Consequences of Later-Life Divorce and Widowhood for Adult Well-Being: A Call for the Convalescence Model

Abstract

Later-life marital dissolution increasingly occurs through divorce, not widowhood, and this is reflected in the rising rate of divorce among older adults and gains in life expectancy. Yet the consequences of later-life divorce for individual well-being are poorly understood. Prior work on later-life marital dissolution focuses on bereavement, positing either a short-term, acute grief period or chronic, unrelenting grief. The divorce literature offers an analogous bifurcated explanation in the stress-adjustment perspective on divorce. Adjustment either occurs during a brief crisis period or is elusive as divorce creates chronic strain. After describing these entrenched conceptual models of postdissolution adjustment, we introduce a third trajectory that is emergent in recent, longitudinal studies: the convalescence model. Drawing on Pearlin’s stress process perspective, we theorize why spousal loss and divorce alike may result in a protracted period of recovery. We conclude with directions for future research to encourage conceptual and methodological advancements in the field.

Studies of marital dissolution in later life (age 50 years and older) have focused on widowhood, largely overlooking older adult divorce. This singular focus was appropriate when divorce in later life was uncommon. In 1990, for instance, fewer than 10% of all divorces occurred among persons aged 50 or older (Brown & Lin, 2012). Nowadays, the exclusion of divorce as a dissolution pathway effectively mischaracterizes older adult marital dissolution, given that one-third of dissolutions among older adults occur through divorce rather than spousal death (Brown, Lin, Hammersmith, & Wright, 2018). The divorce rate among adults aged 50 and older has doubled since 1990. This doubling of the “gray divorce rate,” as it is termed in the literature, coupled with the aging of the population means that more than one-quarter of today’s divorces involve an older adult (Brown & Lin, 2012).

Despite the recent acceleration in older adult divorce, theoretical developments have not kept pace. In particular, the consequences of divorce for older adults remain poorly understood. The historical emphasis on widowhood in the gerontological literature means that the available theoretical frameworks for understanding later-life marital dissolution center on bereavement (Stroebe, Hansson, Schut, & Stroebe, 2008). Although divorce, like spousal death, is a stressful life event, it is not readily apparent that bereavement would align with the divorce experience. Nor would we necessarily expect that the short- and long-term effects of spousal death should map onto the corresponding effects of gray divorce for individual health and well-being. At the same time, theories from the vast literature on divorce, such as the stress-adjustment perspective (Amato, 2000), are unlikely to mesh seamlessly with older adults’ experience of divorce, as older adults are at a stage of the life span marked by particular strengths and vulnerabilities (Charles, 2010).

Our goal is to evaluate the existing theoretical frameworks guiding research on the outcomes associated with spousal death and divorce during the second half of life. We consider the concept of acute grief that is at the center of work on the aftermath of spousal death as well as the alternative explanation that spousal death results in chronic grief. In addition, we examine the stress-adjustment perspective that dominates the divorce literature, including the competing models on whether divorce operates as a short-term temporary crisis period or a long-term persistent chronic strain. Ultimately, we assert that each of these models has limited utility for describing the consequences of marital dissolution during the second half of life. Instead, we develop a novel theoretical framework for a construct first introduced by Lin, Brown, Wright, and Hammersmith (2019) as a post hoc explanation for the pattern of adjustment after marital dissolution that they observed among older adults: the convalescence model. Unlike the acute grief and crisis models, the convalescence model posits that marital dissolution has long-term negative effects on well-being. Yet contrary to the chronic grief and chronic strain models, which stipulate that the negative effects of dissolution persist indefinitely, the convalescence model characterizes the effects as protracted but eventually receding. The convalescence model describes a long, slow recovery, which is a trajectory that until now has been absent from the marital dissolution literature.

Background

The U.S. widowhood rate among adults aged 50 and older (“older adults” hereafter) has declined over the past 40 years. In 1980, 27 people became widowed for every 1,000 married individuals older than age 50 (Sweet & Bumpass, 1987), whereas that number was 19 in 2018 (authors’ calculation using 1-year estimates from the American Community Survey). During the same period, the overall U.S. divorce rate, which peaked in 1980, leveled off and since has declined slightly. In 1980, 20 people divorced for every 1,000 married individuals (Clarke, 1995), whereas that number was 15 in 2018 (authors’ calculation using 1-year estimates from the American Community Survey). This overall trend masks considerable variation by age group, however. Since 1990, the divorce rate for adults younger than age 35 has decreased, whereas for those aged 35 and older it has increased (Kennedy & Ruggles, 2014; Wu, 2017). Indeed, the largest rise occurred among older adults. The divorce rate among adults aged 50 and older doubled from five divorcing persons per 1,000 married persons in 1990 (Brown & Lin, 2012) to 10 divorcing persons in 2018 (authors’ calculation using 1-year data from the American Community Survey). In 2010, about 600,000 people experienced a gray divorce, and this figure is expected to rise to more than 800,000 by 2030 (if the gray divorce rate from 2010 holds) simply because of the aging of the population (Brown & Lin, 2012).

Compared to those of prior generations, the marriages of today’s older couples are increasingly likely to end through divorce rather than spousal death (Brown et al., 2018). Yet the bereavement and divorce literatures remain distinct, with the latter focused primarily on younger adults, whose outcomes may not mirror those of their older counterparts who experience gray divorce. The demographic shifts in later-life marital dissolution signal the importance of developing broader theories that encompass various dissolution pathways and their implications for older adult well-being. Gerontologists in particular must widen their focus to include marital dissolution that occurs not just through spousal death but also through gray divorce.

Bereavement: Acute Versus Chronic Grief

Spousal death is considered the most stressful life event during adulthood (Miller & Rahe, 1997). Not surprisingly, a large body of research addresses adjustment to spousal loss. In general, bereavement scholars have conceptualized adjustment to spousal death as either acute grief that involves an initial decline in well-being, followed by recovery in a few weeks to up to two years (Wortman & Silver, 2001), or chronic grief, from which the bereaved do not return to preloss levels of well-being (Stroebe, Hansson, Stroebe, & Schut, 2001). Despites the recent advances in research that include unpacking heterogeneity in postloss adjustment among the bereaved (Carr & Springer, 2010), much of the work on bereavement still reflects this bifurcated approach of conceptualizing grief as either acute or chronic (Stroebe et al., 2008).

The preponderance of early bereavement studies has been limited in approach, relying either on cross-sectional data that lacked preloss well-being measures (e.g., Carnelley, Wortman, Bolger, & Burke, 2006; Wilcox et al., 2003) or on two waves of data consisting of just one point of data before and another shortly after spousal death (e.g., Lee & DeMaris, 2007). By design, these studies cannot ascertain whether bereaved persons eventually return to preloss levels of well-being and how long it takes individuals to recover. The recent availability of longitudinal data to follow individuals before, during, and long after marital dissolution makes it possible to assess whether individuals recover from initial declines in health and well-being after marital dissolution and how long it takes them to return to predissolution levels.

Several such studies using longitudinal data have lent support to the acute grief model. Using 9 years of data from the annual British Household Panel Study (BHPS), conducted in the 1990s, Wade and Pevalin (2004) found that widow(er)s experienced a greater risk for mental health problems beginning in the year immediately preceding spousal death and returned to preloss levels the year following the death. Attempting to unpack heterogeneity among the bereaved, Maccallum, Galatzer-Levy, and Bonanno (2015) showed that a majority of respondents (71%) in the Health and Retirement Study (HRS, 1994–2010) were resilient because they experienced a modest elevation in depressive symptoms after spousal death, which quickly dissipated and reverted to preloss levels, resembling acute grief. Fewer studies have examined the consequences of spousal death for physical health (Zhang, Liu, & Yu, 2016), but Brockmann and Klein (2004), using the German Socio-Economic Panel Study (GSOEP, 1984–1988), revealed that mortality risks rose precipitously during the first 2 years after spousal death and then attenuated to preloss levels thereafter, in support of the idea that spousal death is an acute stressor.

The chronic grief model also has received empirical support. For instance, Maccallum et al. (2015) identified a nontrivial share of respondents (11%) in the HRS who experienced a significant elevation in depressive symptoms after spousal death. Although the heightened depressive symptoms gradually subsided, they did not return to preloss levels at the end of the 6-year postloss study period. Another study using monthly data (1993–2002) from the Centers for Medicare and Medicaid Service and the Vital Status file (Elwert & Christakis, 2006) observed that the transition to widowhood increased mortality risk during the first few months of bereavement, followed by a monotonic decline, although the risk remained elevated from the second year onward. Additionally, Dupre and Meadows (2007), using the HRS data, noted that the longer the time spent in widowerhood, the higher was men’s risk of experiencing diabetes, cancer, heart attack, or stroke onset, which suggests that the effect of spousal death on physical health is long term, at least for men.

A few studies have emerged that signal that adjustment to spousal loss does not neatly align with either the acute grief model (recovery within two years) or the chronic grief model (indefinitely elevated poor health). Lucas, Clark, Georgellis, and Diener (2003), using 15 waves of the GSOEP data, showed that life satisfaction dropped significantly during the first year of widowhood and then gradually increased. It took eight years, on average, for widow(er)s to return to pre-widowhood levels of life satisfaction. Granted, this study was based on a general adult population and thus may not accurately reflect bereavement experiences in later life. Using a nationally representative sample of individuals aged 51–61 observed for 14 years (1994–2008), Sasson and Umberson (2014) found that the transition to widowhood elevated depressive symptoms, but depressive symptoms reverted to pre-widowhood levels in about eight years. This conclusion was corroborated by Lin et al. (2019) using the same data source (HRS) with individuals aged 50 and older over a slightly longer period (1998–2014): They termed this pattern the convalescence model.

The emergence of this pattern of adjustment to loss—a long-term decline in well-being from which individuals ultimately recover—does not conform to the arcs espoused by existing models, instead mirroring a third pathway that combines elements of both the acute grief model and the chronic grief model. This novel insight was possible only with the recent availability of data that include not only preloss well-being measures but also postloss measures at multiple time points spanning more than a decade, making the protracted recuperation process visible. Furthermore, the recent advances in statistical modeling strategies that track within-person change over time and permit person-specific discontinuities in the trajectory before and after the event (Singer & Willett, 2003), as used in all three studies (Lin et al., 2019; Lucas et al., 2003; Sasson & Umberson, 2014), help researchers to chart the course of adjustment more precisely.

In sum, with the availability of longitudinal data and the advances in statistical modeling strategies, the field has made great strides in better understanding adjustment to bereavement. Both the acute and the chronic grief models have been supported in the literature. Nevertheless, the prevailing dichotomous approach to conceptualizing the consequences of spousal death either as acute grief that operates for roughly two years or less or as chronic grief that seemingly persists indefinitely ignores a third pathway, namely a protracted period of decreased well-being that ultimately culminates in recovery after several years: the convalescence model.

The Divorce-Stress-Adjustment Perspective: Crisis Versus Chronic Strain

As is spousal death, divorce is a stressful life event (Miller & Rahe, 1997), often leading to poor health and lower well-being. The divorce-stress-adjustment perspective, first articulated by Paul Amato in his 2000 Journal of Marriage and Family decade-in-review article, is a theoretical framework commonly adopted by scholars to study the consequences of divorce (e.g., Booth & Amato, 1991; Johnson & Wu, 2002; Kalmijn, 2017). This perspective views marital dissolution as a process, beginning while the marriage is still intact and continuing after the divorce is finalized. The perspective encompasses two distinct processes that offer competing views on how the divorce experience and subsequent adjustment unfold as either a temporary crisis period or an enduring chronic strain.

For some individuals, the disruptive effects of divorce are short-lived, confined to a crisis period that lasts for just one or two years after divorce. Divorce operates as an acute stressor that elevates psychological stress around the time of divorce, but the stress subsides precipitously and returns to predivorce levels about two years after divorce (Booth & Amato, 1991), which is akin to a temporary crisis. According to this model, it takes individuals a relatively short amount of time to adjust to singlehood and regain their footing following divorce.

For others, divorce operates as a chronic strain that is marked not only by elevated stress levels around the time of divorce but also for years thereafter. Individual well-being generally does not revert to predivorce levels, reflecting the enduring hardships linked to divorce, such as the decline in economic resources, loss of emotional support, and ongoing challenges involved in solo parenting and coparenting with an ex-spouse (Amato, 2000).

Studies that followed individuals before, during, and after divorce have lent support to the crisis model. For instance, two British studies (Blekesaune, 2008; Wade & Pevalin, 2004), using the BHPS data with different study period lengths (15 years and 9 years, respectively), revealed that people who were divorced experienced a greater risk for poor mental health leading up to divorce, and then their mental health typically returned to preloss levels the year following the divorce. Likewise, Brockmann and Klein (2004), using German data (GSOEP, 1984–1988), observed that mortality risks escalated during the first two years after divorce and then attenuated to predivorce levels thereafter, resembling a crisis period.

The chronic strain model also has received empirical support. Johnson and Wu (2002) used the Marital Instability over the Life Course (1980–1992) data and showed that psychological distress increased leading up to divorce but did not decline following marital disruption. Similarly, Lucas (2005), using German data (GSOEP), noted that life satisfaction declined after divorce and did not return to predivorce levels by the end of the study period (6 years). We are not aware of a study that supports the chronic strain model for physical health. Rather, two longitudinal studies (Kalmijn, 2017; Lorenz, Wickrama, Conger, & Elder, 2006) found that physical health did not become worse immediately after divorce. It is plausible that chronic diseases develop more slowly than psychological distress and thus do not manifest until many years after divorce (Hughes & Waite, 2009).

Similar to the bereavement literature, some empirical evidence about adjustment to divorce is at odds with both the chronic strain and the crisis model. This evidence is not in line with the chronic strain model because the divorced person eventually recovered. Nor is the pattern consistent with the crisis model, because recovery took twice as long as expected according to the crisis model. Kalmijn (2017), using annual data from the Swiss Household Panel (1999–2014), revealed that individuals initially became more depressed and less satisfied with life after divorce but then adjusted to the transition and returned to predivorce levels in five to seven years. Combining retrospective and prospective data from the HRS, Dupre, Beck, and Meadows (2009) showed that mortality risks increased during the first one to four years after a divorce and then attenuated to predivorce levels thereafter.

All the studies reviewed above are based on divorces that primarily occurred earlier in the life course to people younger than age 50. Unlike divorce at younger ages, gray divorce often does not involve issues related to single parenthood and coparenting, and thus it may be less prone to parenting stress and continued conflict with the former spouse. Nonetheless, older adults may be affected more severely than their younger counterparts by the loss of resources accumulated through marriages because those who have experienced gray divorce have fewer years left to recoup their financial losses and rebuild their social networks. Hence, whether the consequences of gray divorce are more aligned with the crisis or the chronic strain model—or perhaps another model altogether—is not entirely clear.

As far as we know, only one longitudinal study (Lin et al., 2019) has examined the consequences of gray divorce for individual well-being. Using a nationally representative sample of older adults from the HRS (1998–2014), Lin et al. (2019) found that older adults who got divorced exhibited higher levels of depressive symptoms than their married counterparts even before divorce occurred. Depressive symptoms rose at the transition to divorce, but they gradually subsided and returned to predivorce levels in about four years, suggesting that people ultimately recover from gray divorce, but that the recovery is protracted. This trajectory spurred the authors to propose a third pathway, the convalescence model. Although they briefly introduced this concept in the concluding section of their article, it was a post hoc explanation for their finding. As we argue in the remainder of this article, this third trajectory merits inclusion alongside the established models in the literature.

The Convalescence Model

Despite their use of different labels, the existing models to explain the effects of spousal death are analogous to models that describe the consequences of divorce. As Figure 1 depicts, both the acute grief model and the crisis model posit an immediate drop in well-being after marital dissolution, followed by a relatively quick recovery within two years (indicated by the dotted line). Likewise, the chronic grief and chronic strain models signal a persistent negative effect of widowhood and divorce, respectively, as well-being does not revert to predissolution levels (indicated by the dashed line). Yet none of these models describes the postdissolution trajectory of well-being that has been uncovered in recent longitudinal studies: an initial drop in well-being following dissolution from which individuals eventually make a slow, protracted recovery (indicated by the solid line). We explicate a theoretical foundation for this model, drawing on Pearlin’s stress process perspective (Pearlin, Menaghan, Lieberman, & Mullan, 1981) to describe why the convalescence model is an apt descriptor of the experience of marital dissolution in later life.

Figure 1.

Models Theorizing Adjustment to Later-Life Marital Dissolution.

Theorizing the Convalescence Model

A range of theoretical frameworks guide studies on bereavement (Archer, 2008) and divorce (Zhang et al., 2016). In particular, the stress process perspective, introduced by Leonard I. Pearlin nearly 40 years ago (Pearlin et al., 1981), has played a significant role in providing the foundation for understanding how stressful life events, such as marital dissolution, are associated with health, as well as how these events shape well-being through mechanisms that include personal and social resources (or a lack thereof) that mitigate or exacerbate the health impacts of stress (Aneshensel & Avison, 2015). The mechanisms outlined in the stress process perspective are well suited to explain how and why the convalescence model characterizes the adjustment process following marital dissolution in later life.

The stress process perspective consists of three core components: stressors that provoke stress, mediators or moderators, and health outcomes (Pearlin & Bierman, 2013). The sources of stress include life events (i.e., primary stressors) and their associated strains (i.e., secondary stressors). Adverse life events like spousal death and divorce often incite stress, and strains accompanied by the events, such as economic hardship, social isolation, and single parenthood, further aggravate the stress (Pearlin & Johnson, 1977). Mediators and moderators are mechanisms that encompass social or personal resources and coping. With support from family or friends and effective coping, the negative effects of the life events and the strains on stress can be quickly mitigated, whereas the lack of social support or the use of ineffective coping strategies can lead to persistent stress. Finally, stress affects people’s lives through its manifestation in poor health.

Scholarly discussions of the risk and protective factors that explain individual differences in adjustment to bereavement have identified mechanisms connecting spousal death and health outcomes that are similar to those articulated by Pearlin et al. (1981). For instance, Stroebe, Folkman, Hansson, and Schut (2006) conceptualized two major sources of risk or protective factors: inter- or intrapersonal resources and coping. Interpersonal resources involve factors external to the bereaved person, such as social support, intervention programs, and family dynamics. Intrapersonal resources are characteristics intrinsic to the bereaved person, including attachment styles, personalities, and age-related frailty. Coping encompasses cognitive and behavioral adaptation and emotion regulation. Together these factors determine how well bereaved individuals adjust to spousal loss.

Likewise, the divorce-stress-adjustment perspective (Amato, 2000) echoes Pearlin’s stress process perspective (1981), delineating both risk and protective factors that explain variation in individual adjustment to divorce. Diminished interpersonal resources such as conflict between ex-spouses and smaller networks of supportive kin and friends following divorce are linked to poorer postdivorce adjustment. In contrast, intrapersonal resources such as higher socioeconomic status and better health, as well as constructive coping skills, buffer the negative ramifications of divorce for individual well-being.

Pearlin’s stress process perspective that emphasizes how social and personal resources and coping mediate or moderate the course of adjustment to life stress provides a foundation for understanding why a slow, protracted recovery would best depict the experience of later-life marital dissolution. Older adults differ from younger adults in their repertoires of interpersonal and intrapersonal resources to combat stressful experiences. Old age is often characterized as a time of loss through deteriorating health, losing friends or family members, and declining cognitive functioning, all of which diminish the capacity of older adults to alleviate stress using effective coping strategies. Thus, although adverse life events reduce one’s sense of control in life at any age, the erosive effect is particularly pronounced among older adults (Cairney & Krause, 2008). Loss of personal and social resources makes older adults more vulnerable to adverse life events.

At the same time, older adults are resilient, as their adroit emotion regulation helps them spring back from and adapt to changed life circumstances. Charles (2010) posited that older adults are better than younger adults in emotion regulation because they realize that the time they have left to live is growing shorter. According to socioemotional selectivity theory (Carstensen, 1992), two main goals motivate human interaction. One centers on information and knowledge acquisition for the future, and the other seeks emotional states and emotionally meaningful activities of the present. When the time left to live grows shorter, as during old age, emotional goals of interaction would be perceived as more important than information goals. Thus, older adults are more likely than younger adults to direct their efforts toward maintaining emotional well-being and so they would engage in successful emotion regulation strategies, for example, by curtailing less desired interaction.

Older adults are also better than younger adults at emotion regulation because they accrue experience and knowledge from years lived (Charles, 2010). A study conducted by Mirowsky and Ross (2001) showed that both new and persistent economic hardship was associated with lower levels of depressive symptoms for older adults relative to younger adults, in support of the age-as-maturity hypothesis (Mirowsky & Ross, 1992). People become more mature by navigating the problems of daily life. Older adults who have survived negative life events in the past learn how to cope and feel confident about their ability to overcome similar or new problems in the present.

Hence, older adults are more vulnerable than younger adults, yet an awareness of the time they have left to live and their experiences and knowledge accrued over the years give older adults better emotion regulation in the face of adversity. It may take older adults more time than younger adults to recoup resources lost because of marital dissolution, but they eventually bounce back, returning to predissolution levels of well-being. The convalescence model as illustrated by Lin et al. (2019) and Sasson and Umberson (2014) reflects this protracted recuperation process.

In the stress process perspective, Pearlin (1989) also noted that uncontrollable changes tend to be perceived as more stressful. In the event of marital dissolution in later life, spousal death may be perceived as more stressful than divorce, because spousal death is involuntary, whereas divorce often involves mutual consent. This is supported by Lin et al.’s (2019) study, which showed that after marital dissolution, widowed persons faced a more extended period to recovery than did individuals who got divorced. The recovery time following widowhood was double that for gray divorce (8 vs. 4 years, respectively) (Lin et al., 2019).

Future Research

Despite mounting empirical evidence of distinctive postdissolution trajectories of well-being among older adults, scholars have been slow to challenge the dominant dichotomous paradigm that pits short-term reversible effects against enduring persistent effects. Lin et al. (2019) recently called attention to a third trajectory of postdissolution well-being, the convalescence model, to describe depressive symptoms following either spousal death or divorce in later life that are marked by a long, slow recovery that is reminiscent of a pattern of convalescence. The entrenchment in the literature of the bifurcated models may be partially due to inertia (see Roberto, Blieszner, & Allen, 2006), but it also results from the methodological challenges that arise when studying the consequences of later-life marital dissolution. Given the recent availability of longitudinal data that follow older adults over an extended period of time and advances in statistical analyses that chart within-person change over time, scholars have the tools to evaluate the application of the convalescence model for a range of well-being indicators and to refine the model as needed.

Methodological Challenges

Researchers studying the consequences of later-life marital dissolution face at least three methodological challenges: (a) selection versus causation, (b) status versus transition, and (c) short versus long follow-up periods after dissolution. These challenges may obscure the patterns of decline and eventual recovery that constitute the lived experiences of older adults after spousal loss or divorce.

First, although marital dissolution often diminishes health and well-being, one cannot rule out social selection as an explanation (Amato, 2010). Widowhood and divorce tend to be selective of individuals who have poor health and lower well-being. Indeed, several studies have shown that individuals who become widowed or divorced already have experienced poorer mental health before dissolution than their married counterparts (Blekesaune, 2008; Johnson & Wu, 2002; Lee & DeMaris, 2007; Wade & Pevalin, 2004), likely because of caregiving stress (for widowhood) or marital strain (for divorce). Without taking into account predissolution well-being, researchers are prone to overestimate the negative effects of marital dissolution. A prospective approach that begins several years before marital dissolution is critical for illuminating whether individuals return to predissolution levels of well-being and, if so, how long it typically takes them to recover.

The second issue that arises when studying the consequences of marital dissolution is the potential to confuse marital status (between-person differences) and marital transition (within-person change) effects. Marital status indicates one’s current state of being married, separated, divorced, widowed, or never married, whereas marital transition captures the progression from one marital status to the next (Zhang et al., 2016). Prior research has consistently shown that those who are widowed or divorced have poorer mental health, more chronic conditions and mobility limitations, and shorter life expectancies than their married counterparts during later life (Hughes & Waite, 2009; Manzoli, Villari, Pirone, & Boccia, 2007; Zhang & Hayward, 2006). These studies are based on between-person comparisons by contrasting those who are widowed or divorced with those who are married. Additionally, studies on marital status often do not distinguish those who experienced marital dissolution earlier in life from those who became widowed or divorced in later life. The former have had years to recover from the stress brought about from marital dissolution, yet the latter are still adjusting to the relatively recent transition (Wilcox et al., 2003). Further, those who are well-adjusted in the aftermath of marital dissolution (i.e., those with better health and more economic resources) are more likely to repartner than those who are not (Brown, Lin, Hammersmith, & Wright, 2019). Therefore, to better understand the trajectories of adjustment to spousal death and divorce, researchers need to rely on methods that gauge the transition from one marital status to the next and to track within-person change over time.

The third issue related to studying whether the consequences of marital dissolution are more aligned with the crisis, chronic, or convalescence model is that it requires a study period that extends for more than a few years after marital dissolution. In addition, the study period would need to include indicators of health and well-being that are measured at multiple time points, preferably with equal intervals. For instance, many studies using the Changing Lives of Older Couples study follow older adults to 6 or 18 months after spousal death (e.g., Bonanno et al., 2002; Carr, House, Wortman, Nesse, & Kessler, 2001). That duration of 6 to 18 months may be too short to observe widow(er)s who eventually complete a long, slow recovery, resulting in erroneously characterizing the trajectory as chronic grief rather than convalescence. In addition, although the Marital Instability Over the Life Course study spans more than a decade, the interval between two adjacent follow-up interviews is four to five years apart (Johnson & Wu, 2002), which is arguably too long to capture short-term effects that result in well-being changes between interviews.

In summary, to better understand adjustment to later-life marital dissolution, researchers need longitudinal data sets that track individuals before, during, and after marital dissolution to take into account selection, to capture transitions (vs. statuses), and to test whether adjustment trajectories are more consistent with the crisis, chronic, or convalescence models. These methodological hurdles must be surmounted to accurately characterize the responses of older adults to marital dissolution and thus should serve as guideposts for future research to ensure appropriate methodological rigor.

New Directions

The convalescence model offers a new perspective for understanding how adults in the second half of life respond following marital dissolution. The availability of several data sources (e.g., Health and Retirement Study; Panel Study of Income Dynamics; National Social Life, Health, and Aging Project; National Health and Aging Trends Study) that prospectively follow nationally representative samples of older adults coupled with advances in a variety of growth-curve-modeling strategies enable researchers to rigorously investigate the applicability of the convalescence model to postdissolution adjustment and to develop the model conceptually. We offer several avenues for future research.

First, although Pearlin’s stress process perspective is a sound framework for explicating the theoretical underpinnings of the convalescence model, future research may benefit from integrating additional theoretical perspectives to expand the utility of the convalescence model. For instance, it may be fruitful to apply cumulative disadvantage theory (Dannefer, 2003) or the life course perspective (Elder, 1994) to an examination of the intersection of age, gender, and race-ethnicity for well-being after spousal death or divorce.

Second, older adults are heterogeneous in terms of their individual strengths and vulnerabilities, and thus we do not expect the convalescence model to characterize the postdissolution adjustment process for all older adults. For older adults who have a lot of strengths but few vulnerabilities, their experience may align more with the crisis models, whereas for older adults who have many vulnerabilities but meager strengths, their experience may be more congruent with the chronic models. Recent research (Lin et al., 2019; Sasson & Umberson, 2014) using the HRS that supports the convalescence model has relied on growth-curve modeling to predict overall population averages, which could obscure possible heterogeneity. Maccallum et al. (2015) used latent growth mixture analyses to model heterogeneity in postdissolution adjustment following widowhood, but their approach drew on a limited subsample of the HRS and did not examine marital dissolutions that occurred through divorce. Future research should unpack older adult population heterogeneity in both gray divorce and widowhood. Moreover, prior studies have identified factors that may account for variation in adjustment to later-life marital dissolution such as gender, marriage order, and marital duration, but these factors do not appear to predict various adjustment trajectories, at least in a limited number of studies (e.g., Bowen & Jensen, 2017; Lin et al., 2019; Sasson and Umberson, 2014). It is important to further explore these differentials using different data and to examine additional risk and protective factors.

Third, the two studies supporting the convalescence model for adults who experience later-life marital dissolution (Lin et al., 2019; Sasson & Umberson, 2014) are based on the same data source (i.e., HRS). Even though the HRS is advantageous because it contains a large sample of adults that is representative of diverse racial-ethnic groups and the full range of education and income distributions in the older population, the convalescence model awaits confirmation using other data sources. Also, both existing studies focus on depressive symptoms and thus the extent to which the convalescence model applies to other domains of health and well-being is unknown. A handful of longitudinal studies have examined how spousal death is related to mortality and illness onset (Brockmann & Klein, 2004; Elwert & Christakis, 2006), but because death and chronic diseases are absorbing states from which people cannot escape, these studies do not reveal whether the convalescence model is applicable to physical health for the bereaved. Two other studies relying on the general adult population (Kalmijn, 2017; Lorenz et al., 2006) suggested that physical health problems seem to be less affected by divorce, at least in the short term. Whether a similar pattern is also observed for gray divorce is unclear. Because chronic diseases develop more slowly than psychological distress (Hughes & Waite, 2009), we suspect that the convalescence model would be more applicable to mental health than physical health, but this awaits confirmation in further studies. In addition to addressing whether the convalescence model is evident for adjustment trajectories using other domains of well-being, it would be worthwhile for scholars to explore whether the model is particularly suited to older adults or is more broadly applicable to adults of all ages by examining whether convalescence also characterizes the adjustment experiences of younger adults.

Fourth, another pathway to achieve resilience among older adults is following a turning point in life (Bennett, 2010). For instance, prior studies have shown that repartnering after divorce, either through remarriage or through cohabitation, significantly reduces distress and depressive symptoms and increases life satisfaction regardless of whether the divorce occurred earlier or later in life, although the improvement seems to be temporary (Blekesaune, 2008; Kalmijn, 2017; Lin et al., 2019). Future research should consider other turning points (e.g., retirement, empty nest, health decline) that may accelerate or decelerate the protracted recovery and examine how long the effects last.

Last, future research should examine whether the convalescence model can be applied to dissolution of other types of unions, including cohabitation, same-sex unions, and non-coresidential partnerships (i.e., dating and living apart together, or LAT). Cohabitation has grown rapidly among older adults, increasing from 1 million in 2000 (Brown, Lee, & Bulanda, 2006) to 4 million in 2016 (Stepler, 2017). Same-gender unions are also probably on the rise, although estimates for older adults are scant. A 2015 study estimated that 0.4% of U.S. adults aged 50 and older were in same-gender cohabiting unions in 2013 (Manning & Brown, 2015). There are no national estimates for the numbers of older adults in dating or LAT relationships, but these relationship forms are increasingly common in later life (Brown & Wright, 2017). If relationships reflect a continuum of social attachment and commitment (Ross, 1995) where the married arguably have the highest level of commitment and daters have the lowest, then do people in relationships with different levels of commitment experience different adjustment trajectories after dissolution?

Conclusion

Gray divorce is on the rise and widowhood is waning, yet gerontological research largely retains its singular focus on widowhood. Despite the recent acceleration in divorce among older adults, theoretical developments have not kept pace. In particular, the consequences of divorce for older adults remain poorly understood. A handful of longitudinal studies have uncovered a pattern of postdissolution adjustment among older adults that is incongruous with the acute versus chronic grief models in the bereavement literature and the crisis versus chronic strain models in the stress-adjustment perspective on divorce (Lin et al., 2019; Sasson & Umberson, 2014). This third adjustment pathway, the convalescence model (Lin et al., 2019), describes a protracted but eventual recovery.

Drawing on Pearlin’s stress process perspective (Pearlin et al., 1981), we offer a rationale for the theoretical underpinnings of the convalescence model. Old age is often associated with a time of loss. Declining health and cognitive functioning, combined with the loss of family members and friends, diminish the personal and social resources that older adults have to mitigate the negative impacts of marital dissolution. Yet most older adults are resilient, as they have better emotion regulation than younger adults for coping with stressful life events (Carstensen, 1992; Charles, 2010; Mirowsky & Ross, 2001). Thus, the convalescence model appropriately describes this long, slow recovery in the aftermath of later-life marital dissolution.

Theoretical innovations such as the introduction of the convalescence model have the potential to transform practice and policy (Roberto et al., 2006). The growing empirical evidence that adjustment to widowhood or divorce is often a prolonged process could inform responses of both practitioners and policy makers on the front lines of ensuring the health and well-being of vulnerable elders who have experienced marital disruption. By conceptually developing the convalescence model, researchers can illuminate the mechanisms undergirding the slow path to recovery for older adults after marital dissolution, identifying specific resources and challenges that shape the arc of the recovery process and that may be malleable for interventionists, service providers, and social policy.

At the same time, practitioners would benefit from awareness that many older adults face an extended period of recovery following divorce or spousal loss in later life that persists for several years. Working in concert, researchers and practitioners can create evidence-based programs to support these older adults, who are likely at higher risk of economic precarity and social isolation (Connidis & Barnett, 2019; Lin, Brown, & Hammersmith, 2017; Wright, Hammersmith, Brown, & Lin, 2019). These risks may affect older adults unevenly, with the most disadvantaged taking the longest to recover, which underscores the importance of devising services and policies that support an increasingly diverse older adult population whose needs and vulnerabilities differ across racial-ethnic, gender, social class, and other dimensions of privilege and oppression (Connidis & Barnett, 2019). Likewise, levels of formal and informal support differ according to one’s partnership status and history (Connidis & Barnett, 2019). Trends such as gray divorce are contributing to the rise of kinlessness among older adults (Margolis & Verdery, 2017), which signals a lack of family members to provide care in later life that will shift the burden of caregiving to formal sources of support. Even as the availability of informal sources of support shrinks, the need for such support is expanding, as people live longer with fewer family members and in particular being less likely to have a spouse (Connidis & Barnett, 2019).

The rapidly changing familial experiences of older adults, including marital dissolution that increasingly occurs through gray divorce, necessitate novel policy responses. Our assertion that adjustment to marital dissolution in later life is often protracted heightens the salience of developing innovative policy solutions that bolster support to the aged, especially those who lack the informal, familial sources of support that older adults have traditionally relied on. A recent assessment of the family policy landscape identified family caregiving of the aged as the topic that is likely to take center stage in the family policy arena in the years to come (Ooms, 2019). Indeed, the policy challenges presented by an aging population are compounded by the failure of social policy to keep pace with the growing complexity of families, young and old alike (Bogenschneider & Wadsworth, 2019). The majority of caregiving is provided at home by family members, but caregivers are frequently at the margins of both social services and health-care delivery (Ooms, 2019), which may compound stress and worsen the health of both caregivers and care recipients. Weaving together informal and formal supports takes on a new urgency for service providers and policy makers in an era of changing family ties that are fundamentally reshaping the aging experience.

As the aging of the population continues to accelerate and the marital narrative of older adults is increasingly varied and diverse, it is essential that family gerontological theories reflect these new realities. The sizable literature on the outcomes associated with widowhood and divorce has remained rather static from a theoretical standpoint, with the classic models of acute grief and crisis as well as chronic grief and strain firmly entrenched. Despite extensive research, alternative conceptualizations, including the convalescence model, have been absent from the literature until quite recently. We argue that both conceptual and methodological challenges have constrained investigations of the adjustment trajectories following marital dissolution. The emergence of the convalescence model, coupled with the availability of longitudinal data and analytic strategies that track within-person change over time, provides an opportunity for scholars to pursue new frontiers in research on the consequences of marital dissolution for older adults that will have important ramifications for practice and policy.