) | 14,117 (18.1) | 14,469 (19.1) | 12,714 (16.8) | 11,145 (15.0) |
| Some college | 77,652 (25.5) | 17,179 (22.0) | 21,674 (28.6) | 18,202 (24.0) | 20,597 (27.7) |
| College | 67,807 (22.3) | 15,551 (19.9) | 15,686 (20.7) | 17,349 (22.9) | 19,221 (25.9) |
| Graduate | 70,792 (23.3) | 16,956 (21.7) | 14,024 (18.5) | 19,957 (26.3) | 19,855 (26.7) |
| Missing | 8221 (2.7) | 2469 (3.2) | 2146 (2.8) | 2224 (2.9) | 1382 (1.9) |
| Smoking status | |||||
| No | 177,754 (58.5) | 48,596 (62.2) | 45,620 (60.1) | 43,607 (57.5) | 39,931 (53.8) |
| Yes | 121,398 (39.9) | 28,539 (36.5) | 29,173 (38.5) | 30,959 (40.8) | 32,727 (44.1) |
| Missing | 4898 (1.6) | 981 (1.3) | 1061 (1.4) | 1229 (1.6) | 1627 (2.2) |
| Alcohol drinking | |||||
| No | 28,175 (9.3) | 10,022 (12.8) | 9159 (12.1) | 6139 (8.1) | 2855 (3.8) |
| Yes | 267,455 (88.0) | 66,040 (84.5) | 64,684 (85.3) | 67,490 (89.0) | 69,241 (93.2) |
| Missing | 8420 (2.8) | 2054 (2.6) | 2011 (2.7) | 2166 (2.9) | 2189 (2.9) |
| Hypertension status | |||||
| No | 176,152 (57.9) | 48,360 (61.9) | 43,900 (57.9) | 42,941 (56.7) | 40,951 (55.1) |
| Yes | 127,898 (42.1) | 29,756 (38.1) | 31,954 (42.1) | 32,854 (43.3) | 33,334 (44.9) |
| ASCVD history | |||||
| No | 288,688 (94.9) | 74,827 (95.8) | 74,145 (97.7) | 71,217 (94.0) | 68,499 (92.2) |
| Yes | 15,362 (5.1) | 3289 (4.2) | 1709 (2.3) | 4578 (6.0) | 5786 (7.8) |
| BMI groups | |||||
| < 18.5 | 3086 (1.0) | 1136 (1.5) | 734 (1.0) | 724 (1.0) | 492 (0.7) |
| 18.5– < 25 | 68,975 (22.7) | 20,146 (25.8) | 15,160 (20.0) | 17,872 (23.6) | 15,797 (21.3) |
| 25– < 30 | 89,952 (29.6) | 23,589 (30.2) | 21,060 (27.8) | 22,841 (30.1) | 22,462 (30.2) |
| ≥ 30 | 121,722 (40.0) | 30,141 (38.6) | 29,731 (39.2) | 31,143 (41.1) | 30,707 (41.3) |
| Missing | 20,315 (6.7) | 3104 (4.0) | 9169 (12.1) | 3215 (4.2) | 4827 (6.5) |
| Deprivation index | |||||
| 1st quartile | 76,797 (25.3) | 8759 (11.2) | 14,594 (19.2) | 18,017 (23.8) | 35,427 (47.7) |
| 2nd quartile | 78,288 (25.7) | 18,605 (23.8) | 4689 (6.2) | 30,120 (39.7) | 24,874 (33.5) |
| 3rd quartile | 75,520 (24.8) | 13,289 (17.0) | 27,323 (36.0) | 23,560 (31.1) | 11,348 (15.3) |
| 4th quartile | 73,445 (24.2) | 37,463 (48.0) | 29,248 (38.6) | 4098 (5.4) | 2636 (3.5) |
Fig. 1.
A Radon exposure by 3-digit zip code level. B Spatial distribution of participants
Fig. 2.
Distribution of radon exposure level in the study populations
A total of 1334 incident MIs (incidence rate: 0.14 per 100 person-years) and 1869 incident strokes (incidence rate: 0.19 per 100 person-years) were identified from the EHRs (Table 2). The incidence rates were higher in the 3rd and 4th quartiles of radon exposure.
Table 2.
Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in the study population
| Radon exposurea | MI | Stroke | |||||
|---|---|---|---|---|---|---|---|
| Incidence no | Incidence rate, per 100 person-years |
HR (95% CI)b | Incidence no | Incidence rate, per 100 person-years |
HR (95% CI)b | ||
| Overall | |||||||
| Per doubling | 1334 | 0.14 | 1.46 (1.03–2.07) | 1869 | 0.19 | 1.58 (1.02–2.47) | |
| 1st quartile | 260 | 0.10 | Ref | 389 | 0.16 | Ref | |
| 2nd quartile | 126 | 0.05 | 0.73 (0.25–2.15) | 139 | 0.06 | 0.49 (0.18–1.30) | |
| 3rd quartile | 473 | 0.21 | 2.47 (1.32–4.62) | 700 | 0.28 | 2.47 (1.17–5.22) | |
| 4th quartile | 475 | 0.22 | 2.18 (1.17–4.06) | 641 | 0.26 | 2.36 (1.14–4.87) | |
| Female | |||||||
| Per doubling | 512 | 0.09 | 1.39 (0.90–2.14) | 878 | 0.15 | 1.59 (0.98–2.58) | |
| 1st quartile | 104 | 0.07 | Ref | 203 | 0.13 | Ref | |
| 2nd quartile | 46 | 0.03 | 0.62 (0.39–0.99) | 72 | 0.05 | 0.47 (0.17–1.29) | |
| 3rd quartile | 179 | 0.13 | 2.52 (1.15–5.49) | 297 | 0.22 | 2.35 (1.07–5.17) | |
| 4th quartile | 183 | 0.14 | 2.44 (1.15–5.20) | 306 | 0.23 | 2.41 (1.14–5.10) | |
| Male | |||||||
| Per doubling | 610 | 0.16 | 1.41 (0.99–2.03) | 707 | 0.18 | 1.74 (1.10–2.74) | |
| 1st quartile | 135 | 0.13 | Ref | 153 | 0.14 | Ref | |
| 2nd quartile | 65 | 0.07 | 0.75 (0.29–1.93) | 56 | 0.06 | 0.48 (0.17–1.33) | |
| 3rd quartile | 215 | 0.24 | 2.34 (1.33–4.13) | 252 | 0.27 | 2.84 (1.37–5.88) | |
| 4th quartile | 195 | 0.25 | 2.04 (1.16–3.56) | 246 | 0.30 | 2.74 (1.32–5.66) | |
| Non-smoking | |||||||
| Per doubling | 449 | 0.11 | 1.41 (0.96–2.09) | 580 | 0.15 | 1.70 (1.06–2.71) | |
| 1st quartile | 87 | 0.09 | Ref | 116 | 0.12 | Ref | |
| 2nd quartile | 38 | 0.05 | 0.72 (0.25–2.10) | 45 | 0.05 | 0.45 (0.17–1.21) | |
| 3rd quartile | 159 | 0.17 | 2.53 (1.27–5.02) | 189 | 0.20 | 2.66 (1.22–5.78) | |
| 4th quartile | 165 | 0.18 | 2.38 (1.25–4.54) | 230 | 0.25 | 2.65 (1.26–5.58) | |
| Smoking | |||||||
| Per doubling | 683 | 0.12 | 1.46 (0.98–2.16) | 1024 | 0.19 | 1.64 (1.01–2.66) | |
| 1st quartile | 150 | 0.09 | Ref | 241 | 0.14 | Ref | |
| 2nd quartile | 73 | 0.05 | 0.66 (0.22–1.99) | 83 | 0.06 | 0.51 (0.16–1.57) | |
| 3rd quartile | 239 | 0.18 | 2.31 (1.31–4.09) | 365 | 0.27 | 2.40 (1.10–5.23) | |
| 4th quartile | 221 | 0.19 | 1.95 (1.09–3.49) | 335 | 0.28 | 2.50 (1.18–5.33) | |
CI confidence interval, HR hazard ratio, MI myocardial infarction
aLithium concentration range for each quartile: 0.56–0.89 pCi/L for 1st quartile, 0.90–1.13 pCi/L for 2nd quartile, 1.14–1.70 pCi/L for 3rd quartile, 1.71–5.16 pCi/L for 4th quartile
bThree stratified terms were included: sex at birth, race and ethnicity, and age at 10-year interval. Adjusted for household income, current smoking status, alcohol drinking status, hypertension status, atherosclerotic cardiovascular disease history, BMI groups, education, and deprivation index quartiles of residential address
Risk associated with residential radon exposure
Higher radon exposure level was significantly associated with increased risks for both MI and stroke (Table 2). Per doubling increase of radon exposure level was associated with HR = 1.46 (95% CI 1.04–2.08) for MI and HR = 1.60 (95% CI 1.02–2.49) for stroke. Compared to the 1st quartile (0.56–0.89 pCi/L) of radon exposure, the 2nd quartile (0.90–1.13 pCi/L) did not demonstrate any significant associations, whereas the 3rd (1.14–1.70 pCi/L) and 4th (1.71–5.16 pCi/L) quartiles was associated with increased risks for MI (HR = 2.48, 95% CI 1.32–4.65 for the 3rd quartile; HR = 2.20, 95% CI 118–4.10 for the 4th quartile) and stroke (HR = 2.48, 95% CI 1.17–5.24 for the 3rd quartile; HR = 2.38, 95% CI 1.16–4.89 for the 4th quartile). Model fit test results demonstrated that adding radon exposure variables significantly improved the model (all P-values < 0.0001, Table S2 in Additional File 1).
When stratified by sex, higher radon exposure remained significantly associated with increased risks for MI and stroke in both males and females, with no substantial difference in the associations between the two sexes (Table 2), as P-values for interaction were all larger than 0.10 (Table S3 in Additional File 1). For instance, compared to the 1st quartile of radon exposure, the 4th quartile demonstrated similar associations with MI in females (HR = 2.37, 95% CI 1.18–4.78) and males (HR = 2.05, 95% CI 1.17–3.59) and also similar associations with stroke in females (HR = 2.47, 95% CI 1.17–5.22) and males (HR = 2.71, 95% CI 1.32–5.57).
Again, when stratified by current smoking status, higher radon exposure remained significantly associated with increased CVD risks in both smokers and nonsmokers (Table 2), with P-values for interaction larger than 0.10 (Table S3 in Additional File 1). Take stroke as an example: per doubling increases in radon exposure was associated with HR = 1.70 (95% CI 106–2.71) in non-smoking participants vs. HR = 1.64 (95% CI 1.01–2.66) in smoking participants.
The nonlinear analysis corroborated that higher radon exposure was associated with increased risks for MI and stroke (Fig. 3). For both MI and stroke risks, a steep slope was observed beginning from 1 pCi/L, and the risk reached a plateau at 1.5 pCi/L. Compared to the nonlinear trend, estimates based on the log-2 transformed values in the linear model appeared to underestimate the HR associated with radon exposure.
Fig. 3.
Associations of radon exposure with myocardial infarction and stroke risks from the log-2 transformed linear model and the nonlinear spline model. A Myocardial infarction. B Stroke
Sensitivity analyses that additionally adjusted for PM2.5 exposure (Table S4 in Additional File 1), included All of Us participants of all ages (Table S5 in Additional File 1), excluded participants who have been living in the address for less than three years (Table S6 in Additional File 1), and unadjusted for hypertension and ASCVD history (Table S7 in Additional File 1) all yielded consistent results. Particularly, when additionally adjusted for PM2.5 exposure level or unadjusted for hypertension and ASCVD history, the associations of radon exposure were strengthened.
Discussion
In this large, nationwide cohort study of 304,050 participants in the All of Us Research Program, we observed significant associations between residential radon exposure and increased risks of incident MI and stroke. These associations persisted across sexes and demonstrated a dose–response relationship, with risks rising steeply at radon levels above 1 pCi/L, a level much lower than the current EPA action level of 4 pCi/L for lung cancer mitigation [34]. Our findings contribute novel evidence to the growing body of research suggesting that radon, a ubiquitous environmental carcinogen, may be associated with CVD burdens, indicating that current action levels may not fully address the public health burden posed by radon exposure.
In this study, the majority of participants resided in areas with low residential radon exposure levels, averaging 1.36 pCi/L, a value comparable to the national average of 1.3 pCi/L reported by EPA, suggesting geographic representativeness of the study population. EPA recommends actions if the radon level is 4 pCi/L or above and advises caution and consideration if the radon level is 2 to 4 pCi/L. However, in this study, even within a low-exposure range lower than 2 pCi/L, we still observed significant dose-dependent increases in MI and stroke risks. Critically, our nonlinear exposure–response analysis identified a steep risk gradient beginning at 1 pCi/L, a level far lower than the EPA’s lung cancer-based guideline. These findings challenge the adequacy of existing radon mitigation policies, which do not account for CVD.
Our main models were adjusted for hypertension and ASCVD history. However, the role of the preexisting CVD conditions in the association between radon exposure and CVD risks remains debatable. Some researchers consider the preexisting conditions as intermediate processes on the causal pathway from radon exposure to MI and stroke. Adjustment for them may therefore attenuate the true associations by blocking part of the mediating effect. Alternatively, one could argue that if radon’s effect on MI and stroke operates exclusively by increasing the risk of hypertension and ASCVD, then public health efforts could focus solely on managing these conditions, potentially making a radon-specific action level unnecessary. In our sensitivity analysis that was not adjusted for hypertension and ASCVD history, we indeed observed slightly strengthened associations, suggesting the preexisting conditions did partially, but not completely, explain the associations of radon exposure. Future studies should investigate these alternative pathways linking radon exposure to CVD.
Our results align with emerging toxicological and epidemiological data linking radon to CVD [7–11, 22–24]. The observed associations are biologically plausible given the radioactivity of radon. The natural radon decay releases ionizing radiation that induces oxidative stress and triggers inflammatory responses in vascular tissues, leading to endothelial dysfunction and promoting atherosclerosis [35, 36]. Experimental studies have shown that even low doses of ionizing radiation can damage vascular cells through mechanisms such as lipid peroxidation and cytokine release [37]. Additionally, ionizing radiation can induce DNA damage and cellular senescence in vascular endothelial and smooth muscle cells, further compromising cardiovascular health [38]. Moreover, radon is soluble in blood and bone marrow [39], where its ionizing radiation may induce somatic mutations that drive clonal proliferation of circulating leukocytes or hematopoietic stem cells capable of accelerating inflammatory, atherosclerotic, or thromboembolic processes [40], thereby predisposing individuals to stroke and other CVD.
A strength of this study is that it was a large, nationwide, diverse longitudinal study. The comprehensive EHR data allowed us to retrieve key variables. The high-resolution radon exposure data derived from over six million measurements across the USA offers a more precise and contemporary exposure assessment than earlier EPA zone-based categorizations [26], which also allowed us to conduct a dose–response analysis to understand whether the current radon action level was enough to protect the public’s health. The diverse populations in All of Us make the results more generalizable compared with a prior study restricted to certain regions or racial, ethnic, and socioeconomic groups [27]. Moreover, our research was strengthened by the rich longitudinal nature of the All of Us Research Program, capturing incident CVD diagnosis with ongoing EHRs.
Several limitations should be considered. First, the incidence rates of MI and stroke were 0.14 and 0.19 per 100 person-years in this study. In comparison, the US Centers for Disease Control and Prevention reports 805,000 new cases for MI and 795,000 new cases for stroke every year, which roughly corresponds to an incidence rate of 0.23 per 100 person-years for both conditions, given a population size of 342 million in the USA [41, 42]. However, underdiagnosis appears to be nondifferential in relation to radon exposure because the participants and All of Us investigators were not aware of the research on radon exposure when the data were collected. The nondifferential underdiagnosis would bias the estimates toward the null in epidemiologic research. Second, the All of Us Research Program has not been linked to the National Death Index yet. Mortality status in All of Us is currently reported by each health provider organization. Therefore, it is possible that some deaths, either from CVD or other causes, were not recorded. The missingness of mortality data is not related to individuals’ radon exposure level and thus is unlikely to lead to bias through the missing-not-at-random mechanism. If mortality status was missing completely at random, the effect on our estimates would be negligible aside from a minor loss of efficiency. If mortality status was missing at random conditional on other measured covariates such as demographics or health system, our analytic approach (multiple imputation) would yield unbiased estimates. Third, the radon exposure was assigned at the 3-digit zip code level, which may not perfectly reflect individuals’ exposure levels and individual household-level factors (e.g., radon mitigation systems, ventilation practices). Also, this exposure was not time-varying or cumulative, and could not accurately reflect individuals’ cumulative exposure dose. Consequently, we could not apply exposure lags to reduce concerns about reverse causation. For example, participants with chronic CVD may spend more time indoors. These limitations could lead to nondifferential misclassification but in some circumstances may also bias associations away from the null. However, the radon prediction model used in our study integrates over six million household-level measurements and includes variables reflecting housing characteristics, which helps improve ecological precision relative to prior studies that relied solely on EPA zone-based categories. Therefore, the exposure at the 3-digit zip code level in this study, although not optimal, still provided crucial evidence for the disparities between subpopulations. Fourth, although we adjusted for current smoking status (yes, no) based on questionnaire data, this measure may not fully capture lifetime smoking intensity, duration, or changes over the follow-up period. As prior studies suggested that smoking may modify radon’s effect on CVD [22], incomplete characterization of smoking exposure could prevent us from understanding the complex relationship between smoking, radon exposure, and cardiovascular outcomes. Fifth, our study could not account for terrestrial gamma radiation, which is correlated with indoor radon concentrations due to shared geological sources. Previous research has suggested associations between gamma radiation exposure and circulatory diseases [43]. Because of the lack of terrestrial gamma radiation, we cannot rule out the confounding effect of gamma radiation.
Conclusion
In conclusion, our findings provide novel evidence of an association between residential radon exposure and increased risks of MI and stroke in the general population. While the current EPA action level is 4 pCi/L, our results suggest that risks may emerge at levels well below this action level. These findings call for further research incorporating household-level exposure measures and lifetime residential histories to confirm the association and to inform whether the current action level adequately protects against cardiovascular risk.
Supplementary Information
Additional file 1. Method S1: Full OMOP concept list for heart attack. Method S2: Full OMOP concept list for stroke. Table S1: Descriptive statistics of radon exposure level in the study population. Table S2: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in the study population, additionally adjusted for PM2.5. Table S3: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in the All of Us population of all ages. Table S4: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in study population that have been living in the address for 3 or more years. Table S5: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in the All of Us population of all ages. Table S6: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in study population that have been living in the address for 3 or more years. Table S7: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in study population, not adjusted for hypertension and ASCVD history. Figure S1: Flowchart for participant selection in the study. Figure S2: Distribution of duration for living in the current address among the study participants.
Acknowledgements
The authors thank the dedicated All of Us field staff and community partners for their support of this work.
Abbreviations
- ASCVD
Atherosclerotic cardiovascular disease
- CI
Confidence interval
- CVD
Cardiovascular disease
- EHR
Electronic health record
- HR
Hazards ratio
- MI
Myocardial infarction
- PM2.5
Fine particulate matter
Authors’ contributions
J.L., J.P., and B.A. conceived the study; H.A. and B.A. acquired the data; J.L., J.Z. and Y.Y. conducted the data analysis; J.L., C.O., and A.N. interpreted the results; J.L. wrote the manuscript draft. All authors read and approved the final manuscript.
Funding
This research was supported by funding from NIH grants P30ES027792, R03HL172114, and 1OT2OD036445.
Data availability
The radon exposure data are publicly available in the link: https:/osf.io/3shc4. The All of Us data are available on the All of Us Researcher Workbench upon approval by the All of Us Advisory Committee.
Declarations
Ethics approval and consent to participate
This research was reviewed by the University of Chicago Institutional Review Board (IRB). Because this research used publicly available de-identified data, it is exempt from IRB approval.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Samet JM. Radon and lung cancer. JNCI J Natl Cancer Inst. 1989;81(10):745–58. [DOI] [PubMed] [Google Scholar]
- 2.Turner MC, Krewski D, Chen Y, Pope CA III, Gapstur S, Thun MJ. Radon and lung cancer in the American Cancer Society cohort. Cancer Epidemiol Biomarkers Prev. 2011;20(3):438–48. [DOI] [PubMed] [Google Scholar]
- 3.Gaskin J, Coyle D, Whyte J, Krewksi D. Global estimate of lung cancer mortality attributable to residential radon. Environ Health Perspect. 2018;126(5):057009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.VoPham T, DuPré N, Tamimi RM, James P, Bertrand KA, Vieira V, et al. Environmental radon exposure and breast cancer risk in the Nurses’ Health Study II. Environ Health. 2017;16:1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Messier KP, Serre ML. Lung and stomach cancer associations with groundwater radon in North Carolina, USA. Int J Epidemiol. 2017;46(2):676–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Barbosa-Lorenzo R, Barros-Dios JM, Raíces Aldrey M, Cerdeira Caramés S, Ruano-Ravina A. Residential radon and cancers other than lung cancer: a cohort study in Galicia, a Spanish radon-prone area. Eur J Epidemiol. 2016;31:437–41. [DOI] [PubMed] [Google Scholar]
- 7.Kataoka T, Shuto H, Naoe S, Yano J, Kanzaki N, Sakoda A, et al. Radon inhalation decreases DNA damage induced by oxidative stress in mouse organs via the activation of antioxidative functions. J Radiat Res. 2021;62(5):861–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Blomberg AJ, Nyhan MM, Bind M-A, Vokonas P, Coull BA, Schwartz J, et al. The role of ambient particle radioactivity in inflammation and endothelial function in an elderly cohort. Epidemiology. 2020;31(4):499–508. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Baselet B, Sonveaux P, Baatout S, Aerts A. Pathological effects of ionizing radiation: endothelial activation and dysfunction. Cell Mol Life Sci. 2019;76:699–728. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mavragani IV, Laskaratou DA, Frey B, Candéias SM, Gaipl US, Lumniczky K, et al. Key mechanisms involved in ionizing radiation-induced systemic effects. A current review. Toxicol Res. 2016;5(1):12–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Taylor BK, Smith OV, Miller GE. Chronic home radon exposure is associated with higher inflammatory biomarker concentrations in children and adolescents. Int J Environ Res Public Health. 2022;20(1):246. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bonomini F, Tengattini S, Fabiano A, Bianchi R, Rezzani R. Atherosclerosis and oxidative stress. Histol Histopathol. 2008. 10.14670/HH-23.381. [DOI] [PubMed] [Google Scholar]
- 13.Madamanchi N, Hakim Z, Runge M. Oxidative stress in atherogenesis and arterial thrombosis: the disconnect between cellular studies and clinical outcomes. J Thromb Haemost. 2005;3(2):254–67. [DOI] [PubMed] [Google Scholar]
- 14.Lu L, Zhang Y, Chen C, Field RW, Kahe K. Radon exposure and risk of cerebrovascular disease: a systematic review and meta-analysis in occupational and general population studies. Environ Sci Pollut Res Int. 2022;29(30):45031–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Rage E, Caër-Lorho S, Laurier D. Low radon exposure and mortality among Jouac uranium miners: an update of the French cohort (1946–2007). J Radiol Prot. 2017;38(1):92. [DOI] [PubMed] [Google Scholar]
- 16.Navaranjan G, Berriault C, Do M, Villeneuve PJ, Demers PA. Cancer incidence and mortality from exposure to radon progeny among Ontario uranium miners. Occup Environ Med. 2016;73(12):838–45. [DOI] [PubMed] [Google Scholar]
- 17.Xiang-Zhen X, Lubin JH, Jun-Yao L, Li-Fen Y, Sheng LQ, Lan Y, et al. A cohort study in southern China of tin miners exposed to radon and radon decay products. Health Phys. 1993;64(2):120–31. [DOI] [PubMed] [Google Scholar]
- 18.Marsh JW, Tomášek L, Laurier D, Harrison JD. Effective dose coefficients for radon and progeny: a review of ICRP and UNSCEAR values. Radiat Prot Dosimetry. 2021;195(1):1–20. [DOI] [PubMed] [Google Scholar]
- 19.Turner MC, Krewski D, Chen Y, Pope CA III, Gapstur SM, Thun MJ. Radon and nonrespiratory mortality in the American Cancer Society cohort. Am J Epidemiol. 2012;176(9):808–14. [DOI] [PubMed] [Google Scholar]
- 20.Amini M, Zayeri F, Salehi M. Trend analysis of cardiovascular disease mortality, incidence, and mortality-to-incidence ratio: results from global burden of disease study 2017. BMC Public Health. 2021;21(1):1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kim SH, Park JM, Kim H. The prevalence of stroke according to indoor radon concentration in South Koreans: nationwide cross section study. Medicine. 2020;99(4):e18859. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zhang Y, Yan Q, Angley M, Lu L, Miller EC, Judd SE, et al. Smoking modifies the association between radon exposure and incident ischemic stroke: the REGARDS study. Stroke. 2023;54(11):2737–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Buchheit SF, Collins JM, Anthony KM, Love S-AM, Stewart JD, Gondalia R, et al. Radon exposure and incident stroke risk in the Women’s Health Initiative. Neurology. 2024;102(4):e209143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Anthony KM, Collins JM, Love S-AM, Stewart JD, Buchheit SF, Gondalia R, et al. Radon exposure, clonal hematopoiesis, and stroke susceptibility in the Women’s Health Initiative. Neurology. 2024;102(2):e208055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Environmental Protection Agency. What is EPA’s action level for radon and what does it mean? 2024 [Available from: https://www.epa.gov/radon/what-epas-action-level-radon-and-what-does-it-mean.
- 26.Li L, Coull BA, Zilli Vieira CL, Koutrakis P. High-resolution national radon maps based on massive indoor measurements in the United States. Proc Natl Acad Sci USA. 2025;122(3):e2408084121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.All of Us Research Program Investigators. The “all of us” research program. N Engl J Med. 2019;381(7):668–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Burrows NR, Li Y, Gregg EW, Geiss LS. Declining rates of hospitalization for selected cardiovascular disease conditions among adults aged≥ 35 years with diagnosed diabetes, US, 1998–2014. Diabetes Care. 2018;41(2):293–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Chandler PD, Clark CR, Zhou G, Noel NL, Achilike C, Mendez L, et al. Hypertension prevalence in the all of us research program among groups traditionally underrepresented in medical research. Sci Rep. 2021;11(1):12849. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Li L, Coull BA, Koutrakis P. A national comparison between the collocated short-and long-term radon measurements in the United States. J Expo Sci Environ Epidemiol. 2023;33(3):455–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.National Institutes of Health. All of Us Research Hub 2023 [Available from: https://www.researchallofus.org/.
- 32.Brokamp C, Beck AF, Goyal NK, Ryan P, Greenberg JM, Hall ES. Material community deprivation and hospital utilization during the first year of life: an urban population–based cohort study. Ann Epidemiol. 2019;30:37–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Shah AD, Bartlett JW, Carpenter J, Nicholas O, Hemingway H. Comparison of random forest and parametric imputation models for imputing missing data using MICE: a CALIBER study. Am J Epidemiol. 2014;179(6):764–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Environmental Protection Agency. The National Radon Action Plan 2021–2025 2022 [Available from: http://www.radonleaders.org/resources/nationalradonactionplan.
- 35.Lesbek A, Omori Y, Bakhtin M, Kazymbet P, Tokonami S, Altaeva N, et al. Systematic review and meta-analysis of inflammatory biomarkers in individuals exposed to radon. Biomedicines. 2025;13(2):499. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Kataoka T. Study of antioxidative effects and anti-inflammatory effects in mice due to low-dose X-irradiation or radon inhalation. J Radiat Res. 2013;54(4):587–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Little MP, Azizova TV, Bazyka D, Bouffler SD, Cardis E, Chekin S, et al. Systematic review and meta-analysis of circulatory disease from exposure to low-level ionizing radiation and estimates of potential population mortality risks. Environ Health Perspect. 2012;120(11):1503–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Bhattacharya S, Asaithamby A. Ionizing radiation and heart risks. In Seminars in cell & developmental biology. Academic Press; 2016. Vol. 58, pp. 14–25. [DOI] [PubMed]
- 39.Hinrichs A, Schmitt M, Papenfuß F, Roth M, Fournier C, Kraft G, et al. Radon solubility in different tissues after short term exposure. Int J Environ Res Public Health. 2023;20(3):1773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Soudet S, Jedraszak G, Evrard O, Marolleau J-P, Garçon L, Pietri MS. Is hematopoietic clonality of indetermined potential a risk factor for pulmonary embolism? TH Open. 2021;5(03):e338–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Centers for Disease Control and Prevention. Heart Disease Facts 2024 [Available from: https://www.cdc.gov/heart-disease/data-research/facts-stats/?CDC_AAref_Val=https://www.cdc.gov/heartdisease/facts.htm.
- 42.Centers for Disease Control and Prevention. Stroke Facts 2024 [Available from: https://www.cdc.gov/stroke/data-research/facts-stats/?CDC_AAref_Val=https://www.cdc.gov/stroke/facts.htm.
- 43.Huang S, Garshick E, Vieira CL, Grady ST, Schwartz JD, Coull BA, et al. Short-term exposures to particulate matter gamma radiation activities and biomarkers of systemic inflammation and endothelial activation in COPD patients. Environ Res. 2020;180:108841. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Additional file 1. Method S1: Full OMOP concept list for heart attack. Method S2: Full OMOP concept list for stroke. Table S1: Descriptive statistics of radon exposure level in the study population. Table S2: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in the study population, additionally adjusted for PM2.5. Table S3: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in the All of Us population of all ages. Table S4: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in study population that have been living in the address for 3 or more years. Table S5: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in the All of Us population of all ages. Table S6: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in study population that have been living in the address for 3 or more years. Table S7: Adjusted HR and 95% confidence intervals for myocardial infarction and stroke according to residential radon exposure in study population, not adjusted for hypertension and ASCVD history. Figure S1: Flowchart for participant selection in the study. Figure S2: Distribution of duration for living in the current address among the study participants.
Data Availability Statement
The radon exposure data are publicly available in the link: https:/osf.io/3shc4. The All of Us data are available on the All of Us Researcher Workbench upon approval by the All of Us Advisory Committee.