2025 Volume 30 Pages 23
Background: Studies have shown that married couples often share similar lifestyles, as well as lifestyle-associated conditions such as diabetes, hypertension, and hyperlipidemia. This study aims to prospectively investigate the association between an individual’s development of a non-communicable disease and the subsequent development of the same condition in their spouse.
Methods: This population-based cohort study utilized 12 waves of annual prospective surveys from 2005 onwards in Japan, with a discrete-time design. A total of 9,417 middle-aged couples (18,834 participants; discrete-time observations = 118,876) were included. Each participant whose spouse had developed one of six conditions was propensity score-matched with five controls whose spouses had not been diagnosed with the condition: diabetes [n = 1374 vs n = 6870], hypertension [n = 2657 vs n = 13285], hypercholesterolemia [n = 3321 vs n = 16605], stroke [n = 567 vs n = 2835], coronary heart disease (CHD) [n = 1093 vs n = 5465] or cancer [n = 923 vs n = 4615]. Using conditional logistic regression, we assessed participants’ development of the same condition within three years following their spouse’s diagnosis.
Results: Participants whose spouses had developed diabetes, hypertension, hypercholesterolemia, or CHD were more likely to develop the same condition within three years. The odds ratios (ORs) and 95% confidence intervals (CIs) were: 1.96 (1.53–2.50), 1.20 (1.06–1.36), 1.63 (1.47–1.81) and 1.43 (1.05–1.95), respectively. No significant associations were observed in stroke [1.69 (0.80–3.58)] or cancer [1.08 (0.75–1.54)].
Conclusion: Spouses of individuals recently diagnosed with certain metabolic conditions are at a higher risk of developing those conditions themselves. These findings may provide valuable guidance for targeting and personalizing chronic disease screening and prevention efforts.
Non-communicable diseases (NCDs) account for the largest number of deaths worldwide [1] and are closely associated with lifestyle-associated risk factors, such as smoking, physical inactivity and obesity [2]. Spouses often exhibit similar behavioral patterns as one another, which could be due to the development of shared habits after marriage or due to assortative mating (i.e., a tendency of individuals to choose a spouse with similar characteristics, such as body mass index, or habits such as smoking) [3, 4]. Such shared lifestyle behaviors can affect partners’ mutual health and may place spouses, both individually and together, at higher risk of NCDs.
Prior literature has suggested when one spouse has a NCD, the other may be more likely to have the same condition themselves. Specifically, cross-sectional studies and their meta-analysis have found that when one spouse had conditions like diabetes, hypertension or hypercholesterolemia, the other was more likely to have the same condition [5–9]. Although these results are complementary they are also conflicting; their findings are constrained by study designs that do not allow inference of causation [10, 11]. For diabetes, prospective cohorts and their systematic review found that spouses of individuals who were newly diagnosed with diabetes developed diabetes at much higher rates than individuals whose spouses did not have the condition [9, 12–14]. In contrast, hypertension and hypercholesterolemia have not been assessed in prospective studies. Furthermore, while a spousal concordance might be expected for cardio-cerebrovascular diseases, considering shared risk factors, this relationship has not yet been explored.
This issue is important, as understanding the relationship between spouses’ health could inform targeted strategies for disease prevention and screening. To address this gap, we analyzed data from the Longitudinal Survey of Middle-aged and Elderly Persons in Japan, examining a broad spectrum of NCDs, including diabetes, hypertension, hypercholesterolemia, coronary heart disease, stroke, and cancer. We hypothesized that a spouse’s development of an NCD would increase the likelihood of the same condition developing in their partner within the subsequent three years.
We used data from the Longitudinal Survey of Middle-aged and Elderly Persons conducted by the Japanese Ministry of Health, Labour and Welfare (MHLW). The purpose of this national survey was to follow middle-aged and elderly men and women to continuously investigate changes in their health, employment, and social activities and to obtain basic data for planning and implementing policies for the elderly [15]. We obtained permission from the MHLW to use the data. The study was reviewed and approved by the Research Ethics Committee of the Osaka International Cancer Institute (no. 1508119060).
The study included Japanese people who were aged 50–59 years as of 31 October 2005; these individuals were selected by a two-stage random sampling procedure. Specifically, the MHLW collected health-related information using self-administered questionnaires from 276,682 households that were randomly sampled from 5280 districts in Japan. From these data, 2515 districts were randomly sampled for the Longitudinal Survey of Middle-aged and Elderly Persons. The sample was not weighted. The survey was conducted every year with the same participants via face-to-face interviews (1st–5th year) or via mail (6th–12th year). In the first survey, 33,815 residents responded, and were followed-up. The number of respondents in the 2nd–12th waves were 31,403, 29,772, 28,492, 27,591, 25,157, 23,672, 22,288, 21,556, 20,680, 20,101 and 19,513. Participants were included for these analyses if they were married and lived with their spouse, and both spouses responded to the questionnaire in the first wave.
Variables Explanatory variablesThe explanatory variables were the spouse’s development of any of the six following conditions: diabetes, hypertension, hypercholesterolemia, CHD, stroke, and cancer. Participants were asked whether they had been diagnosed with any of these conditions by doctors at during the previous year in all 12 waves. When one spouse answered that he or she had not been diagnosed in previous waves but had been newly diagnosed during the most recent year, they were regarded as developing the condition. In the first questionnaire in 2005, participants were asked whether they had been previously diagnosed with any of the conditions (not during the previous year). Therefore, 2006 was the first year when participants could be considered to have newly developed a condition. If participants developed a condition twice, they were excluded the second time.
Outcome variablesThe outcome of interest for this study was the spouses’ development of diabetes, hypertension, hypercholesterolemia, CHD, stroke, or cancer within 3 years of their partner developing the same condition [12].
CovariatesWe measured the following potential covariates: age (continuous), sex, education (junior high school, high school, technical school or junior college, or university or higher), household income (0–30,000, 30,001–60,000, 60,001–90 000, or 90,001+ USD), employment status (office worker, manual worker, unknown or unemployed), number of cigarettes smoked each day (0, 1–10 cigarettes/day, 11–20 cigarettes/day, 21–30 cigarettes/day, or 31+ cigarettes/day), alcohol intake (every day, 1–6 times/week, 1–3 times/month, rarely or never) and physical exercise (active and participated in moderate [e.g., walking or jogging] or vigorous [e.g., aerobics or swimming] physical activity at least once a week, or inactive).
Statistical analysisIn each analysis, individuals were included if they had no diagnosis at baseline and had information on their spouse’s status from the previous year. We created discrete-time design modules, in which each wave of participants was treated as an analytical unit. Cases were participants whose spouse had developed a condition within one year. Controls were participants whose spouse did not develop a condition throughout the survey period (Fig. 1).
Study design. One participant whose spouse developed the disease, along with five propensity score-matched controls whose spouses were not diagnosed, are followed up for three years.
Participants’ spouses’ development of condition was not equally scattered, so we conducted a propensity score (PS) analysis to evaluate the association with individual development of the disease. PS matching was used to account for the differences. PS (the probability of spouses’ development of diabetes, hypertension, hypercholesterolemia, CHD, stroke and cancer for each participant ranging from 0–1) was calculated by multivariable logistic regression using the following measured covariates; age, sex, education, household income, employment status, physical activity, smoking status, drinking status. Using this score, we selected five matched couples to enlarge the power without compromising the quality of the matches [12]. We used a nearest neighbor matching algorithm without replacement. To judge the success of each PS matching in terms of creating groups that looked similar on the observed covariates, we used standardized differences; the difference in proportions between the exposed and control groups divided by the standard deviation in the exposed group. Generally, a standard difference of 0.1 indicates a potentially meaningful imbalance (Supplementary 1).
When PS matching created an acceptable balance, incidence proportion and odds ratios (ORs) were calculated. We used a conditional logistic regression model and calculated the incidence of the diseases within 3 years with or without the other spouse’s prior diagnosis [12].
First, we determined the association between one spouse’s prior development of a condition and the onset of the same condition in the other spouse within 3 years of the development. We calculated ORs for each condition (main analysis). Additionally, we examined the association between one spouse’s development of the above-mentioned six conditions and the onset of these six conditions individually in the other spouse (secondary analysis). Therefore, we created 36 (6 × 6) case and control groups. All analyses were performed using SAS version 9.4 software (SAS Institute Inc, Cary, NC).
Of 33,815 participants in the survey, 18,946 (9,473 couples) were married and living together. In some cases from the first survey in 2005, answers were not obtained from both couples; therefore, 112 (56 couples) were removed, with 18,834 (9,417 couples) remaining for analysis (Table 1). The discrete-time number was 118,876. From that, 1,374 cases in which one spouse developed diabetes in the previous year were matched with 6,870 controls with no diagnosis of diabetes. Similarly, 2,657 cases with hypertension were matched with 13,285 controls without hypertension, 3,321 cases with hypercholesterolemia were matched with 16,605 controls without hypercholesterolemia, 567 cases with stroke were matched with 2,835 controls without stroke, 1093 cases with CHD were matched with 5,465 controls without CHD, and 975 cases with cancer were matched with 4,875 controls without cancer (Table S1).
| Husbands (N = 9,417) | Wives (N = 9,417) | |||
|---|---|---|---|---|
| n | % | n | % | |
| Age | ||||
| 50–54 | 5675 | 60.3 | 3048 | 32.4 |
| 55–59 | 3742 | 39.7 | 6369 | 67.6 |
| Job | ||||
| Not employed | 432 | 4.4 | 2858 | 29.4 |
| Manual work | 2433 | 25.0 | 1293 | 13.3 |
| Office work | 5923 | 61.0 | 4303 | 44.3 |
| Unknown | 619 | 6.4 | 950 | 9.8 |
| Missing | 10 | 0.1 | 13 | 0.1 |
| Household annual income (USD) | ||||
| 0–30 000 | 1944 | 20.6 | 1944 | 20.6 |
| 30 001–60 000 | 4014 | 42.7 | 4014 | 42.7 |
| 60 001–90 000 | 1316 | 14.0 | 1316 | 14.0 |
| 90 001– | 679 | 7.2 | 679 | 7.2 |
| Missing | 1464 | 15.5 | 1464 | 15.5 |
| Education | ||||
| Junior high school | 1673 | 17.2 | 1339 | 13.8 |
| High school | 4194 | 43.2 | 4609 | 47.4 |
| Technical school or junior college | 691 | 7.1 | 2210 | 22.8 |
| University (4 years) or more | 2137 | 22.0 | 618 | 6.4 |
| Missing | 722 | 7.4 | 641 | 6.6 |
| Smoking | ||||
| Non-smoker | 4875 | 50.2 | 8134 | 83.7 |
| <10 cigarette per day | 429 | 4.4 | 385 | 4.0 |
| 11–20 cigarette per day | 1882 | 19.4 | 551 | 5.7 |
| 21–30 cigarette per day | 1559 | 16.0 | 148 | 1.5 |
| 31– cigarette per day | 621 | 6.4 | 28 | 0.3 |
| Missing | 51 | 0.5 | 171 | 1.8 |
| Alcohol drinking | ||||
| Excessive drinker* | 4090 | 42.1 | 1002 | 10.3 |
| Moderate drinker** | 4353 | 44.8 | 4835 | 49.8 |
| Non-drinker | 1212 | 12.5 | 3851 | 39.6 |
| Missing | 59 | 0.6 | 26 | 0.3 |
| Physical activity | ||||
| Active† | 2391 | 24.6 | 2674 | 27.5 |
| Inactive‡ | 6513 | 67.0 | 6234 | 64.2 |
| Missing | 513 | 5.3 | 509 | 5.2 |
| Comorbid conditions | ||||
| History of diabetes | 897 | 9.2 | 391 | 4.0 |
| Development of diabetes during follow-up | 1259 | 13.0 | 707 | 7.3 |
| History of Hypertension | 1868 | 19.2 | 1324 | 13.6 |
| Development of hypertension during follow-up | 2671 | 27.5 | 1919 | 19.8 |
| History of hypercholesterolemia | 856 | 8.8 | 707 | 7.3 |
| Development of hypercholesterolemia during follow-up | 2345 | 24.1 | 2296 | 23.6 |
| History of stroke | 160 | 1.6 | 70 | 0.7 |
| Development of stroke during follow-up | 525 | 5.4 | 280 | 2.9 |
| History of ischemic heart disease | 368 | 3.8 | 138 | 1.4 |
| Development of ischemic heart disease during follow-up | 949 | 9.8 | 537 | 5.5 |
| History of cancer | 151 | 1.6 | 181 | 1.9 |
| Development of cancer during follow-up | 859 | 8.8 | 636 | 6.5 |
*Excessive drinker: heavy drinker who consumes 15 drinks or more per week for men or 8 drinks or more per week for women, or binge drinker who consumes 5 or more on a single occasion for men or 4 or more drinks on a single occasion for women
†Active: take part in moderate (such as walking or jogging) or vigorous (such as aerobics or swimming) physical activity 4–5 times/week or more
During the three-year follow-up, 3.6% of propensity-score matched participants whose spouse did not develop diabetes, and 6.8% of participants whose spouse had developed diabetes, developed diabetes. The corresponding incidences were 11.5% and 13.5% for hypertension, 11.1% and 17.1% for hypercholesteremia, 1.4% and 2.1% for stroke, 3.4% and 4.8% for CHD and 3.6% and 3.9% for cancer. For diabetes, hypertension hypercholesterolemia, and CHD participants whose spouse had been developed the condition had significantly greater odds of developing the same condition within 3 years compared to participants whose spouse had not been diagnosed with the condition. The ORs (95% confidence interval [CI]) were 1.96 (1.53–2.50), 1.20 (1.06–1.36), 1.63 (1.47–1.81) and 1.43 (1.05–1.95), respectively (Table 2). The same association was not observed for stroke [1.69 (0.80–3.58)] and cancer [1.08 (0.75–1.54)].
| Total number | Participant’s incidence of diabetes, n (%) |
Odds ratio * | |
| Spouse without prior development of diabetes | 6870 | 246 (3.6) | 1 [ref] |
| Spouse with prior development of diabetes | 1374 | 93 (6.8) | 1.96 (1.53–2.50) |
| Total number | Participant’s incidence of hypertension, n (%) |
||
| Spouse without prior development of hypertension | 13285 | 1527 (11.5) | 1 [ref] |
| Spouse with prior development of hypertension | 2657 | 359 (13.5) | 1.20 (1.06–1.36) |
| Total number | Participant’s incidence of hypercholesterolemia, n (%) |
||
| Spouse without prior development of hypercholesterolemia | 16605 | 1843 (11.1) | 1 [ref] |
| Spouse with prior development of hypercholesterolemia | 3321 | 567 (17.1) | 1.63 (1.47–1.81) |
| Total number | Participant’s incidence of stroke, n (%) |
||
| Spouse without prior development of stroke | 2835 | 40 (1.4) | 1 [ref] |
| Spouse with prior development of stroke | 567 | 12 (2.1) | 1.69 (0.80–3.58) |
| Total number | Participant’s incidence of heart disease, n (%) |
||
| Spouse without prior development of CHD | 5465 | 188 (3.4) | 1 [ref] |
| Spouse with prior development of CHD | 1093 | 53 (4.8) | 1.43 (1.05–1.95) |
| Total number | Participant’s incidence of cancer, n (%) |
||
| Spouse without prior development of cancer | 4875 | 177 (3.6) | 1 [ref] |
| Spouse with prior development of cancer | 975 | 38 (3.9) | 1.08 (0.75–1.54) |
CHD, coronary heart disease.
(*) The odds ratio was calculated using conditional logistic regression following propensity score matching.
When a participant’s spouse was diagnosed with diabetes, the participant had high odds of developing only diabetes but not other conditions (Table 3). When a participant’s spouse was diagnosed with hypertension, the participant had high odds of developing hypertension and hypercholesterolemia [1.15 (1.03–1.28)]. When a participant’s spouse was diagnosed with hypercholesterolemia, the participant had high odds of developing hypercholesterolemia, stroke [1.29 (1.00–1.65)] and diabetes [1.20 (1.02–1.41)]. In no case was a spouse’s prior development of a condition associated with the participant’s development of cancer.
| Spouse’s prior development | OR (95% CI) of developing Diabetes | OR (95% CI) of developing Hypertension | OR (95% CI) of developing Hypercholesterolemia | OR (95% CI) of developing Stroke | OR (95% CI) of developing CHD |
OR (95% CI) of developing Cancer |
|---|---|---|---|---|---|---|
| Diabetes | 1.96 (1.53–2.50) | 1.17 (0.97–1.42) | 1.11 (0.94–1.32) | 1.43 (0.94–2.17) | 1.18 (0.88–1.59) | 0.83 (0.60–1.13) |
| Hypertension | 1.21 (0.98–1.46) | 1.20 (1.06–1.36) | 1.15 (1.03–1.28) | 1.28 (0.96–1.69) | 1.16 (0.95–1.40) | 1.05 (0.86–1.27) |
| Hypercholesterolemia | 1.20 (1.02–1.41) | 0.95 (0.85–1.07) | 1.63 (1.47–1.81) | 1.29 (1.00–1.65) | 1.03 (0.86–1.23) | 1.16 (0.98–1.39) |
| Stroke | 1.38 (0.89–2.14) | 1.16 (0.85–1.58) | 1.51 (1.15–1.96) | 1.69 (0.80–3.58) | 1.61 (1.02–2.55) | 1.47 (0.95–2.26) |
| CHD | 0.90 (0.64–1.27) | 1.13 (0.90–1.41) | 1.24 (1.02–1.50) | 2.13 (1.39–3.25) | 1.43 (1.05–1.95) | 0.97 (0.68–1.36) |
| Cancer | 1.19 (0.85–1.66) | 1.11 (0.88–1.40) | 1.16 (0.94–1.43) | 1.05 (0.63–1.75) | 1.12 (0.79–1.59) | 1.08 (0.75–1.54) |
OR, odds ratio; CI, confidence interval; CHD, coronary heart disease.
The odds ratio was calculated using conditional logistic regression following propensity score matching.
This prospective cohort study of couples showed that when an individual developed diabetes, hypertension, hypercholesterolemia, or CHD, their spouse was more likely to develop the same condition within 3 years. Previous cohort studies and meta-analyses have shown that similar associations were observed only in diabetes [12]. Our research extended the evidence showing that this was not only the case for diabetes [OR 1.96 (1.53–2.50), incidence proportion 6.8% (case) and 3.6% (control)], but also hypertension [1.20 (1.06–1.36), 13.5% and 11.5%], hypercholesterolemia [1.63 (1.47–1.81), 17.1% and 11.1%] and CHD [1.43 (1.05–1.95), 4.8% and 3.4%] are more likely to occur in spouses of already diagnosed counterparts.
Individuals who live together are likely to engage in many of the same activities, such as eating, physical activity, watching television and smoking, and also experience the same environmental factors [16–18]. Because couples generally have different genetic makeups, our results indicated that shared environmental and behavioral factors could account for the development of diseases. The significant concordance in diabetes, hypertension and hypercholesterolemia suggested that the environment shared by couples had an important role in development of disease. A consequent association for CHD, for which diabetes, hypertension and hypercholesterolemia are risk factors, was also found [19]. However, no such association was found in the development of cancer, a condition that is influenced by both genes and environment differently, depending on the cancer site [20, 21]. Given that cancer was less attributable to lifestyle compared to other NCDs, the lack of association between an individual’s development of cancer and their spouse’s concurrent development is consistent with expectations.
Awareness of an increased risk of developing the same condition as one’s spouse may encourage individuals to adopt preventive behaviors and engage in regular screenings. For public health personnel, a couple-centered approach for spouses, where one spouse has been diagnosed with a noncommunicable disease, could enhance prevention efforts. Our findings also suggest that physicians should consider not only patients’ medical histories but also those of their spouses to comprehensively assess health risks. For instance, if a patient’s spouse has diabetes, the physician can highlight the patient’s elevated risk and recommend focused preventive measures, such as intensified blood glucose monitoring.
Our results further revealed that when a spouse developed a specific condition, the other was more likely to develop the same condition rather than a different one. Although not statistically significant, the odds ratios for spouses both having diabetes, hypertension, or high cholesterol were higher than for other conditions. This suggests that shared risk may stem not only from similar lifestyle habits but also from specific disease-related factors.
While lifestyle changes prompted by a spouse’s health status—such as adjustments to diet or physical activity—can support overall health, these general changes may have limitations. Effective prevention and management may require strategies tailored to each condition’s unique risk factors. Specifically, preventing diabetes, hypertension, and high cholesterol may benefit from individualized approaches that consider the distinct causes and environmental factors linked to each disease.
LimitationsThis study has several limitations. First, participants with a newly-diagnosed spouse may have been more aware of the early symptoms of a particular condition, and this could have made them more likely to consult their physician and be screened. We adjusted for whether participants had health checkups, but there still may have been overestimation. Second, diagnosis of a condition was self-reported by the participants and this may have caused misclassification. Third, the participants were all selected from Japanese opposite-sex couples, aged 50–59 years in 2005 (1st wave), and the results may not be applicable to other populations or other age groups. Fourth, the rate of missing outcome data was 1%–2%. Couples who had separated or in which one spouse had died or dropped out of the study during the study interval were excluded. This may have resulted in an underestimation of the observed associations because couples who dropped out may have been more likely to develop the diseases. Finally, some of the important covariates, including calory intake, may be missed, or there are residual confounders.
We considered six conditions and found that if a spouse was diagnosed with diabetes, hypertension hypercholesterolemia or CHD, the other spouse, who might share the same risk factors, had a higher chance of developing the same condition. This finding suggests a need for increased attention to the spouses of individuals who are diagnosed with one of the above-mentioned conditions.
non-communicable disease
CHDcoronary heart disease
ORodds ratio
CIconfidence interval
We used data from the Longitudinal Survey of Middle-aged and Elderly Persons conducted by the Japanese Ministry of Health, Labour and Welfare (MHLW). The purpose of this national survey was to follow middle-aged and elderly men and women to continuously investigate changes in their health, employment, and social activities and to obtain basic data for planning and implementing policies for the elderly. We obtained permission from the MHLW to use the data. The study was reviewed and approved by the Research Ethics Committee of the Osaka International Cancer Institute (no. 1508119060).
AcknowledgmentsWe thank Dr Julia Mortimer for her English language editing.
Conflict of interest statementThe authors declare that they have no conflicts of interest.
FundingThe study is supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI Grants (15K19256, 18H03062 and 21H04856).
Data availabilityData cannot be shared for privacy or ethical reasons.
Consent for publicationAll authors have provided their consent for publication.
Authors’ contributionsStudy conception and design: UT and TT. Data acquisition: TT. Data analysis and interpretation: UT, IH. Drafting manuscript: UT. Critical revision: TT and IH. All authors read and approved the final manuscript.
