INTRODUCTION
Traditional Understanding of Bereavement and Health
Much of the literature on health outcomes in the bereaved focuses on the approximately 800,000 older people who lose a spouse each year, and suggests that in the months following the loss of a spouse, widows and widowers are at higher risk for depression (Ott et al., Reference Ott, Lueger and Kelber2007), PTSD (Zisook et al., Reference Zisook, Chentsova-Dutton and Shuchter1998), impaired immune response (Phillips et al., Reference Phillips, Carroll and Burns2006), increased use of health care services (Prigerson et al., Reference Prigerson, Maciejewski and Rosenheck2000; Charlton et al., Reference Charlton, Sheahan and Smith2001) increased health care costs (Prigerson et al., Reference Prigerson, Maciejewski and Rosenheck2000), sudden death (Ott et al., Reference Ott, Lueger and Kelber2007), and death from all causes (Hart et al., Reference Hart, Hole and Lawlor2007) than are members of the general population (Clayton, Reference Clayton1973; Jacobs & Ostfeld, Reference Jacobs and Ostfeld1977; Zisook et al., Reference Zisook, Chentsova-Dutton and Shuchter1998; Prigerson et al., Reference Prigerson, Maciejewski and Rosenheck2000; Ott et al., Reference Ott, Lueger and Kelber2007) Many studies have also been completed describing the grief experience of other adults such as caregivers, children, parents, siblings, and friends; however, very few identify health outcomes in the bereaved that may not be detected until months or years after a significant loss. Those studies that have been completed suggest that the nature of the bereaved adult's relationship with, involvement in the care of, and cause of death of the deceased individual significantly affect his or her risk for adverse mental and physical health outcomes (Prigerson et al., Reference Prigerson, Bierhals and Kasl1997; Schulz et al., Reference Schulz, Beach and Lind2001; Brazil et al., Reference Brazil, Bedard and Willison2002). Those with close relationships to the deceased, a high level of involvement in care, and who experience a loss to a sudden or violent death tend to exhibit more adverse outcomes, whereas those with distant relationships, no or little involvement in caregiving, and who experience a loss from a chronic condition or expected circumstance tend to follow the expected grief trajectory more closely.
Though studies report wide variation in the adjustment of bereaved adults as manifested psychological symptoms (including depression and PTSD) and the duration of the grief response (Charlton et al., Reference Charlton, Sheahan and Smith2001; Ott et al., Reference Ott, Lueger and Kelber2007), for most adults the typical grief course occurs in stages, as described by Kubler Ross (Reference Kubler-Ross1969), and results in lower grief scores and higher levels of adjustment over time (Zisook et al., Reference Zisook, Chentsova-Dutton and Shuchter1998; Ott et al., Reference Ott, Lueger and Kelber2007). In his book Bereavement: Studies of Grief in Adult Life, Colin Murray Parkes (Reference Parkes1998) named this time of adjustment “psychosocial transition.” He described the variable concept of bereavement as an adaptation to a different life while cherishing the memory of the person who has died—rather than a process with a defined beginning and ending (Parkes, Reference Parkes1998).
Although the grief course and subsequent health risks appear to be well defined for adults experiencing expected losses such as an aged spouse or parent, it remains unclear whether the experience and subsequent health risks are similar for those experiencing the loss of a child. This review will provide an analysis of the literature surrounding health outcomes in bereaved parents.
Bereaved Parents
Children, who make up about 29% of the United States' population, account for 2% of all the deaths annually (Institute of Medicine, 2004). The most recent publication of National Vital Statistics indicates that approximately 56,000 of them die each year (Hoyert et al., Reference Hoyert, Hsiang-Ching and Smith2005). This figure includes neonates, infants, children, and adolescents who die from a wide range of conditions such as premature birth, SIDS, congenital abnormalities, trauma, neurodegenerative disorders, HIV/AIDS, and cancer (McAliley et al., Reference McAliley, Hudson-Barr and Gunning2000; Carter et al., Reference Carter, Howenstein and Gilmer2004; Rallison & Moules, Reference Rallison and Moules2004). These numbers suggest that, estimating conservatively, about 100,000 American parents are left to grieve the loss of a child each year (Fletcher, Reference Fletcher2002). In many cases, siblings, grandparents, and other family members are also profoundly affected, and Seecharan et al. (Reference Seecharan, Andresen and Norris2004) report that, when including such extended family members, approximately 19% of the U.S. adult population has experienced the death of a child.
In many different studies, parents have described the death of a child as a unique, complicated, stressful, dramatic, profound, disruptive, and devastating experience (Gilliss et al., Reference Gilliss, Moore and Martinson1997; Znoj & Keller, Reference Znoj and Keller2002; Dyregrov, Reference Dyregrov2004; Arnold et al., Reference Arnold, Gemma and Cushman2005; Hinds et al., Reference Hinds, Oakes and Hicks2005). In her landmark cross-sectional study in 1979, Sanders found that parents suffer more intense grief after the death of a child than other adults do after the death of either a spouse or a parent (Sanders, Reference Sanders1979; Middleton et al., Reference Middleton, Raphael and Burnett1998; Sirki et al., Reference Sirki, Saarinen-Pihkala and Hovi2000; Fletcher, Reference Fletcher2002; Seecharan et al., Reference Seecharan, Andresen and Norris2004). A similar, but longitudinal study, conducted by Middleton et al. (Reference Middleton, Raphael and Burnett1998), confirmed Sanders' findings at 1 month, 10 weeks, 7 months, and 13 months after the death of a child. Other studies suggest that parental grief is typically more intense for those going through it than is the loss of any other adult including friends, partners, and adult siblings (Sirki et al., Reference Sirki, Saarinen-Pihkala and Hovi2000; Fletcher, Reference Fletcher2002; Seecharan et al., Reference Seecharan, Andresen and Norris2004).
Several studies report that the intense emotions characterized as parental grief may last from 1 to 9 years after the child's death (McClowry et al., Reference McClowry, Davies and May1987; Theut et al., Reference Theut, Pedersen and Zaslow1989; Gilliss et al., Reference Gilliss, Moore and Martinson1997; Laakso & Paunonen-Ilmonen, Reference Laakso and Paunonen-Ilmonen2002), but many others suggest that feelings of sadness and loss remain throughout the bereaved parent's life, that is, “grief gets different, it doesn't get better” (Talbot, Reference Talbot1996; Woolley, Reference Woolley1997; Romesberg, Reference Romesberg2004, p. 163; Arnold et al., Reference Arnold, Gemma and Cushman2005). Talbot (Reference Talbot1996) describes the experience as “remaining in a perpetual state of bereavement” (p. 67) A study by Neidig and Dalgas-Pelish (Reference Neidig and Dalgas-Pelish1991) clearly illustrates this point. They found that parents who had lost a child 2–20 years prior to the interview had grief scores (The Texas Revised Inventory of Grief Part II: Present Feelings) similar to those of parents who has lost a child only 3 months to 2 years prior to entering the study.
Table 1 provides an illustration of this disagreement and a sample of the variety of research findings on the duration of grief in parents who have lost a child. Lang et al. (Reference Lang, Gottlieb and Amsel1996) suggest that the disagreement is likely due to the relatively small number of studies on parental bereavement and the fact that the majority of them are cross-sectional in nature, each focusing on a separate time frame in the grief experience.
Table 1. Disagreement among authors about duration of parental grief
Three primary causes for the unique intensity of parents' grief responses have been proposed in the literature, including disruption of the natural order, parental feelings of failure, and interruption of the family structure. Disruption of the natural order refers to the fact that in most developed nations, since early in the 20th century, the death of a child has become a relatively rare event (Davies, Reference Davies2004). Parents expect to die before their children, so when the “natural” process is reversed, society, including the parent's social support system (friends, relatives, religious organization, and even health care providers) is unprepared to respond appropriately to the grieving parent's needs (Fletcher, Reference Fletcher2002; Davies, Reference Davies2004). This lack of support leaves parents vulnerable to complicated grief responses and potentially to adverse health outcomes.
Second, the theory of parental feelings of failure refers to the parent's inability to prevent the child's death and to fulfill the role of protector. Though this may vary with cause of death (sudden or accidental, acute illness, or prolonged decline from a chronic condition), the resulting guilt may compound grief emotions and overwhelm the parent's coping mechanisms (Fletcher, Reference Fletcher2002; Hasui & Kitamura, Reference Hasui and Kitamura2004). Third, interruption of the family structure leaves the parent without the stability provided by their familiar and comfortable household, which again can leave them vulnerable to psychological stress and complicated mourning. Grieving parents must not only work to overcome intense sadness and loneliness after the loss of a child, but also must reorganize and restructure family roles and responsibilities to compensate for the absent member (Fletcher, Reference Fletcher2002; Lang et al., Reference Lang, Goulet and Amsel2004).
While theories are under study regarding the causes of complicated grief in parents, much disagreement exists in the literature about how that grief affects the physical health and well-being of the parents. Some researchers such as Birenbaum et al. (Reference Birenbaum, Stewart and Phillips1996) and Kvikstad and Vatten (Reference Kvikstad and Vatten1996) suggest that parents' physical health is not adversely affected by such a death, whereas other studies suggest that bereaved parents are at increased risk for a decline in physical health (Birenbaum et al., Reference Birenbaum, Stewart and Phillips1996; Kvikstad & Vatten, Reference Kvikstad and Vatten1996; Murphy et al., Reference Murphy, Lohan and Braun1999, Reference Murphy, Tapper and Johnson2003; Znoj & Keller, Reference Znoj and Keller2002; Lang et al., Reference Lang, Goulet and Amsel2004). Murphy et al. (Reference Murphy, Lohan and Braun1999) found that 81% of mothers and 85% of fathers reported their health as “excellent” 1 year after a child's death; however, she also notes that in her study, 70% of mothers and 53% of fathers report visiting a physician for care during the early stages of bereavement.
Other studies have presented findings suggesting that bereaved mothers may be more at risk for developing an illness, for having difficulty recovering from an injury or illness, and for dying of a disease than their peers who have not lost a child (Birenbaum et al., Reference Birenbaum, Stewart and Phillips1996; Murphy et al., Reference Murphy, Lohan and Braun1999, Reference Murphy, Tapper and Johnson2003; Znoj & Keller, Reference Znoj and Keller2002; Lang et al., Reference Lang, Goulet and Amsel2004), and that bereaved fathers may be more at risk for death from unnatural causes such as suicide, accident, or homicide than are their nonbereaved peers (Goodenough et al., Reference Goodenough, Drew and Higgins2004). Many of these findings, however, have also been criticized by those suggesting that the adverse health outcomes are typically psychosomatic in nature or related to risky behaviors (Birenbaum et al., Reference Birenbaum, Stewart and Phillips1996).
This article provides a review and analysis of the current conflicts in the literature and will address the associations between parental grief and health outcomes, including mortality risk and morbidity.
METHODS
Initially, a search was conducted using OVID-Medline (1996–November Week 2, 2007), CINAHL (1982–December Week 1, 2007), and PsycINFO (1967–December Week 3, 2007) using variations of the terms “parental grief and bereaved parents” combined with “health,” “illness,” “morbidity,” and “mortality.” All causes of death were included. Subsequently, the bibliographies of each source were reviewed for more potential sources. Thirty-four articles were initially identified in the search. However, after reviewing the titles and abstracts, only 17 articles were deemed appropriate for inclusion in the review. Table 2 contains a complete list of reviewed articles.
Table 2. Reviewed articles with measures and findings
In narrowing the field of articles, nine were excluded because the authors defined parental grief as grief of a child who had lost one or both parents. Seven were excluded because they focused on mortality as child death rather than parent's risk of death after losing a child. Finally, one article could not be located and so was excluded. No review articles were identified on the topic or included in this review.
RESULTS
Parental Grief and Mortality
Eight of the 17 articles reviewed discussed parental mortality after the death of a child. Four studies found an increased risk of death in bereaved parents both by suicide and from illnesses including cancer and myocardial infarction (MI; Levav et al., Reference Levav, Kohn, Iscovich, Abramson, Tsai and Vigdorovich2000; Li et al., Reference Li, Hansen and Mortensen2002a; Qin & Mortensen, Reference Qin and Mortensen2003; Davies, Reference Davies2006) and four studies found no significant difference in mortality risk between bereaved parents and the general population in either all-cause mortality or from specific illnesses including cancer and stroke (CVA; Levav et al., Reference Levav, Friedlander and Kark1988; Kvikstad & Vatten, Reference Kvikstad and Vatten1996; Li et al., Reference Li, Johansen and Olsen2003a, Reference Li, Johnsen and Olsen2003b).
All authors who investigated suicide and suicidal ideation agreed that bereaved parents are at higher risk for completed suicide than are members of the general population. Qin and Mortensen (Reference Qin and Mortensen2003) note that although being a parent is protective against suicide in the general population, the risk of suicide in parents increases significantly above that of the general population if the child dies, especially if the child is young (age 1–6 years) at the time of death (odds ratio [OR] 4.88, 95% confidence interval [CI] 3.23–7.39). Although more investigation is needed into the methods of suicide employed by this population, Davies (Reference Davies2006) suggests that the movement toward home-based palliative care has helped to increase parents' risk of death by suicide after the death by providing grieving parents with increased access to potentially lethal doses of opiods—those prescribed to control the pain of their dying children.
The disagreements, on the other hand, that exist in the literature regarding risk of mortality in bereaved parents primarily concern risk of death from cancer and cardiovascular events (stroke and MI). Levav et al. (Reference Levav, Kohn, Iscovich, Abramson, Tsai and Vigdorovich2000) found an increase in risk of death from cancer if the diagnosis was made prior to the death of their child, but no significant increase if the diagnosis was made after the child had already died. Kvikstad and Vatten (Reference Kvikstad and Vatten1996), however, found no increased risk of death from cancer in bereaved parents, but their study included only parents diagnosed with cancer after the loss of a child. Thus, their findings actually did not conflict with those of Levav et al. Further study is needed in this area to confirm or reject the hypothesis that parents diagnosed with cancer before losing a child have increased risk of death, whereas others do not.
The other area of disagreement among studies surrounds major cardiovascular events including myocardial infarction and CVA. Both articles reporting findings about cardiovascular-related mortality in bereaved parents came out of the same study in which Cox Proportional Hazards analyses were performed on data from the Denmark National Register. J. Li was the primary author on several papers resulting from this study. In the first article, Li et al. (Reference Li, Hansen and Mortensen2002a) present a study of the risk of fatal myocardial infarction in parents who lost a child. Similar rates of fatal MI between bereaved parents and matched controls were noted for the first 6 years of follow-up; however, from follow-up years 7–17, bereaved parents were significantly more likely to suffer a fatal MI than were controls (relative rate [RR] 1.58, 95%CI 1.08–2.30).
Unlike the findings for MI, Li et al.'s (2003b) findings regarding risk of fatal CVA in bereaved parents were nonsignificant. They found no increased risk of death from stroke (hemorrhagic or nonhemorrhagic) among bereaved parents versus their matched control group.
The use of the “hard” outcome, mortality, strengthened all of these studies because it is not often misdiagnosed, nor is it subject to the subjective interpretation found in self-report questionnaires. The use of the reliable national registry data by Li et al. also strengthened their analyses because the data set is both large and well managed by the government. The very limited number of studies in this field, however, provide very little support for the relationship between the psychosocial insult of a child's death and subsequent risk of death in the parents. More research is needed to provide a better understanding of the existence of a relationship and potential mechanisms that contribute to the linkage.
Parental Grief and Morbidity
Thirteen of the 17 articles reviewed discussed morbidity in bereaved parents after the death of a child. Nine studies found an increased risk of illness (including anxiety, depression and suicidal ideation, cancer, nonfatal myocardial infarction, multiple sclerosis, drug/alcohol abuse) and/or poor health self-ratings among bereaved parents (Vance et al., Reference Vance, Najman and Boyle1994, Reference Vance, Najman and Thearle1995; Murphy et al., Reference Murphy, Lohan and Braun1999, Reference Murphy, Tapper and Johnson2003; Levav et al., Reference Levav, Kohn, Iscovich, Abramson, Tsai and Vigdorovich2000; Li et al., Reference Li, Hansen and Mortensen2002a, Reference Li, Johansen and Hansen2002b, Reference Li, Johansen and Bronnum-Hansen2004a; Kreicbergs et al., Reference Kreicbergs, Valdimarsdottir and Onelov2004), and four found no significant excess risk for morbidity from specific diseases (cancer, nonfatal CVA, and exacerbation of inflammatory bowel disease) in parents who had lost a child versus those who had not (Birenbaum et al., Reference Birenbaum, Stewart and Phillips1996; Kvikstad & Vatten, Reference Kvikstad and Vatten1996; Li et al., Reference Li, Johnsen and Olsen2003b, Reference Li, Norgard and Precht2004b).
The findings of all authors examining adverse psychological outcomes including anxiety, depression, and suicidal ideation agree that bereaved parents are at higher risk for psychological disorders after the death of a child than are control parents of live children (Vance et al., Reference Vance, Najman and Thearle1995; Murphy et al., Reference Murphy, Tapper and Johnson2003; Qin & Mortensen, Reference Qin and Mortensen2003; Kreicbergs et al., Reference Kreicbergs, Valdimarsdottir and Onelov2004; Davies, Reference Davies2006). Measurement of symptoms of psychological disorders was based primarily upon questionnaires created by the researchers drawing on the Center for Epidemiologic Studies Depression (CESD) Scale, though Vance et al. (Reference Vance, Najman and Thearle1995) reported use of the Delusions-Symptoms-States Inventory of Anxiety and Depression (Bedford & Foulds, Reference Bedford and Foulds1977) and Murphy et al. (Reference Murphy, Tapper and Johnson2003) used the Brief Symptom Inventory (BSI; Johnson et al., Reference Johnson, Murphy and Dimond1996). Although these instruments have all been validated and it seems reasonable to assume that bereaved parents would be anxious and/or depressed, none of the studies mentions the use of a medical (psychiatrist/psychologist) exam or International Classification of Diseases (ICD) diagnosis code as “hard” evidence of the outcome of interest.
Four other morbidities were examined in the reviewed articles, but sufficient findings do not exist on any specific illnesses examined to provide a thorough exploration of the relationship between parental grief and illness. However, this review will provide a look at the literature that exists so far. Single studies reviewed the incidences of multiple sclerosis (MS) and exacerbation of inflammatory bowel disease (IBS). Although psychological stress is commonly thought to be related to or to facilitate exacerbations and worsening of both conditions (Li et al., Reference Li, Johansen and Bronnum-Hansen2004a, Reference Li, Norgard and Precht2004b), in the study of bereaved parents, Li et al. found a relationship between parental bereavement and MS incidence, but not bereavement and length of hospitalization for IBS.
Two studies were completed in which the researchers examined cardiovascular (CV) morbidity outcomes (nonfatal MI and nonfatal stroke) in bereaved parents. Like the CV mortality outcomes, both of the CV morbidity articles were based on the large population-based study completed in Denmark (Li et al., Reference Li, Hansen and Mortensen2002a, Reference Li, Johnsen and Olsen2003b). Findings from the study revealed that similar rates of nonfatal MI were noted in both bereaved parents and in matched controls for the first 6 years of follow-up. However, from years 7–17 of follow-up, bereaved parents were significantly more likely to suffer a nonfatal MI than were controls (RR 1.31, 95%CI 1.09–1.57). Li et al. (Reference Li, Hansen and Mortensen2002a) suggest that these findings lend support to the body of literature linking high levels of stress to adverse cardiac events.
In direct contrast to the stress literature and Li's MI findings, however, Li et al. (Reference Li, Johnsen and Olsen2003b) found no increased risk of either hemorrhagic or nonhemorrhagic nonfatal CVA in bereaved parents when compared to matched controls, even after 18 years of follow-up. So the link between death of a child and subsequent adverse cardiovascular outcomes in parents is not well established and requires further study.
Finally, the most clear example of disagreement in the literature regarding parental grief and morbidity health outcomes is in the area of cancer. Cancer is also the most thoroughly examined morbidity risk for bereaved parents, with three articles on the subject. Two studies (Levav et al., Reference Levav, Kohn, Iscovich, Abramson, Tsai and Vigdorovich2000; Li et al., Reference Li, Johansen and Hansen2002b) reported increased cancer risk in bereaved mothers, whereas one (Kvikstad & Vatten, Reference Kvikstad and Vatten1996) found no significant difference in relative risk between bereaved and nonbereaved mothers. It is interesting to note that though they reported opposing findings, both the study by Li et al. (Reference Li, Johansen and Hansen2002b) and the study by Kvikstad and Vatten (Reference Kvikstad and Vatten1996) were population-based studies in similar countries (Denmark and Norway, respectively) using national registries, matched controls, and Cox Proportional Hazard Analysis.
Overall, a greater number of studies (nine) reported findings that support the hypothesis that a relationship exists between parental grief and morbidity. However, it is clear that there are still large gaps in this area of research and further study is necessary.
DISCUSSION
This review of the literature on risk of morbidity and mortality in parents who have lost a child reveals that both commonalities and significant disagreements exist as part of the current state of the science. Most researchers appear to be working with the assumption that the death of a child acts as a psychological stressor to the parents, and it has been hypothesized that this stress may adversely affect health either directly through immunologic pathways or indirectly by encouraging or contributing to risky health behaviors such as poor diet and exercise habits (Baum & Grunberg, Reference Baum, Grunberg, Cohen, Kessler and Gorden1995; Cohen et al., Reference Cohen, Kessler, Gorden, Cohen, Kessler and Gorden1995; Li et al., Reference Li, Johnsen and Olsen2003b). From this review, it is clear that more consistent, well-designed research is needed to allow for more comparison among studies to occur and for the risks a child's death poses to parents' health to be fully identified and more thoroughly understood.
Three primary areas of concern with existing studies have surfaced during this review including lack of consistency in measurement for psychological (particularly grief) variables and “soft” self-report health outcomes, questionable methodologies in bereavement research in general, and the lack of a uniform definition of bereaved parents.
Measurement of Parental Grief and Outcomes
First, with regard to measurement, the use of grief instruments to measure parents' psychological distress after the loss of a child is of great concern. Lang et al. (Reference Lang, Gottlieb and Amsel1996) report that many of the grief inventory instruments commonly used in research studies were developed based upon women's grief experiences and may not accurately capture fathers' perspectives on grief . Additionally, many researchers examining psychological or self-report health outcomes report using scales created by the authors for the purpose of their own studies (Vance et al., Reference Vance, Najman and Boyle1994; Murphy et al., Reference Murphy, Lohan and Braun1999; Kreicbergs et al., Reference Kreicbergs, Valdimarsdottir and Onelov2004). Though some state that their new instruments are based on valid and reliable instruments such as the CESD and BSI, no formal psychometric testing of the newly created instruments was reported in the literature, and the reader is left wondering whether the new instrument appropriately captured the variable of interest in both parents. This concern is lent further support when the reader considers that though several of the studies reviewed here do report differences in self-report outcomes between the mothers and fathers, mothers' grief and distress scores are typically higher (Vance et al., Reference Vance, Najman and Boyle1994; Birenbaum et al., Reference Birenbaum, Stewart and Phillips1996; Kreicbergs et al., Reference Kreicbergs, Valdimarsdottir and Onelov2004). It is unclear whether these measurements truly reflect more distress and poorer health in mothers or whether the instruments failed to appropriately capture the fathers' experiences.
A similar weakness in this area was that several studies utilized “soft” health and morbidity outcomes that were based on self-report such as the Duke–UNC Health Profile (Birenbaum et al., Reference Birenbaum, Stewart and Phillips1996), the BSI (Murphy et al., Reference Murphy, Tapper and Johnson2003), and the Delusions-Symptoms-States profile (Vance et al., Reference Vance, Najman and Thearle1995). Although these instruments have all had sound psychometric analyses, the results would have been more substantive and more strongly evidence based if outcomes had included “hard” measures such as presence or absence of disease in addition to the self-report measures, or if soft findings such as anxiety and depression had been confirmed by physicians, medical record review, or analysis of health care utilization.
Methodological Challenges
The second area of concern identified in this review is methodological challenges in working with a population of bereaved parents. The most commonly noted methodological challenges in parental bereavement literature, including small sample size, retrospective design, and limited follow-up, are all present to some degree in the articles reviewed here. Many of the studies included in this review were population-based studies that utilized national registry data with hundreds of thousands of subjects, but others fell victim to the common methodological problem of small sample size. Because since the early 20th century the death of a child has become an uncommon occurrence in developed nations, it can often be difficult to recruit an adequate sample size in a single hospital setting. Sample sizes in this review ranged from as few as 80 subjects (Birenbaum et al., Reference Birenbaum, Stewart and Phillips1996) to 21,062 bereaved parents and greater than 300,000 controls in the population-based studies. Although the population-based studies are helpful and feasible in countries with socialized medicine and national registries such as in Denmark, Norway, and Israel, currently such research is difficult if not impossible with bereaved parents in the United States because national health data are available only for those with Medicare or Medicaid coverage, which includes primarily the elderly and those with low levels of family income. This suggests the need for more concerted effort to create large data sets that will help researchers to capture the bereavement experience of parents in the United States.
The second methodological concern is that prospective studies are typically preferable to retrospective ones whenever possible. The majority of studies on parental bereavement and health have been done retrospectively; however, because most have investigated the bereaved parents of children who died suddenly and unexpectedly, it was not possible to identify these parents before the bereavement event. A single study, however, by Birenbaum et al. (Reference Birenbaum, Stewart and Phillips1996) employed a prospective study design in which the researchers collected data on parent's health before the death of their child from cancer and at three points during the first year after the child's death. The design, however, did not seem to function as it was intended, as a baseline measure of health from which bereavement would theoretically contribute to changes. This is because the parents were recruited when the child was very near death, sometimes within a few hours. (In one case, the child died the same day the parents enrolled in the study.) These parents were expecting the death of the child and were likely already in a state of anticipatory grief, which may already have affected their health status. This consideration casts doubt on their findings that “parents' health was not adversely affected by a child's death from cancer” (p. 105).
The final methodological concern of interest is related to the length of follow-up for many studies. At the time children are born, most parents are in their 20s and 30s, a typically healthy age for most people. Therefore, if children die before they turn 18, overall, parents are still in a relatively healthy age group. So, it would be expected that a fairly long-term follow-up would be necessary to capture any potential increased risk of disease such as cancer, stroke, adverse cardiovascular events, or psychological disorders and risky behaviors in this population. However, other researchers have also noted that length of time since the child's death could play a role in risk of morbidity and mortality in bereaved parents (Murphy et al., Reference Murphy, Tapper and Johnson2003), so the optimum follow-up time remains unclear.
In the studies examined in this review, the authors used a wide variety of follow-up times. Whereas some researchers who used national registry data boasted follow-up times of 18–24 years (Levav et al., Reference Levav, Kohn, Iscovich, Abramson, Tsai and Vigdorovich2000; Li et al., Reference Li, Johansen and Hansen2002b), others used cross-sectional designs that included data collection at a single point after the child's death and no follow-up (Vance et al., Reference Vance, Najman and Boyle1994; Kreicbergs et al., Reference Kreicbergs, Valdimarsdottir and Onelov2004) or utilized follow-up times as short as 8 months (Vance et al., Reference Vance, Najman and Thearle1995), 1 year (Birenbaum et al., Reference Birenbaum, Stewart and Phillips1996), 2 years (Murphy et al., Reference Murphy, Lohan and Braun1999), or 60 months (Murphy et al., Reference Murphy, Tapper and Johnson2003), but that did not appear adequate to capture morbidity and mortality outcomes. A complete list of follow-up times is available in Table 2.
Definition of Child
The third and final concern identified in this review is the lack of a consistent definition of child among studies. It is unclear whether child should refer only to those individuals less than 18 years old or to any person with a living parent. To make this review as thorough as possible, all articles on bereaved parents were selected—including two articles that examined bereaved parents of adult children who had died either in an accident or in a war (Levav et al., Reference Levav, Friedlander and Kark1988, Reference Levav, Kohn, Iscovich, Abramson, Tsai and Vigdorovich2000). As was stated earlier, many studies have suggested that once a parent loses a child, he (or she) remains a “bereaved parent” for the rest of his (or her) life (Rubin, Reference Rubin1999; Levav et al., Reference Levav, Kohn, Iscovich, Abramson, Tsai and Vigdorovich2000; Hasui & Kitamura, Reference Hasui and Kitamura2004); however, several researchers have raised the issue that the age of the child at the time of his or her death might potentially affect the parents' grief process and risk for adverse health outcomes (Murphy et al., Reference Murphy, Tapper and Johnson2003).
One study found that the highest risk of morbidity in the form of poor mental health and suicide was in parents of children who died at very young ages (Qin & Mortensen, Reference Qin and Mortensen2003), but the study did not include bereaved parents of adult children. Similarly, Levav et al. (2000) found an increased risk of death from cancer in bereaved parents of adult children, but this study did not include bereaved parents of children under the age of 18 years and it disagrees with the findings of Li et al. (Reference Li, Johansen and Olsen2003a), who examined bereaved parents of young children and found no difference in cancer survival in that group compared to matched controls. Clearly, more research is needed in this area to determine the most appropriate inclusion criteria for studies of bereaved parents, which will allow for optimum comparison among studies.
Limitations
This review is limited by the very small amount of literature on health outcomes in bereaved parents. The strongest of the existing literature focuses on predominately white, wealthy, European countries with socialized medical care and parents of children who died unexpectedly or tragically. Although this review depicts the overall health risks for parents who lose a child, it remains unclear to what extent various aspects of deceased children and their families such as culture (Costa et al., Reference Costa, Hall and Stewart2007), socioeconomic status (Koop & Strang, Reference Koop and Strang1997), mode of reimbursement (public or private insurance), cause of death (Zisook et al., Reference Zisook, Chentsova-Dutton and Shuchter1998), enrollment in hospice (Christakis & Iwashyna, Reference Christakis and Iwashyna2003), and involvement in care (McCorkle et al., Reference McCorkle, Robinson and Nuamah1998) play a role the subsequent health of parents.
Conclusions
The literature provides very mixed results on the relationship between parental grief and subsequent morbidity and mortality risks to the parents. This review and analysis revealed that, in general, many of the current studies have several problems, which may make their results questionable and/or unreliable. These include biased grief measurement tools, methodological concerns such as small sample size, retrospective design, and inadequate follow-up periods, and lastly, indecision about the comparing of literature concerning bereaved parents of adult children with that concerning bereaved parents of young children. Additionally, this review revealed a gap in the parental bereavement literature. Characteristics of the deceased child and family and circumstances surrounding the child's death have yet to be examined in relation to parental grief, and it remains unclear whether these variables play a role in the subsequent health of parents.
Based on these findings, it is clear that more methodologically sound research is necessary to clarify the relationship between parental grief after the death of a child and the parents' subsequent morbidity and mortality risks.
ACKNOWLEDGMENT
The author gratefully acknowledges the contributions of Dr. Ruth McCorkle to the ideas presented in this article.
