Thursday September 8 2011
Many factors influence parents' risk of death
The BBC reported that “Bereaved parents die of 'broken heart'”, and went on to say that “Parents who lose a baby before its first birthday are more likely to die early themselves”.
This news story is based on a study that examined the association between experiencing the death of a child within its first year of life and parental risk of death. The researchers compared the mortality risk between parents whose child had been stillborn or died within one year and parents whose child had lived beyond one year.
The research found that parents who lost a child within its first year of life were at two to four times the risk of dying within 15 years compared with parents whose child had not died.
The design of this study means that, although it can show an association between two factors, it cannot prove that one caused the other. In addition, the size of the increase in risk was, in terms of absolute risk, comparatively small. Further research in a larger sample and with more information on the parents is needed.
Where did the story come from?
The study was carried out by researchers from the University of York and the University of Stirling. Funding was provided by the Carnegie Trust for the Universities of Scotland. The study was published in the peer-reviewed journal BMJ: Supportive and Palliative Care.
Generally, the media reported the results of the study accurately. The BBC appropriately reported that this study cannot tell us whether or not the death of a child causes an increased risk of parental death, and that the relationship may be accounted for by factors such as poor parental health. However, the headline claiming that these parents die of a ‘broken heart’ may be misleading.
What kind of research was this?
This retrospective cohort study examined the association between experiencing the death of a child and the parents’ risk of death. Using death registers, the researchers randomly selected a sample of parents whose child had lived for a year or more and parents whose child had been stillborn or died within its first year of life. They then compared the risk of death between the two groups of parents.
A retrospective cohort study is able to describe associations between two factors (in this case, death of a child and parents’ risk of death) but cannot determine causation. There may be several factors that influence a parent’s risk of dying that researchers are not able to account for after the fact, some of which may contribute to both infant and parental mortality, such as parental ill health.
What did the research involve?
The researchers selected two groups of parents, those whose child had been stillborn or died during their first year of life, and those whose child had lived through their first year. They then determined which of these parents had died by 2006, calculated the risk of dying for each of the groups, and compared the risks to determine whether or not parents who had lost a child were at increased risk of dying themselves.
The study was conducted in two parts. In part one the researchers collected data from the Scottish Longitudinal Study (SLS), which links data from a variety of administrative sources, including census data, vital events registries, such as birth and death registries, and NHS data. The researchers drew a sample of 270,385 people from the 1991 census, which accounted for 5.3% of the Scottish population. From this sample, they determined how many of the people had had a child (either live or stillbirth), and then further divided the group into non-bereaved parents (those whose child had not died during the first year of life) and bereaved parents (those whose child was stillborn or died during their first year of life).
The researchers then analysed the dataset to determine how many of the parents in each group had died by 2006. Using these data, they calculated the chance of death for each of the groups within 15 years of the death (for bereaved parents) or birth (for non-bereaved parents) of the child. They then compared these two risks to determine whether or not bereaved parents were at a higher risk of dying within the 15 years after losing a child.
In part two of the study, the researchers used data from the Office for National Statistics Longitudinal Study Dataset, which contains information from the 1971, 1981 and 1991 censuses for about 1% of the English and Welsh populations. Information in this dataset is linked to data from a variety of other sources, including birth and death registries; however, it was only possible to link information for the mothers in this part of the study.
The researchers collected and analysed data separately for each census year cohort (or group). Similar to the methods used in the first part of the study, they determined how many of the people in the sample had had a child, and then further divided the group into bereaved and non-bereaved mothers. For each census group, they then determined how many of the bereaved and non-bereaved mothers had died by 2006. They used these data to calculate the chance of dying in each group of mothers over 15, 25 and 35 years (for the 1991, 1981 and 1971 census cohorts, respectively). As with the first part of the study, they compared the two groups to determine whether or not bereaved parents were at higher risk of dying, and whether or not this risk was continuously high over many years.
What were the basic results?
In the Scotland study, the researchers identified 738 bereaved parents and 50,132 non-bereaved parents. By 2006, they found that:
- Of the bereaved parents, 15 had died or been widowed. This represents a 2.03% risk of death.
- Of the non-bereaved parents, 482 had died or been widowed. This represents a 0.96% risk of death.
- This equated to bereaved parents being more than twice as likely to die (at 2.11 times greater risk) in the 15-year time period than non-bereaved parents (relative risk 2.11, 95% confidence interval 1.27 to 3.52).
In the England and Wales study, the researchers identified:
- 1,272 bereaved and 40,524 non-bereaved mothers in the 1971 cohort
- 827 bereaved and 40,381 non-bereaved mothers in the 1981 cohort
- 662 bereaved and 39,969 non-bereaved mothers in the 1991 cohort
For the 1971 cohort, the researchers found that by 2006:
- Of the bereaved mothers, 178 had died or been widowed. This represents a 14% risk of death over 35 years.
- Of the non-bereaved mothers 4,489 had died or been widowed. This represents an 11.3% risk of death over 35 years.
- This equated to bereaved mothers having a 1.24 times greater risk of death than non-bereaved mothers.
For the 1981 cohort, the researchers found that by 2006:
- Of the bereaved mothers, 50 had died or been widowed. This represents a 6% risk of death over 25 years.
- Of the non-bereaved mothers, 1,623 had died or been widowed. This represents a 4% risk of death over 25 years.
- This equated to bereaved mothers having a 1.5 times greater risk of death than non-bereaved mothers.
For the 1991 cohort, the researchers found that by 2006:
- Of the bereaved mothers, 40 had died or been widowed. This represents a 6% risk of death over 15 years.
- Of the non-bereaved mothers, 509 had died or been widowed. This represents a 1.3% risk of death over 15 years.
- This equated to bereaved mothers having a 4.74 times greater risk of death than non-bereaved mothers.
How did the researchers interpret the results?
The researchers concluded that “bereaved parents are between two and four times more likely to die or become widowed in the first 10 years after the experience of stillbirth or the death of their child than non-bereaved parents”. They say that this risk reduces over time, but that even at 35 years mothers who have lost a child still have a higher risk of death than non-bereaved mothers.
They say that further large-scale studies should be conducted that analyse the data according to the cause of parents’ deaths, and to determine whether or not there are additional risk factors that are associated with the parents’ deaths.
This study examined the association between parental grieving and risk of death. Due to the study design, it is not possible to say conclusively that losing a child causes increased risk of death.
The study has several limitations that are important to note:
- The results presented in the study and newspapers represent the risk to bereaved parents relative to the risk to non-bereaved parents. While these results point to a 2-4-fold increase in risk of death, it is important to remember that the absolute difference in risk was actually relatively low. In Scotland, of the non-bereaved parents, 0.96% died within 15 years of their child’s death compared with 2.03% of the bereaved parents, representing an increase in risk of 1.07%. This was marginally higher at between 2% and 4.7% higher in the England and Wales part of the study.
- While the number of people involved in the study was large overall, the size of the groups differed greatly and there was only a small number of deaths in the bereaved groups. This may mean that the bereaved sample was not representative of the population at large, and caution should be exercised before generalising these findings to the whole of the UK.
- The researchers did not analyse the data according to cause of the parents’ deaths, which they say makes it difficult to determine what puts the parents at increased risk of dying. They say that the small size of their study did not allow them to determine things like the contribution of parental suicide and changes caused by the stress of bereavement, such as a poorly functioning immune system and increased use of alcohol.
- A reverse causation is possible in that underlying factors such as serious health conditions could put both parent and child at increased risk of death.
The researchers point out that this study only considered parents as bereaved if their child had died within one year, thus it is possible that parents were included in the non-bereaved group if their child had died after one year. They also say that the small sample size and lack of background data prevented them from analysing the data by potentially important characteristics, such as socio-economic and health status.
Overall, further research is needed before an accurate picture of the impact of parental grieving on risk of death can been drawn.