Article Text

other Versions

PDF
Childlessness, parental mortality and psychiatric illness: a natural experiment based on in vitro fertility treatment and adoption
  1. Esben Agerbo1,2,
  2. Preben Bo Mortensen2,
  3. Trine Munk-Olsen2,3
  1. 1CIRRAU—Centre for Integrated Register-based Research, Aarhus University, Aarhus, Denmark
  2. 2National Centrefor Register-Based Research, Aarhus University, Aarhus, Denmark
  3. 3The Danish Clinical Registers – A National Quality Improvement Programme, Aarhus N, Denmark
  1. Correspondence to Professor Esben Agerbo, CIRRAU—Centre for Integrated Register-based Research, Aarhus University, Aarhus DK-8000 C, Denmark; ea{at}ncrr.dk

Abstract

Background Childlessness increases the risk of premature mortality and psychiatric illness. These results might, however, be confounded by factors that affect both the chance of parenthood as well as the risk of premature death and psychiatric illness.

Methods Using population-based health and social registers, we conducted a follow-up study of 21 276 childless couples in in vitro fertility treatment.

Results The crude death rate ratio in women who become mothers to a biological child is 0.25 (95% CI 0.16 to 0.39). In other words, childless women experience a fourfold higher rate of death, that is, 4.02 (2.56 to 6.31). The analogous death rate in fathers is approximately halved: 0.51 (0.39 to 0.68) and 0.55 (0.32 to 0.96) associated with having a biological child and an adopted child, respectively. With substance use disorders being the exception, none of the crude rates of psychiatric illness in parents of a biological child were statistically distinguishable from the rates in the childless. These findings were slightly confounded by age, calendar year, income, education, somatic comorbidity and marital breakup.

Conclusions Mindful that association is not causation, our results suggest that the mortality rates are higher in the childless. Rates of psychiatric illness do not appear to vary with childlessness, but the rate of psychiatric illness in parents who adopt is decreased.

Statistics from Altmetric.com

Introduction

Studies have shown a u-shaped relationship between parity and mortality, with the risk being lowest among those with two children.1 ,2 Few studies have focused on the childless in relation to those with only one child,1 ,3 and it remains unclear whether childlessness has an impact on mortality, or whether childlessness is a mediating factor. Furthermore, research seldom disentangles effects of parenthood from effects of marital status,4–6 or separates the voluntary childless and the involuntary childless.7

Psychiatric illness increases the risk of childlessness,8–10 but few studies have investigated whether childlessness causes psychiatric illness.11–14 The thorny issue lies in separating effects of childlessness from shared selection effects into psychiatric illness and childlessness.15

We have established a cohort of couples in in vitro fertilisation treatment (IVF), and consider the event of becoming a parent as a natural experiment, that is, we assume that becoming a parent varies through some naturally occurring circumstances that are exogenous to the outcome.16 Our goals are to determine whether the rate of death and psychiatric illness is related to having a child, and whether the rate varies across causes of death and psychiatric illnesses.

Methods

Data were obtained by linking population-based registers using the unique personal identification number, which is assigned to all Danes. The Danish Civil Registration System contains dates of births, deaths and emigrations. All biological and legal offspring, including adopted children, are registered. The Danish IVF Register was established on 1 January 1994 and covers all public and private IVF-treatments until 31 December 2005.17 All other registers are updated up to 31 December 2008, ensuring a follow-up of at least 3 years. The Danish Causes of Death Register contains dates and causes of all deaths.18 The Danish Psychiatric Central Register includes all admission dates and diagnoses since 1969 (outpatients since 1994).19 The Danish National Hospital Register holds records of all contacts with somatic hospitals since 1978.20 The Integrated Database for Longitudinal Labour Market Research contains information on income, education and identifiers of spouses and cohabitees.

The cohort consists of women who were registered in the IVF register and, who in the previous year, were living with their partner. Women, or the partner, who were not childless, or had suffered from a psychiatric illness, were excluded. Individuals were followed from entry into the IVF register and until the date of death, the first psychiatric illness, emigration or until the study ends, that is, 31 December 2008. Dates when a child was born or adopted were modelled as time-dependent covariates. Statistics were based on Cox regressions.

Results

Table 1 shows rate ratios of death and psychiatric illness associated with becoming a parent compared with individuals who are childless. In the period from 1994 to 2005, 21 276 childless couples were registered in the IVF register, and 15 210 (1564) children were born (adopted) during the 1 87 000 person-years of follow-up. A total of 96 women and 220 men died, corresponding to death rates of 51 and 117 per 1 00 000 person-years, 710 females and 553 males were diagnosed with a psychiatric illness corresponding to rates of 385 and 299 per 1 00 000 person-years.

Table 1

Rate ratios of death and psychiatric illness in relation to childlessness among 21 276 couples in IVF treatment (95% CI)

The crude death rate ratio in mothers to a biological child is 0.25 (95% CI 0.16 to 0.39), so that childless women experience a fourfold higher rate of death (4.02 (2.56 to 6.31)). Mortality rates of cancer, circulatory diseases and accidents are 0.22 (0.12 to 0.39), 0.22 (0.06 to 0.80) and 0.44(0.06 to 3.25). No mothers died of other causes. There is a decreased death rate associated with adoption: 0.67 (0.32 to 1.38).

The crude death rate ratio in fathers is approximately halved: 0.51 (0.39 to 0.68) and 0.55 (0.32 to 0.96) associated with having a biological child and an adopted child. Cause-specific death rates of cancer, circulatory diseases, accidents and external causes in men with a biological child are 0.59 (0.36 to 0.97), 0.62 (0.33 to 1.17), 0.74 (0.26 to 2.15) and 0.39 (0.16 to 0.94). The age-adjusted male death rate is 70% higher than the female rate: 1.69 (1.31 to 2.16). Childless men experience a rate which is 2.7 times more than the rate in women with a biological child, and thus, less marked than the analogous rate in childless women.

None of the crude rate ratios of psychiatric illness in parents of a biological child are statistically distinguishable from the rates in the childless. Competing risk analyses indicate (pwomen=0.24 and pmen=0.30) that the rates associated with substance use disorders (women: 0.55 (0.28–1.09) and men: 0.61 (0.38–0.98)), psychotic disorders (0.56 (0.23–1.37) and 1.24 (0.54–2.88)), affective disorders (1.02 (0.75–1.38) and 1.03 (0.70–1.50)), neurotic disorders (0.82 (0.64–1.04) and 0.93 (0.69–1.24)), and other psychiatric illnesses (1.15 (0.72–1.84) and 0.73 (0.44–1.21)), are of similar magnitude.

Parents who adopted children had reduced rates of psychiatric illness: the rates are approximately half: 0.52 (0.35–0.78) and 0.46 (0.30–0.73) in women and men, respectively. A decrease is observed across the diagnostic subgroups. Substance use disorders (women, 0.59 (0.17–1.97) and men, 0.34 (0.11–1.11)), affective disorders (0.41 (0.19–0.88) and 0.77 (0.37–1.61)), neurotic disorders (0.53 (0.29–0.98) and 0.42 (0.20–0.91)), and other psychiatric illnesses (0.95 (0.37–2.40) and 0.33 (0.08–1.36)).

The rate ratios appear to be slightly confounded by age and year, perhaps with the exception of death rates in men. This confounding is primarily caused by age. The additional adjustments for education, income and somatic comorbidity have only a marginal impact. Marital breakup, number of IVF treatments, own and spousal psychiatric illness, only have a limited mediating effect.

Discussion

This study finds that men and particularly women who become parents have a decreased rate of death. The decrease is also observed in the cause-specific mortality rates and appears to be slightly confounded by education, income and somatic comorbidity. The rate of psychiatric illness is of the same order of magnitude in parents of a biological child as in the childless, with substance use disorder perhaps being the exception. By contrast, the rate of psychiatric illness is approximately halved in men and women who adopt a child.

This is presumably the first study to use the outcome of IVF treatment as a natural experiment to investigate the impact of childlessness on mortality and psychiatric illness. Besides focusing on the effect of involuntary childlessness, a virtue of our design is that unobserved factors, presumably, are balanced. As our study is based on a natural experiment, the results are less likely to be due to reverse causation or confounding. The event of becoming a parent may not be exogenous, but could rely on residual confounding due to, for instance, unobserved comorbidity, so that parents and non-parents are not exchangeable. However, the rates are relatively unchanged when observed factors are taken into account, which suggests that the event of becoming a biological parent fulfils the criteria for being a natural experiment.

It has been hypothesised that the excess mortality in the nulliparous, and those with only one child, originate from selection effects in the form of poor health, unhealthy behaviour and lower levels of social support and relationships.1 ,3 However, research rarely distinguishes between the voluntary and involuntary childless, and research tends to focus on parity and family size rather than contrasting parents and non-parents. In our study, effects of healthy behaviour, social support and relationships must originate from the event of becoming a parent rather than from a marital partner.

The rate ratios in parents who adopt are surprising. Perhaps this is related to survival bias, as Danish adoption regulations mandate that a fertility treatment must be completed before an adoption process can begin. Alternatively, parents who adopt could be a selected group, as prospective parents’ health and economic condition are evaluated by the National Adoption Board. The rate of adoption is higher among affluent parents, 1.75 (1.46–2.08), but unrelated to the Charlson comorbidity index (p=0.93). Thus, the event of becoming an adoptive parent may not be exogenous given the covariates and the assumption for the natural experiment may be violated.

Our analyses show that childlessness has little impact on the rate of psychiatric illness. This lends support to the conjecture that previous reports of higher risks in the childless8–10 may actually reflect reverse causality, as individuals with an undiagnosed or insidious psychiatric illness are more likely to be childless and, subsequently, diagnosed with a psychiatric illness. However, our results show that individuals who adopt are at lower risk of psychiatric illness. It is likely that this lower rate is an artefact as mentioned above. Consistent with previous studies, our results suggest that the rate of substance use is higher in the childless.

This study should be viewed in the context of its limitations. Besides relying on population-based registers, and the fact that the first three treatment attempts are free of charge in Denmark, the findings may not generalise to other involuntary childless individuals who are not seeking in vitro fertility treatment. Although income and number of IVF treatments are considered in the analyses, affluent individuals could have a higher success rate by purchasing treatment sessions in private clinics, and as affluent individuals tend to live longer and experience less psychiatric illness, this may induce a downward bias in the rate ratios. Our results are adjusted for somatic comorbidity, income and educational attainment which correlate with health and longevity, yet residual confounding may still influence our results. The relative paucity of events is a shortcoming, which suggests that the competing risks analyses should be given less emphasis. Some people with psychiatric illnesses are never hospitalised, and these illnesses are unevenly distributed according to, for example, social class and marital status.

Mindful that association is not causation, our results suggest that the mortality rates are higher in the childless. Rates of psychiatric illness do not appear to vary with childlessness, but the rate of psychiatric illness in parents who adopt is decreased.

What is already known on this subject

  • Research suggests that childlessness increases the risk of premature mortality and psychiatric illness. These results might, however, be confounded by factors that affect both the chance of becoming a parent, as well as the risk of premature death and psychiatric illness

What this study adds

  • Our study is the first to capitalise on the natural experiment of becoming parents, which arises in the cohort of individuals who undergo in vitro fertilisation treatment

  • Mindful that association is not causation, our study suggests that the mortality rates are higher in the childless. By comparison, the mortality rates seem to be less diminished in women who adopt a child

  • Our study suggests that the rate of psychiatric illness is of the same order of magnitude in biological parents as in childless individuals, possibly with the exception of substance use disorder

View Abstract

Footnotes

  • Contributors EA designed the study, acquired the data, did all statistical analyses and wrote all drafts of the paper. TMO helped revising the manuscript. All authors interpreted the findings and approved the final manuscript.

  • Funding The National Centre for Register-based Research is supported by the Stanley Medical Research Institute. Trine Munk-Olsen receives funding from the Danish Medical Research Council (Reference number: 09–063642/FSS). Neither the Stanley Medical Research Institute nor the Danish Medical Research Council had any role in the study design, analysis and interpretation of data, in the writing of the report, or in the decision to publish the paper.

  • Competing interests None.

  • Ethics approval The Danish Data Protectency and the Danish National Board of Health.

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

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.

References