Article Text

Download PDFPDF
Physical health conditions and subsequent union separation: a couple-level register study on neurological conditions, heart and lung disease, and cancer
  1. Niina Metsä-Simola1,
  2. Elina Einiö1,
  3. Riina Peltonen1,
  4. Pekka Martikainen1,2,3
  1. 1 Population Research Unit, University of Helsinki, Helsinki, Finland
  2. 2 Laboratory of Population Health, Max-Planck-Institute for Demographic Research, Rostock, Germany
  3. 3 Department of Public Health Sciences, Stockholm University, Stockholm, Sweden
  1. Correspondence to Niina Metsä-Simola, Population Research Unit, University of Helsinki, Helsinki, Finland; niina.metsa-simola{at}helsinki.fi

Abstract

Background Studies that assess the role of physical health conditions on separation risk are scarce and mostly lack health information on both partners. It is unclear how the association between physical illness and separation risk varies by type of illness, gender of the ill spouse and age of the couple.

Methods We used Finnish register data on 127 313 couples to examine how neurological conditions, heart and lung disease, and cancer are associated with separation risk. The data included information on medication, hospitalisations, separations and sociodemographic characteristics. Marital and non-marital cohabiting couples aged 40–70 years were followed from 1998 to 2003 for the onset of health conditions and subsequent separation, and Cox regression was used to examine the associations.

Results Compared with healthy couples, the HR of separation was elevated by 43% for couples in which both spouses had a physical health condition, by 22% for couples in which only the male spouse had fallen ill, and by 11% for couples in which only the female had fallen ill. Among older couples, the associations between physical illness and separation risk were even clearer. The association with separation risk was strongest for neurological conditions, and after incidence of these conditions among males, separation risk increased over time. Adjustment for sociodemographic characteristics had little effect.

Conclusions Our findings suggest that poor health may largely strain relationships through disability and associated burden of spousal care, and this should be taken into consideration when planning support services for couples with physical health conditions.

  • cancer
  • heart disease
  • marital status
  • stroke
  • social epidemiology

Data availability statement

Data may be obtained from a third party and are not publicly available. The data underlying this article were provided by Statistics Finland, the National Institute of Health and Welfare, and the Social Insurance Institution by permission. Researcher may apply for permission to use these data from the register holders. Data from Statistics Finland: tutkijapalvelut@tilastokeskus.fi. Data from the National Institute of Health and Welfare and the Social Insurance Institution: info@findata.fi.

Statistics from Altmetric.com

Introduction

Poor health after divorce is a well-established finding,1–6 but studies that assess the role of poor health for the risk of subsequent divorce are scarce. Most focus on the role of mental health,7–10 although physical illness may be a stressor that reduces relationship quality and thus increases the risk of divorce.11 12 Of the studies that do assess how physical health conditions are associated with the probability of subsequent divorce, only a few have used couple-level data,13–15 although separation is a couple-level event.

Among married Norwegian couples, poor baseline subjective health of either spouse predicted divorce during a 15-year follow-up,14 but in analyses based on the Health and Retirement Study (HRS), neither the wife’s nor the husband’s chronic conditions—measured in terms of heart or lung disease, cancer or stroke—were related to the likelihood of divorce among US couples aged 50 years or more.13 However, in a subsample of those aged less than 61 years and initially healthy, the onset of wife’s chronic conditions, but not those of the husband, were shown to increase the probability of divorce.15 Of the different chronic conditions, only wife’s heart problems and stroke onset were associated with slightly increased probability of divorce. To our knowledge, this study using a sample of 2701 couples and 162 divorces is the only one to examine how the association between physical illness and divorce varies by the type of illness.

Previous studies focusing on the social consequences of specific illnesses have suggested that stroke, multiple sclerosis and cancers affecting the central nervous system are associated with an increased risk of divorce.16–19 Because neurological conditions such as multiple sclerosis, epilepsy, Parkinson’s disease and stroke may cause functional difficulties and severe disability,20 21 these are potential pathways from illness to separation. Of specific cancers, cervical cancer has been indicated to increase divorce risk,22 whereas the findings for testicular cancer are mixed.22–24 Yet, two large population-based studies using Danish25 and Norwegian22 register data found no overall effect of cancer on divorce risk, a finding supported by previous studies with smaller samples.23 24 26 27 All these studies are based on individual-level samples, and having information on the health status of only one spouse may lead to an underestimation of the effect of poor health on separation risk, given that the health status of the other spouse remains unknown. With individual-level data, it is also impossible to examine the joint effects of both spouses’ poor health on separation risk.

We study the effects of various physical health conditions on the risk of subsequent separation using a nationally representative register-based sample of middle-aged Finns. We make the following unique contributions: (1) we use couple-level data on both marital and non-marital cohabiting unions (N=127 313); (2) we identify incidence of physical health conditions—including neurological conditions, heart and lung disease, and cancer—of both spouses using administrative registers; (3) we assess how the risk of separation varies depending on the gender of the ill spouse and whether both spouses having a physical health condition has an effect on separation risk on top of the individual risks of the male or female spouse having a health condition; (4) we examine whether separation risk varies depending on the specific physical health condition and by time since its occurrence and (5) we assess if the associations between the gender of the ill spouse, specific physical health conditions and separation risk depend on the age of the couple.

Methods

We use register-based data on Finnish couples, derived from a two-stage random sample. First, a simple random sample covering 14% of individuals aged 40 or over and living in private households at the end of 1997 was drawn from population registers. Second, all household members of the selected individuals were added to the sample. We included all marital and non-marital cohabiting couples with both spouses aged 40–70. Statistics Finland classifies persons living in the same dwelling and of different genders, not living with a married spouse, not being siblings, and with an age difference not exceeding 15 years as non-marital cohabiters.

The register-based data from Statistics Finland include dates of separation from 1998 to 2003, annual sociodemographic information from 1995 to 2003, and mortality follow-up until the end of year 2003. These data were further linked to hospital records, reimbursement for drug costs and purchases of prescription medication from 1995 to 2003. The information on hospitalisations was provided by the National Institute of Health and Welfare, and information on medication purchases and reimbursement by the Social Insurance Institution. The data linkage was done by Statistics Finland using personal identification codes (the Ethics Committee of Statistics Finland’s permission TK-53-373-09).

Physical health conditions were defined using the 10th revision of the International Statistical Classification ofDiseases and Related Health Problems (ICD-10) codes of hospital care periods, the AnatomicalTherapeutic Chemical (ATC) ClassificationSystem codes of prescription medication purchases, and the Finnish special refund right categories (online supplemental table 1). Neurological conditions covered diseases of the nervous system such as multiple sclerosis and Parkinson’s disease, as well as stroke, following the new ICD-11 classification.28 Heart and lung disease included severe diseases such as ischaemic heart disease and chronic obstructive pulmonary disease, but excluded less severe diseases such as hypertension. Cancer included both malignant neoplasms and in situ neoplasms that had required overnight hospital care.

Supplemental material

We adjusted for several covariates. Age (in 1-year intervals), education (tertiary, intermediary and basic) and social class (upper non-manual, lower non-manual, manual and other) of both spouses, and housing tenure (whether homeowner or not) and union type (marital or non-marital cohabiting union) of the couple were all measured at baseline (end of year 1997, except 1995 for social class). Employment status (employed, unemployed, retired and other) of both spouses, household income (quintiles) and coresident children (none or at least one under 18) were measured annually.

Design, participants and statistical analyses

All couples in the sample were living together at the end of year 1997. We first assessed baseline physical health conditions for all these couples—that is, for both spouses—using hospital records and information on prescription medication from the years 1996 and 1997. After excluding couples with baseline physical ill health, the remaining 127 313 couples were followed for new incidence of physical health conditions and subsequent separation from 1 January 1998 to 31 December 2003. Separation refers to the date when either partner moves out of the common household or date of formal divorce for married couples, whichever is the earliest.

We used Cox proportional hazards regression models to analyse how poor health is related to the risk of separation. Because the risk of separation varies by union duration, time since entry into the union was used as the underlying timescale in all models.29 30 We had information on the exact dates of marriages since 1946, and the exact dates of entry into non-marital cohabiting unions from 1987. For 0.7% of married couples, the exact date of marriage could not be found, so for them we used information on entry into cohabitation. After that, the date of union formation was missing for 0.3% of married couples and for 10% of non-married cohabiting couples. These 1544 couples were excluded from the analyses, leaving 116 408 married couples and 10 905 non-marital cohabiting couples in the sample.

The outcome variable in the Cox models was the time from the beginning of union formation until the date of separation, and the follow-up episodes were left-truncated at 1 January 1998. Censoring occurred at death of either spouse (5244 couples), emigration (454), entry into institutional care (689) or the end of the follow-up (31 December 2003) (112,285). Individuals with physical health conditions are more likely than others to enter institutional care, but after entry into care, non-marital separations can no longer be identified. Dates of formal divorce were available and sensitivity analyses were performed to estimate the effect of censoring (online supplemental table 2). Physical health conditions—cardiovascular diseases in particular—may also co-occur with psychiatric morbidity,31 which is known to predict separation.8–10 We, thus, carried out sensitivity analyses to examine the effect of pre-existing psychiatric morbidity on our results by excluding all couples with baseline (1996–1997) psychotropic drug use and psychiatric hospitalisations (online supplemental table 3).

We used physical health conditions of both spouses as time-varying binary covariates, and analysed both the effects of either the male or the female spouse having a health condition, and the effect of both spouses experiencing a health condition. We present the effects of both spouses being ill only for analyses with all physical health conditions combined. In the condition-specific analyses, we censor couples in which both spouses experience the same physical health condition, because the number of separations among these couples is too small (8–20 depending on specific condition) to allow reliable estimation.

We first adjusted for the age of both the male and the female spouse, and then for all other covariates. We present results for all physical health conditions combined and separately for neurological conditions, heart and lung disease, and cancer and then by age group. The age groups are based on the age of the older spouse (usually the man). Age is related to life circumstances that change over the life course. At the age of 40–70 a common life event is children moving out of the family home, with older couples less often having children living with them. We, thus, conducted sensitivity analyses to examine whether the effects of physical health on separation vary for couples living with or without coresident children (online supplemental table 4).

To assess whether separation risk varies by time since the incidence of specific physical health conditions, we used variables measuring time since the incidence of diseases divided into three groups: no new physical health conditions experienced, less than 2 years since the incidence and two or more years since the incidence. Stata (StataCorp. 2019. Stata Statistical Software: Release V.16.: StataCorp) software was used for all the analyses, and the results are presented as HR.

Results

We observed a total of 8641 separations during the follow-up, the number of couples with new incidence of physical health conditions being 35 523. There were 3525 couples in which both spouses fell ill during the follow-up, and new health conditions were slightly more common among male than female spouses (table 1). Males most often developed heart and lung disease, whereas neurological conditions were more common among females.

Table 1

Characteristics of the study cohort, Finnish couples aged 40–70

When age was adjusted for, the HR of separation was 1.11 (95% CI 1.00–1.22) when the female spouse developed any physical health condition, whereas the corresponding effect for the male spouse was 1.22 (95% CI 1.11 to 1.33) (table 2). When both spouses had a physical health condition, the HR of separation was 1.43 (95% CI 1.12 to 1.84), compared with healthy couples. Neurological conditions had the strongest effect on separation risk. The age-adjusted HR was 1.49 (95% CI 1.33 to 1.68) when the male spouse fell ill and 1.28 (95% CI 1.12 to 1.46) when it was the female spouse. Cancers and heart and lung disease were not associated with separation risk. Had we not censored couples when either partner entered institutional care, the association between neurological conditions and the risk of subsequent separation would have been even stronger (online supplemental table 2). The main results, thus, provide conservative estimates.

Table 2

HR of separation (HR) by physical health conditions, Finnish couples aged 40–70 and not in institutional care

Excluding couples with baseline psychiatric morbidity attenuated the associations, so that the age-adjusted HR of separation when the female spouse developed any physical health condition was 1.05 (95% CI 0.92 to 1.12), and 1.11 (95% CI 0.99 to 1.25) when it was the male spouse (online supplemental table 3). However, the effect of neurological conditions on separation risk remained clear in this subset of the study population, particularly for men, with the age-adjusted HR of separation being 1.38 (95% CI 1.18 to 1.62) after the male spouse fell ill.

When the male spouse developed any physical health condition, the separation risk was similar, regardless of time since its occurrence (table 3). However, when the female spouse fell ill, the risk of separation was slightly larger during the first 2 years following the disease and decreased with time. The association between male spouses’ neurological conditions and separation risk increased with time since incidence, from 1.44 (95% CI 1.24 to 1.66) during the 2 years after the incidence to 1.60 (95% CI 1.34 to 1.91) more than 2 years after it. In contrast, cancer of the male spouse was associated with separation risk only during the first 2 years following the male spouse’s cancer diagnosis.

Table 3

HR of separation (HR) by physical health conditions and time since new disease incidence, Finnish couples aged 40–70 and not in institutional care

Among older couples, physical health conditions among male spouses seemed to increase the risk of separation more than among younger couples, but for female spouses a similar age-specific effect was not found (table 4). When the older spouse was aged 55–70 years, the age-adjusted HR of separation was 1.08 (95% CI 0.89 to 1.31) when the female spouse had any physical health condition, compared with 1.12 (95% CI 1.00 to 1.25) among younger couples. However, among older couples, the age-adjusted HR of separation was 1.30 (95% CI 1.11 to 1.52) when the male spouse had any physical health condition, compared with 1.18 (95% CI 1.06 to 1.32) among younger couples. When both spouses had a physical health condition, the age-adjusted HR for separation was 1.58 (95% CI 1.09 to 2.29) among older couples and 1.34 (95% CI 0.96 to 1.87) among younger couples. The gender difference between older and younger couples was mainly due to the particularly large effect of neurological conditions on separation risk among older couples when the male spouse fell ill, the age-adjusted HR being 1.72 (95% CI 1.38 to 2.15). Among older couples, the male spouse’s cancer was also associated with an increased risk of separation, the age-adjusted HR being 1.43 (95% CI 1.08 to 1.91), whereas cancer of the female spouse, and among younger couples cancer of either spouse, was not related to separation risk. There were no clear differences in the HRs of separation between couples living with co-resident children and couples not living with children (online supplemental table 4). In particular, the large effect of the male partner’s neurological condition on separation risk was very similar in both groups. Adjustment for various socioeconomic characteristics had little effect on any of the associations (tables 2–4).

Table 4

HR of separation (HR) by physical health conditions and age of older spouse, Finnish couples aged 40–70 and not in institutional care

Discussion

Physical health conditions increase the risk of separation among couples aged 40–70 years. Compared with healthy couples, the risk of separation was elevated by 43% for couples in which both spouses had a physical health condition, by 22% for couples in which only the male spouse had fallen ill, and by 11% for couples in which only the female had fallen ill. Among a subsample of older couples, the associations between physical illness and separation risk were even stronger, so that compared with healthy couples, both spouses having a health condition was associated with a 58%, and illness of the male spouse with a 30% increased risk of separation when the older spouse was aged 55–70.

Our results contradict previous findings from the HRS sample which showed that comparable physical health conditions—cancer, heart and lung disease, and stroke—were unrelated to the risk of divorce.13 However, in an initially healthy and younger subsample, heart problems and stroke onset of wives, but not of husbands, predicted a 2%–3% increased risk of divorce between biannual waves. Furthermore, in a Norwegian study using self-reported health measures, baseline poor subjective health of adult married men and women predicted about 30% increase in subsequent divorce risk during a 15-year follow-up, the effect being slightly larger after female illness.14 These divergent findings suggest that the association between physical illness and separation risk depends on the specific illness and gender of the ill spouse, as well as the age group, and highlights the need for future studies that assess the effects of specific physical health conditions in various age groups using couple-level data. Our study further extends previous work by showing that the risk of separation was particularly high for couples affected by neurological conditions such as stroke, Parkinson’s disease, multiple sclerosis and epilepsy, but not elevated after heart or lung disease. The finding is of major policy importance since the prognosis of most neurological conditions, including stroke, have improved in recent years,32 and more couples are thus affected by them.

Previously, physical illness has been viewed as a stressor that reduces relationship quality and thus increases the risk of separation.11 12 Our results suggest that functional problems and disability may be an important pathway, since neurological conditions that may lead to severe disability have the strongest effect on separation risk. While adjustment for socioeconomic factors such as income and employment status had very little effect, it seems that disability leads to separation through other mechanisms.

One specific mechanism may be the increased burden of care falling on the other spouse.33 Our results suggest that physical health conditions among male spouses have somewhat stronger effect on separation risk than these conditions among female spouses, and this gender difference was especially pronounced among older couples. Furthermore, the gender difference was most pronounced for neurological conditions. It is possible that these finding relate to women’s higher expectations to provide spousal care.34 Even when the level of disability increases, ill men are more likely to prefer a home setting with only spousal caregiving, whereas women more readily prefer formal care arrangements.35 Our findings also showed that when the male spouse had a neurological condition, the risk of separation increased with time since falling ill, whereas a similar association was not observed after the female spouse fell ill. This suggests that particularly the long-term care burden falling on the female spouse strains relationships. Future studies should assess in more detail how the level of disability and care arrangements are associated with relationship quality and union stability.

Strengths and limitations

The register-based data provided several advantages, including reduced bias related to self-reporting and minimal loss to follow-up of couples with health problems. The data have excellent quality information on the exact time of separation from both marital and non-marital cohabiting unions, and detailed information on sociodemographic characteristics.

While our main findings suggest that couples in which both spouses were ill experienced higher separation risk than couples with only one ill spouse, in the condition-specific analyses the number of couples in which both spouses experience the same physical health condition was too low to allow reliable estimation. Future studies that examine spousal similarities and dissimilarities in specific health problems are still needed. The comparison between age groups was also limited due to small number of divorces. Our results nevertheless suggest that the association between illness and separation risk may be stronger among older couples. The association between poor health and separation risk may also depend on socioeconomic resources, with illness creating a relatively larger burden among couples already in disadvantaged positions, and we welcome future studies that address these moderating effects in more detail.

We could only follow couples until the end of year 2003, and the association between physical health conditions and separation risk may have changed since. However, there are no major societal changes that would support such a hypothesis; divorce rate in Finland has remained similar for over 20 years,36 and the national welfare system has offered universal health insurance to all residents during the whole time period. Nevertheless, we welcome future studies that aim to assess whether changes in treatment and support available to families affected by illness shape the social consequences of physical illness.

The follow-up time of 6 years may also be too short to examine the long-term effects of illness on separation risk. Our results suggested that the effect of male spouses’ neurological conditions on separation risk increased with time since incidence, and a longer follow-up period would have allowed us to estimate whether this risk increases further, settles to a persistently high level, or declines again after more time has passed. It is also possible that some physical health conditions have no short-term effect on separation risk, but a long-term effect still exists.

The register data did not include information on functional disability, and we were thus unable to test whether long-term disability was the pathway from neurological conditions to separation. Furthermore, the register-based definition of non-marital cohabiting unions may exclude individuals living together but officially having separate addresses. It also excludes same-sex couples and couples with an age difference of more than 15 years. These limitations affect the identification of separations, which are based on either partner moving out of the household. Nevertheless, the register-based definition results in quite a similar prevalence of non-marital cohabitating unions as is obtained in survey samples in Finland,37 38 and should thus not present a major source of bias.

Conclusions

Our findings clearly show that neurological conditions, which may cause functional disabilities, are associated with an increased risk of separation. Because of advanced treatment, people are living longer with these conditions and thus the number of affected couples is on the rise. More research is needed to assess whether these conditions predict separation differently in various contexts. Future studies should also examine how relationship dynamics could be considered in the treatment of physical health conditions to prevent these from leading to separation.

What is already known on this subject

  • Divorce is associated with poor health, but previous evidence on separation risk following physical illness is limited and results are mixed. It is unclear whether separation risk after physical illness varies by the type of illness, gender of the ill spouse and by the age of the couple.

What this study adds

  • Neurological conditions in particular seem to increase the risk of separation. The association appears to be stronger when the male spouse falls ill and also among older couples. The results suggest that disability may be an important pathway through which poor health strains relationships. This should be taken into consideration when planning support services for those with incident health problems.

Data availability statement

Data may be obtained from a third party and are not publicly available. The data underlying this article were provided by Statistics Finland, the National Institute of Health and Welfare, and the Social Insurance Institution by permission. Researcher may apply for permission to use these data from the register holders. Data from Statistics Finland: tutkijapalvelut@tilastokeskus.fi. Data from the National Institute of Health and Welfare and the Social Insurance Institution: info@findata.fi.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors This paper and the results have not been published or submitted elsewhere. There is no financial or other conflict of interest that might bias our work. All authors were involved in the design of the study. NM-S wrote the first draft of the manuscript and RP conducted the statistical analyses. All authors critically commented on the manuscript, and have read and approved the final version and believe that it represents honest work. All authors, and no one else, fulfill the criteria for authorship.

  • Funding This work was supported by the Academy of Finland, grant number 308247.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.