Abstract
Mental health professionals have suggested that widowed persons experience heightened psychological distress on dates that had special meaning for them and their late spouse, such as a wedding anniversary or the late spouse’s birthday. This study examined the effects of such occasions on grief, anxiety, and depressive symptoms in a community sample of older widowed persons.
OLS regression models were estimated using data from the Changing Lives of Older Couples (CLOC) study, a large prospective probability study of late-life widowhood. Participants were interviewed prior to and both 6 and 18 months after spousal loss; married matched controls were interviewed at comparable times.
Widowed persons reported heightened psychological distress when interviewed during the month of their late spouse’s birthday, a post-holiday period (January), and in June, a month associated with wedding anniversaries and graduations in the United States. The distressing effects of special occasions on psychological symptoms were evidenced only within the first 6 months postloss, and were not apparent at the 18-month follow-up.
Our results support the clinical observation that persons in the early stages of spousal bereavement are at increased risk of psychological distress at times with special significance to the couple. We highlight methodological and clinical implications.
Mental health professionals and clinicians frequently report that bereaved clients have heightened feelings of grief and sadness on dates that remind them of their lost loved ones (Earls & Wolf, 1963; Hertz, 2002; Musaph, 1990; Rabow, Hauser, & Adams, 2004). Self-help books and Web sites often warn bereaved persons that they may experience increased sadness on their “loved one’s birthday or other special days throughout the year” (Mayo Clinic, 2012). Grief theories suggest that these special dates may revive thoughts of the deceased, with accompanying feelings of sadness associated with the loss (Chow, 2010; Cook, 2001; Rosenblatt, 1983; Sanders, 1999; Simon, 2012). However, we know of no systematic empirical assessments of this claim. In this article, we examine whether older widowed persons experience heightened symptoms of anxiety, depression, and grief during time periods surrounding three important occasions: the late spouse’s birthday, the winter holiday season (months of December and January), and the month of June, a time traditionally associated with family celebrations including wedding anniversaries and graduations. We use data from the Changing Lives of Older Couples study (CLOC), a prospective multiwave study of older widow(er)s and matched controls. The multiwave data allow us to explore the time course of special occasion grief symptoms, and the widow–control design allows us to explore whether seasonal variations in psychological distress are specific to bereaved older adults, or whether such variations reflect general patterns including seasonal affective disorder (SAD; Rosenthal et al., 1984).
Of the significant events that a widowed person encounters during the first year postloss, the anniversary of the death has garnered the most attention. Studies generally concur that the anniversary of the death is a particularly difficult time for bereaved people (Archer, 1999; Cook, 2001). Jacobs, Schaefer, and Ostfeld (1987) studied a group of 108 widows and widowers 13 months postloss, to explore the impact of the death anniversary. Observances of the anniversary were common (88%) and were associated with depressed mood and social functioning disturbances.
Recognition of the 1-year anniversary phenomenon has implications for scientific practice, as well. Bereavement researchers typically schedule follow-up interviews with widowed subjects shortly before (i.e., 9–10 months) or after (i.e., 13 months) the 1-year anniversary (Carr, Nesse, & Wortman, 2006). To conduct an interview near the 1-year anniversary of the death could potentially bias symptoms upward, or cause undue distress to the study participant (Shuchter, 1986). The CLOC study conducted follow-up interviews 6 and 18 months postloss to avoid death anniversary effects. However, the interviews often occurred near other significant events that could affect a widow(er)’s psychological symptoms during the first 2 years postloss.
The impact of special occasions including the late spouse’s birthday, family holidays, and wedding anniversaries has not been investigated empirically. Several descriptive and clinical accounts suggest the importance of these occasions for bereaved individuals, however. For example, Fox (1984) explored children’s grief responses and found that subjects described their parents’ birthdays as difficult. We expect that the birthday of a deceased spouse also may be a meaningful date; it may revive memories of the last birthday celebrated, of happy times shared as a married couple, and the fact that the decedent will no longer experience future birthdays. The widowed person’s friends and family members may not recall the late spouse’s birthday, and the day may pass unobserved by others, creating an increased sense of isolation.
Wedding anniversaries are another occasion that may rekindle thoughts of the decedent and accompanying feelings of grief or sadness. The wedding anniversary is an annual reminder of what one has lost—their long-term relationship with the deceased. For an older widow(er), the wedding date may have been jointly celebrated for decades, but is no longer acknowledged in the same fashion. Thus, the wedding anniversary and the birthday of the deceased spouse may both be occasions marked by increased feelings of sadness and loss.
In addition to dates specific to the couple, holidays and other family celebrations may be particularly difficult for the bereaved. One study of a holiday outreach program found that grief and bereavement were the primary themes of calls placed to a hotline for lonely older adults during the season of Christmas and Hanukkah (Loring, Smith, & Thomas, 1994). The period immediately after such family centered holidays also may be particularly distressing for older widow(er)s. Widowhood may exacerbate the loneliness experienced when holiday activities cease and visiting family members return home. A review of studies evaluating emotional distress during the December holiday season suggests that there is a rise in depressed mood in the period immediate following—rather than during—this period (Friedberg, 1990). However, we know of no studies that have explored holiday distress among recently widowed persons.
Although widowed persons may be at heightened risk of emotional distress in the period around and following the December holidays, the winter season may influence depressive symptomatology in older adults, more generally. Depressive symptoms may be more frequent or intense in the winter months due to SAD (Harmatz, Well, Overtree, Kawamura, Rosal, & Ockene, 2000). The widow–control design of the CLOC study allows us to assess whether seasonal variation in psychological symptoms is more acute among the recently bereaved.
Our Study
To our knowledge, no study has systematically examined the effects of special occasions on psychological symptoms among elderly bereaved persons. This line of inquiry is important for both research and practice. First, if special occasion reactions are an empirically valid phenomenon, they may represent a significant source of variance in bereavement studies and should be accounted for in study designs (Cook, 2001). Second, documenting and understanding such reactions may improve clinicians’ capacity to develop effective and timely interventions at both the individual and population levels.
The CLOC data have distinct advantages for investigating grief reactions to special events or time periods. First, it is a prospective longitudinal study of older adults with detailed information on pre-bereavement functioning, thus it is possible to distinguish within-person change in psychological health pre- and postloss. Second, the sample of bereaved subjects is derived from a representative sample of older adults living in the Detroit Standardized Metropolitan Statistical Area (SMSA), making the results more generalizable than studies based on clinical or self-help samples.
Third, CLOC obtains both general and loss-specific psychological measures; the former includes depressive symptoms and anxiety, whereas the latter includes three subcomponents of grief including yearning, loss-related anxiety, and despair. Theoretical writings on grief identify its key components as feelings of sorrow or sadness, anxiety (or “active distress”), a yearning or longing for the deceased, and a sense of emotional despair or emptiness (Bowlby, 1980; Parkes, 1985). Thus, the data allow us to pinpoint the distinctive symptoms that may be triggered by meaningful occasions postloss. Fourth, CLOC includes bereaved persons and matched married controls, which enables us to distinguish the effects of general seasonal variations in mood, from those related directly to potentially distressing special occasions (Rosenthal et al., 1984).
Finally, the CLOC obtains comprehensive measures of health, well-being, socioeconomic status, and other personal characteristics that are well-documented correlates of psychological adjustment to loss (Carr et al., 2006). Although these correlates of grief symptoms are not plausibly linked to timing of interview, the inclusion of these control measures may help reduce error variance in our study outcomes. Thus, all analyses are adjusted for age, gender, socioeconomic status, own health, spouse’s health (at baseline), and preloss history of depressive symptoms and major depression among the bereaved (Carr et al., 2006).
Our overarching hypothesis is that time periods potentially associated with memories of the late spouse will lead to heightened symptoms of depression, anxiety, and grief during the first year of bereavement. We expect that interviews conducted during the following periods will evince higher levels of psychological distress than interviews conducted at other time periods: within 1 month of a deceased spouse’s birthday; the holiday month of December; the post-holiday month of January; and the month of June, typically associated with weddings and family celebrations such as graduations. The 1-month time frames allow for adequate sample sizes.
The CLOC obtained information on the year but not month of one’s marriage. However, supplemental analyses of data from the Wisconsin Longitudinal Study, another study of older adults conducted in the upper Midwest (and thus has a climate similar to the Detroit MSA) revealed that June was by far the most common month for weddings. More than 16% of weddings occurred in June, a rate twice that of chance (i.e., an 8% chance of a wedding occurring in one of the twelve months of the year).
We further expect that distress associated with special time periods will be evident during the first year after the death of a spouse, yet effects will abate during the second year. Research shows persuasively that symptoms of distress are most acute during the first year following loss (Maciejewski, Zhang, Block, & Prigerson, 2007). Consistent with this pattern, we expect that the potential impact of a birthday, holiday, or wedding anniversary period may subside over time, and will be most pronounced during the first celebration of such an event after the loss.
Methods
Sample
The CLOC is a prospective study of a two-stage area probability sample of 1,532 married individuals from the Detroit SMSA. Respondents were noninstitutionalized English-speaking members of a married couple where the husband was aged 65 or older. Approximately 65% of those contacted for interview participated, consistent with response rates from other Detroit area studies. Baseline face-to-face interviews were conducted in 1987 and 1988. After baseline interviews were completed, CLOC investigators monitored spousal loss by reading obituaries in three Detroit-area newspapers and using monthly death record tapes provided by the State of Michigan. The National Death Index was used to confirm deaths. Women were oversampled at baseline to maximize the number of participants who would become widowed during the study. The data are weighted to adjust for unequal probabilities of selection.
Widowed persons were matched with still-married persons from the baseline sample, by age, race, and sex. The married matched controls were reinterviewed at three follow-up interviews at roughly the same time as their corresponding widowed persons: 6, 18, and 48 months postloss. This matched control–widow strategy enables researchers to explore the distinctive effects of widowhood versus within-person changes due to aging processes.
The CLOC has modest attrition; primary reasons for nonresponse at the 6-month follow-up were refusal to participate (38%) and ill health or death (42%). Supplementary analyses reveal that age and baseline anxiety increase the odds, and home ownership decreases the odds of attrition. Thus, the analytic sample is overrepresentative of slightly younger–old bereaved or married spouses who are residentially stable and have lower levels of anxiety.
We use two analytic samples; the 297 persons (210 widowed persons and 87 matched controls) interviewed at the 6-month follow-up, and the 370 individuals (184 widowed persons and 186 married controls) interviewed at the 18-month follow-up. The relatively small number of controls at the first follow-up reflects a funding cut from the data collection budget, which was reinstated at the following wave.
Measures
Dependent variables.—
Psychological distress is assessed with two general (depressive symptoms and anxiety) and three loss-related (yearning, despair, and loss-related anxiety) outcomes. The two general psychological outcomes are used in analyses contrasting the widowed–control sample, whereas all five outcomes are used in analyses focused on the widowed sample only.
Depressive symptoms (α = .77 at baseline, α = .83 at 6 months, α = .75 at 18 months) were assessed with a subset of nine items from the Center for Epidemiologic Studies depression (CES-D) scale (Radloff, 1977). Respondents indicated how often they experienced nine symptoms in the week prior to interview: felt depressed, everything was an effort, restless sleep, lonely, felt people were unfriendly, did not feel like eating, felt sad, felt that people disliked me, and could not get going. Response categories were hardly ever, some of the time, or most of the time.
Anxiety (α = .86 at baseline, α = .84 at 6 months, α = .80 at 18 months) was assessed with 10 items from the Symptom Checklist 90 Revised (SCL-90-R) (Derogatis & Cleary, 1977). Respondents were asked to indicate how often they experienced each of the following symptoms in the week prior to interview: nervousness or shakiness, trembling, suddenly scared for no reason, fearful, heart pounding or racing, tense and keyed up, spells of terror and panic, so restless you couldn’t sit still, feeling that something bad is going to happen to you, and thoughts and images of a frightening nature. Both scales are standardized, where M = 0 and SD = 1.
Three psychological symptoms related directly to the loss were measured for bereaved persons only. Each scale captures the frequency of symptoms in the month prior to interview. Yearning (α = .75) is a four-item scale that refers to whether one: was longing to have your spouse with you; had painful waves of missing spouse; experienced feelings of intense pain or grief over the loss of spouse; and experienced feelings of grief, loneliness, or missing your spouse. Loss-related anxiety (α = .71) comprises three symptoms: afraid of what lies ahead, extremely anxious and unsettled, and worried about managing day-to-day affairs. Despair (α = .64) comprises three symptoms: feeling that life has lost its meaning; feeling empty inside, like an important part of you is missing; and feeling that life has lost its meaning. Response categories for all grief scale items were as follows: no, never; yes, but rarely, yes, sometimes; and yes, often. Items were derived from widely used grief measures including the Bereavement Index (Jacobs et al., 1986), Present Feelings About Loss (Singh & Raphael, 1981), and the Texas Revised Inventory of Grief (Zisook, DeVaul, & Click, 1982).
Independent variables.—
The key independent variables are widowhood and the date of interview. Widowhood is a dummy variable indicating those who were widowed between the baseline interview and the 6-month follow-up. Timing of interview includes five time points: within 30 days of the late spouse’s birthday, the month of December, the month of January, and the month of June, and all other times (reference category).
Control variables. All analyses are controlled for health at baseline, as well as demographic and socioeconomic characteristics that have been found elsewhere to affect psychological well-being of older adults (Carr et al., 2006). Depressive and anxiety symptoms at baseline were measured the same way as at the 6-month follow-up (Derogatis & Cleary, 1977; Radloff, 1977). Thus, multivariate models for which depressive symptoms and anxiety symptoms are outcomes reveal changes in symptoms pre- and postloss. We also adjust for whether one had at least one major lifetime depressive episode (MDE) prior to the baseline interview. MDE is a dichotomous variable indicating whether one ever had a period of 2 weeks or more when they felt depressed, and they reported at least four of 15 accompanying symptoms typically associated with MDE (e.g., diminished interest in activities and diminished ability to concentrate). We include both measures to better capture both recent and early history of depression, given research suggesting that both may have distinctive effects on adjustment to loss (Bonanno & Kaltman, 2001).
Own and spouse’s physical health at baseline are assessed with the question, “How would you rate your/your spouse’s health at the present time? Would you say it is excellent, good, fair, or poor?” Sociodemographic controls include: the respondent’s age, sex (1 = female), education (years completed), and total household income (natural log) at baseline. Analyses were adjusted for the duration (in months) between the baseline and 6-month interviews. All follow-up interviews were conducted 6 and 18 months following the death of a spouse, yet the time between the baseline and 6-month interviews ranged from 9 to 76 months. Thus, baseline assessments were more temporally distant for respondents who lost a spouse at later dates.
Analytic plan.—
We first conducted bivariate analyses, to contrast the psychological symptoms reported at the 6-month follow-up by bereaved persons in each special occasion interview period, compared with those reported during the remaining times of the year. The bivariate analyses also contrast the depressive and anxiety symptoms reported by widowed persons and matched controls at the 6-month follow-up. Second, we use ordinary least squares (OLS) regression models to contrast widowed and married persons with respect to depressive and anxiety symptoms, and whether these differences were contingent upon the timing of the interview. The evaluation of two-way interaction terms allows us to formally evaluate whether the impact of interview timing on distress symptoms is distinct to widowed persons, or whether these patterns reflect more general seasonal fluctuations in mood. Third, we use OLS regression models to evaluate the effect of each special occasion period on each of the general and loss-related psychological outcomes among bereaved persons 6 months postloss, net of all controls. We conducted parallel analyses for all psychological outcomes at the 18-month follow-up and found no statistically significant patterns. Thus, all analyses presented here focus on the 6-month outcomes only.
Results
Sample Characteristics
Of the 210 widowed persons interviewed 6 months after the loss of a spouse, 13.8% (n = 29) were interviewed within a month of the deceased spouse’s birthday, 4.8% (n = 10) in December (i.e., holidays), 7.1% (n = 15) in January (i.e., post holidays), 10% (n = 21) in June, and 64% (n = 135) at all other times. The CLOC study team did not schedule interviews during the week between Christmas and New Year’s, which may account for the low number of interviews completed in December. The mean age is 69 years, and average educational attainment is 11.7 years. Respondents had been married an average of 42 years.
Bivariate Analysis
Table 1 displays the unadjusted means for all psychological outcomes among widowed persons, broken down by interview date. The far right-hand columns include the overall mean psychological outcomes for all bereaved persons (n = 210) and means for the general psychological outcomes for matched controls (n = 87) at the 6-month follow-up. Two-tailed t tests were conducted to evaluate significant differences between the widowed and married matched control participants. Consistent with prior studies, widowed participants report significantly more symptoms of depression (.412 vs −.143, p < .000) at the 6-month follow-up compared with married matched controls. Anxiety symptoms do not differ significantly.
Means (and Standard Deviations) for Psychological Outcomes at 6-Month Follow-up, Changing Lives of Older Couples (CLOC) Study
| Nonspecial occasion (n = 75) | Within 30 days of spouse birthday (n = 29) | Widowed persons only | Total sample | ||||
|---|---|---|---|---|---|---|---|
| December (n = 10) | January (n = 15) | June (n = 21) | All widowed (n = 210) | All married controls (n = 87) | |||
| Loss-related symptoms | |||||||
| Yearning | 2.52 (.845) | 2.90 (.815) | 2.72 (.702) | 3.07 (.712) | 3.27* (.745) | 2.81 (.815) | n.a. |
| Loss-related anxiety | 1.62 (.831) | 1.88 (1.03) | 1.57 (.535) | 2.47* (1.21) | 2.29* (1.05) | 1.78 (.890) | n.a. |
| Despair | 2.46 (.895) | 2.87† (.728) | 2.58 (.812) | 2.97 (.716) | 3.22** (.836) | 2.63 (.851) | n.a. |
| General psychological symptoms | |||||||
| Depressive symptoms (CES-D) | .09 (1.11) | .796† (1.50) | .577 (1.02) | 1.51** (1.64) | 1.10* (1.33) | .412 (1.21) | −.143*** (.985) |
| Anxiety (SCL-90) | −.175 (.687) | .310 (1.42) | −.162 (.526) | .413 (1.41) | .388 (1.25) | −.010 (.888) | −.133 (.852) |
| Nonspecial occasion (n = 75) | Within 30 days of spouse birthday (n = 29) | Widowed persons only | Total sample | ||||
|---|---|---|---|---|---|---|---|
| December (n = 10) | January (n = 15) | June (n = 21) | All widowed (n = 210) | All married controls (n = 87) | |||
| Loss-related symptoms | |||||||
| Yearning | 2.52 (.845) | 2.90 (.815) | 2.72 (.702) | 3.07 (.712) | 3.27* (.745) | 2.81 (.815) | n.a. |
| Loss-related anxiety | 1.62 (.831) | 1.88 (1.03) | 1.57 (.535) | 2.47* (1.21) | 2.29* (1.05) | 1.78 (.890) | n.a. |
| Despair | 2.46 (.895) | 2.87† (.728) | 2.58 (.812) | 2.97 (.716) | 3.22** (.836) | 2.63 (.851) | n.a. |
| General psychological symptoms | |||||||
| Depressive symptoms (CES-D) | .09 (1.11) | .796† (1.50) | .577 (1.02) | 1.51** (1.64) | 1.10* (1.33) | .412 (1.21) | −.143*** (.985) |
| Anxiety (SCL-90) | −.175 (.687) | .310 (1.42) | −.162 (.526) | .413 (1.41) | .388 (1.25) | −.010 (.888) | −.133 (.852) |
Notes. t Tests were used to identify statistically significant differences between date of interview categories, where the omitted category includes persons interviewed in months other than January, June, December, and month of late spouse’s birthday, labeled here as “nonspecial occasion” dates. t Tests also were used to compare the psychological symptoms of the widowed and married matched control samples. Depressive symptoms and anxiety symptoms are standardized where overall M = 0 and SD = 1.
*p < .05. **p < .01. ***p < .001. †p < .10.
Means (and Standard Deviations) for Psychological Outcomes at 6-Month Follow-up, Changing Lives of Older Couples (CLOC) Study
| Nonspecial occasion (n = 75) | Within 30 days of spouse birthday (n = 29) | Widowed persons only | Total sample | ||||
|---|---|---|---|---|---|---|---|
| December (n = 10) | January (n = 15) | June (n = 21) | All widowed (n = 210) | All married controls (n = 87) | |||
| Loss-related symptoms | |||||||
| Yearning | 2.52 (.845) | 2.90 (.815) | 2.72 (.702) | 3.07 (.712) | 3.27* (.745) | 2.81 (.815) | n.a. |
| Loss-related anxiety | 1.62 (.831) | 1.88 (1.03) | 1.57 (.535) | 2.47* (1.21) | 2.29* (1.05) | 1.78 (.890) | n.a. |
| Despair | 2.46 (.895) | 2.87† (.728) | 2.58 (.812) | 2.97 (.716) | 3.22** (.836) | 2.63 (.851) | n.a. |
| General psychological symptoms | |||||||
| Depressive symptoms (CES-D) | .09 (1.11) | .796† (1.50) | .577 (1.02) | 1.51** (1.64) | 1.10* (1.33) | .412 (1.21) | −.143*** (.985) |
| Anxiety (SCL-90) | −.175 (.687) | .310 (1.42) | −.162 (.526) | .413 (1.41) | .388 (1.25) | −.010 (.888) | −.133 (.852) |
| Nonspecial occasion (n = 75) | Within 30 days of spouse birthday (n = 29) | Widowed persons only | Total sample | ||||
|---|---|---|---|---|---|---|---|
| December (n = 10) | January (n = 15) | June (n = 21) | All widowed (n = 210) | All married controls (n = 87) | |||
| Loss-related symptoms | |||||||
| Yearning | 2.52 (.845) | 2.90 (.815) | 2.72 (.702) | 3.07 (.712) | 3.27* (.745) | 2.81 (.815) | n.a. |
| Loss-related anxiety | 1.62 (.831) | 1.88 (1.03) | 1.57 (.535) | 2.47* (1.21) | 2.29* (1.05) | 1.78 (.890) | n.a. |
| Despair | 2.46 (.895) | 2.87† (.728) | 2.58 (.812) | 2.97 (.716) | 3.22** (.836) | 2.63 (.851) | n.a. |
| General psychological symptoms | |||||||
| Depressive symptoms (CES-D) | .09 (1.11) | .796† (1.50) | .577 (1.02) | 1.51** (1.64) | 1.10* (1.33) | .412 (1.21) | −.143*** (.985) |
| Anxiety (SCL-90) | −.175 (.687) | .310 (1.42) | −.162 (.526) | .413 (1.41) | .388 (1.25) | −.010 (.888) | −.133 (.852) |
Notes. t Tests were used to identify statistically significant differences between date of interview categories, where the omitted category includes persons interviewed in months other than January, June, December, and month of late spouse’s birthday, labeled here as “nonspecial occasion” dates. t Tests also were used to compare the psychological symptoms of the widowed and married matched control samples. Depressive symptoms and anxiety symptoms are standardized where overall M = 0 and SD = 1.
*p < .05. **p < .01. ***p < .001. †p < .10.
Among widowed persons, those interviewed within a month of their late spouse’s birthday reported elevated symptoms of despair and depression compared with those interviewed in the nonspecial event period (2.87 vs. 2.46, and .796 vs .09, respectively), and these differences approached statistical significance (p < .10). However, we found no evidence that they reported significantly higher levels of yearning, loss-related anxiety, or general anxiety, compared with persons interviewed in times not considered special occasions.
We found some evidence of seasonal variation in symptoms among bereaved spouses, with the exception of December. Those interviewed in December did not differ significantly from persons interviewed during other times of year. However, interviews conducted in January and June elicited significantly more symptoms than interviews conducted at other points in the year. January interviews elicited significantly higher levels of loss-related anxiety (2.47 vs 1.62, p < .05) and depressed mood (1.51 vs 09, p < .05), respectively. Widowed persons who were interviewed in June reported significantly more symptoms of yearning, loss-related anxiety, despair, and depression, compared with the reference category.
In Figure 1, we display the (unadjusted) average number of depressive symptoms reported for each of the 12 months in which the follow-up interviews of bereaved persons occurred. Interviews conducted in January (M = 1.51) and June (M = 1.10) yielded the highest mean number of depressive symptoms for the year. Analysis of variance reached significance (F = 2.70; p =. 003) for omnibus differences between months on depressive mood. Post hoc tests showed that all months except June were significantly lower than January.
Depressive symptoms (standardized) at 6-month follow-up, by month of interview, widowed participants in the CLOC (n = 210).
Multivariate Analyses
Interview timing effects on psychological symptoms of widowed versus married controls.—
We next estimated OLS regression models to evaluate whether the effect of interview timing on psychological distress differed significantly for bereaved spouses versus married matched controls. We estimated separate models evaluating the two-way interaction of widowhood status by January interview, June interview, and interview conducted within one month of spouse’s birthday. We did not evaluate interaction terms for December interviews, as the bivariate analyses revealed that December interviews did not yield significantly different reports of depressive symptoms relative to other months. We estimated models for the two general psychological outcomes: depressive and anxiety symptoms. Of the six possible interaction terms evaluated (each of the three dates by each of the two outcomes), only one approached statistical significance. Interviews conducted within one month of the spouse’s birthday were associated with more frequent depressive symptoms among the widowed than among married controls (p < .07). However, these results should be interpreted cautiously due to small sample size. Likewise, the lack of statistically significant interactions for the June and January interviews, and for the outcome of anxiety also may reflect small sample size.
Table 2 shows the effects of the two-way interaction of spouse’s birthday month interview and widowhood status (net of controls) on depressive symptoms 6 months postloss. The interaction term values are plotted in Figure 2. Widowed persons who completed an interview within a month of their late spouse’s birthday evidence depressive symptoms that are nearly a half standard deviation higher than widowed persons interviewed at other times, and nearly two-thirds of a standard deviation higher than married persons interviewed during the month of their spouse’s birthday. By contrast, among married controls, persons interviewed during their spouse’s birthday month evidence slightly fewer (b = −.188) depressive symptoms relative to their counterparts interviewed at other times (i.e., reference group), although this effect is not statistically significant. Thus, we find suggestive evidence that the month of a spouse’s birthday is protective for married persons, yet distressing for the recently bereaved.
OLS Regression Predicting Effect of Widowhood and Being Interviewed Within 30 Days of Spouse’s Birthday, on Depressive Symptoms (CES-D, Standardized), 6-Month Follow-up (N = 297)
| Widowed | −.07 |
| Interview within 30 days of spouse birthday | −.188 |
| Widowed * interviewed within 30 days of birthday | .723† |
| Control variables | |
| Age (in years) | .01 |
| Education (in years) | .01 |
| Total household income (natural log) | −.034 |
| Self-rated health (1 = excellent, 5 = poor) | .109 |
| Spouse’s health (1 = excellent, 5 = poor) | −.02 |
| Sex (1 = female) | .05 |
| Depressive symptoms at baseline | .323*** |
| History of MDE | .356** |
| Months between baseline and Wave 1 interview | .01* |
| Constant | −1.52 |
| Widowed | −.07 |
| Interview within 30 days of spouse birthday | −.188 |
| Widowed * interviewed within 30 days of birthday | .723† |
| Control variables | |
| Age (in years) | .01 |
| Education (in years) | .01 |
| Total household income (natural log) | −.034 |
| Self-rated health (1 = excellent, 5 = poor) | .109 |
| Spouse’s health (1 = excellent, 5 = poor) | −.02 |
| Sex (1 = female) | .05 |
| Depressive symptoms at baseline | .323*** |
| History of MDE | .356** |
| Months between baseline and Wave 1 interview | .01* |
| Constant | −1.52 |
Notes.†p < .10. *p < .05. **p < .01. ***p < .001.
OLS Regression Predicting Effect of Widowhood and Being Interviewed Within 30 Days of Spouse’s Birthday, on Depressive Symptoms (CES-D, Standardized), 6-Month Follow-up (N = 297)
| Widowed | −.07 |
| Interview within 30 days of spouse birthday | −.188 |
| Widowed * interviewed within 30 days of birthday | .723† |
| Control variables | |
| Age (in years) | .01 |
| Education (in years) | .01 |
| Total household income (natural log) | −.034 |
| Self-rated health (1 = excellent, 5 = poor) | .109 |
| Spouse’s health (1 = excellent, 5 = poor) | −.02 |
| Sex (1 = female) | .05 |
| Depressive symptoms at baseline | .323*** |
| History of MDE | .356** |
| Months between baseline and Wave 1 interview | .01* |
| Constant | −1.52 |
| Widowed | −.07 |
| Interview within 30 days of spouse birthday | −.188 |
| Widowed * interviewed within 30 days of birthday | .723† |
| Control variables | |
| Age (in years) | .01 |
| Education (in years) | .01 |
| Total household income (natural log) | −.034 |
| Self-rated health (1 = excellent, 5 = poor) | .109 |
| Spouse’s health (1 = excellent, 5 = poor) | −.02 |
| Sex (1 = female) | .05 |
| Depressive symptoms at baseline | .323*** |
| History of MDE | .356** |
| Months between baseline and Wave 1 interview | .01* |
| Constant | −1.52 |
Notes.†p < .10. *p < .05. **p < .01. ***p < .001.
Effect of depressive symptoms (standardized) at 6-month follow-up, by widowhood status and whether interview was conducted during spouse’s birthday month.
Grief and distress symptoms during special occasion months among the bereaved.—
Our final aim is to evaluate whether the interviews conducted during the months of special occasions are associated with specific grief symptoms among newly bereaved widows and widowers, net of controls. As noted earlier, preliminary analyses showed that interview timing was not significantly associated with grief symptoms at the 18-month follow-up, revealing that anniversary or special occasion reactions are limited to the first year postloss. The OLS regression models in Table 3 estimate the effects of interview timing on psychological symptoms 6 months postloss only. Relative to persons completing interviews during other times of the year, bereaved spouses interviewed in their spouse’s birthday month, January and June evidence elevated symptoms of grief and distress although the specific symptoms manifested differ slightly across outcomes.
Summary of OLS Regression Models Predicting the Effect of Interview Timing on Loss-Related Psychological Outcomes, Adjusted for Sociodemographic and Health Characteristics, Widowed Sample (N = 210)
| Yearning | Despair | Loss-related anxiety | Anxiety | Depressive symptoms | |
|---|---|---|---|---|---|
| Within 30 days of spouse’s birthday | .157 | .366* | .085 | .466** | .530* |
| January interview | .389† | .427† | .663** | .394 | 1.164*** |
| June interview | .436* | .456* | .289 | .135 | .404 |
| Yearning | Despair | Loss-related anxiety | Anxiety | Depressive symptoms | |
|---|---|---|---|---|---|
| Within 30 days of spouse’s birthday | .157 | .366* | .085 | .466** | .530* |
| January interview | .389† | .427† | .663** | .394 | 1.164*** |
| June interview | .436* | .456* | .289 | .135 | .404 |
Notes. Regression models are adjusted for age, sex, education, income, own self-rated health, spouse’s self-rated health, baseline depressive symptoms (for all outcomes except anxiety), baseline anxiety symptoms (for outcome of anxiety only), history of Major Depression Episode (MDE), and months elapsed between baseline 6-month follow-up interview. Depressive symptoms and anxiety symptoms are standardized where overall M = 0 and SD = 1.
†p < .10. *p < .05. **p < .01. ***p < .001.
Summary of OLS Regression Models Predicting the Effect of Interview Timing on Loss-Related Psychological Outcomes, Adjusted for Sociodemographic and Health Characteristics, Widowed Sample (N = 210)
| Yearning | Despair | Loss-related anxiety | Anxiety | Depressive symptoms | |
|---|---|---|---|---|---|
| Within 30 days of spouse’s birthday | .157 | .366* | .085 | .466** | .530* |
| January interview | .389† | .427† | .663** | .394 | 1.164*** |
| June interview | .436* | .456* | .289 | .135 | .404 |
| Yearning | Despair | Loss-related anxiety | Anxiety | Depressive symptoms | |
|---|---|---|---|---|---|
| Within 30 days of spouse’s birthday | .157 | .366* | .085 | .466** | .530* |
| January interview | .389† | .427† | .663** | .394 | 1.164*** |
| June interview | .436* | .456* | .289 | .135 | .404 |
Notes. Regression models are adjusted for age, sex, education, income, own self-rated health, spouse’s self-rated health, baseline depressive symptoms (for all outcomes except anxiety), baseline anxiety symptoms (for outcome of anxiety only), history of Major Depression Episode (MDE), and months elapsed between baseline 6-month follow-up interview. Depressive symptoms and anxiety symptoms are standardized where overall M = 0 and SD = 1.
†p < .10. *p < .05. **p < .01. ***p < .001.
Interviews conducted within 1 month of the late spouse’s birthday are associated with significantly more frequent symptoms of despair, anxiety, and depressive symptoms, whereas interviews conducted in January evidence more frequent symptoms of all five outcomes except anxiety. Interviews conducted in June are significantly associated with symptoms of yearning and despair, but neither form of anxiety nor depression. Interestingly, only one of the five outcomes—despair—is significantly or marginally significantly related to all three indicators of interview timing. The two anxiety outcomes are associated with just one interview date each; loss-related anxiety symptoms are higher among persons interviewed in January, whereas general anxiety symptoms are higher within 1 month of a late spouse’s birthday. Yearning levels are significantly higher when a bereaved person is interviewed in June, and marginally higher when interviewed in January. Depressive symptoms, by contrast, are associated with birthday month and January interviews. Each of these effects remains statistically significant after sociodemographic factors, health, and preloss mental health symptoms are controlled.
Discussion
This study provides some empirical support for the clinical observation that particular special occasions and time periods are especially difficult for bereaved spouses, at least during the early stages postloss. Three main findings emerged from our analysis. First, among bereaved spouses, a range of psychological symptoms including depression, anxiety, yearning, despair, and loss-related anxiety are significantly higher during the months of the late spouse’s birthday, June, and January, relative to other months of the year. Second, interviews during the month of December, a time of major family centered holidays in Judeo-Christian societies, did not evidence significantly elevated levels of distress. However, this may reflect the fact that interviews were purposely not conducted in the week surrounding Christmas and New Year’s. Further, the December holidays are typically spent with family members, who may provide emotional support which protects against symptoms of distress. Third, married persons interviewed within 1 month of their spouse’s birthday reported superior psychological health relative to those interviewed at other points in the year, whereas widowed persons reported elevated symptoms of depression around their spouse’s birthday month, relative to their peers. Finally, these patterns were limited to the first year of loss only; we found no significant associations between interview timing and psychological symptoms at the 18-month interviews. The implications of this study are twofold: special occasion grief reactions are a significant yet time-limited phenomenon and a source of psychological distress among recently bereaved older adults; and special occasions may serve as a methodological confound if not accounted for in bereavement studies.
Overall, we found that special occasion periods are associated with a range of elevated grief and depressive symptoms during the first year of loss. Although the magnitude of these effects varies based on the particular occasion and outcome considered, the impact is substantial—at least in the short-term postloss. For example, the effects of June interviews on yearning (b = .436) and despair (b = .456) were roughly one-half standard deviation. Special events may focus the widowed person’s attention on the death of a loved one, and thus enhance feelings of grief and sadness. Cognitive science suggests that people place disproportionate emphasis where their attention is directed (Loewenstein & Schkade, 1999). Special dates may lead to a re-focusing on earlier painful feelings associated with the death (Rabow, Hauser, & Adams, 2004). Changes in routine and social interactions during these special occasion periods also may affect psychological adjustment. After the excitement and heightened social interaction associated with holidays and celebrations pass, bereaved older adults may return to their regular routines, making their solitude and loss all the more salient.
The specific symptoms evidenced, however, vary across occasion period. For example, the multivariate analyses revealed that symptoms of yearning and despair (but neither depression nor anxiety) were elevated during June interviews. Yearning reflects an intermittent, recurrent, and obtrusive wish or need to recover the person who has died. Despair, similarly, refers to a feeling of emptiness and that a part of one’s self is missing. Attachment theory holds that the essential dimension of grief that distinguishes bereavement from other forms of emotional distress is the act of pining for the lost person (Archer, 1999). Given that June is the most common month of weddings, bereaved older spouses may find themselves longing for their lost relationship on their own wedding anniversary date or when celebrating the wedding anniversaries of their peers, children, and others who still have living spouses. June also is associated with other family centered occasions, including school graduations and Father’s Day in the United States; each of these may be reminders that the deceased spouse is no longer alive to participate in such meaningful celebrations.
January interviews, by contrast, were associated with elevated symptoms for all but one of the psychological outcomes; symptoms of yearning, despair, loss-related anxiety, and depressive symptoms were higher in January. We did not find evidence of a statistically significant difference between the January depressive symptoms of widowed persons and matched controls although this nonsignificant interaction term could reflect small sample size. Interviews conducted in January clearly elicit distress among the bereaved. Elderly widowed persons may be especially vulnerable to post-holiday “blues” (Cook, 1983). Bereaved spouses may feel particularly lonely and socially isolated after the excitement of family holidays has passed, and family members return to their daily routines.
Finally, persons who completed an interview in the month surrounding the birthday of their deceased spouse reported elevated levels of despair, anxiety, and depressive symptoms relative to persons interviewed at other times. Importantly, we found that currently married persons evidenced fewer depressive symptoms when interviewed near their spouse’s birthday, whereas bereaved spouses reported significantly more symptoms, relative to their counterparts interviewed at other times. For older married adults, a spouse’s birthday celebration may bring a sense of happiness and relief that another year has passed and the couple is still together. Further, the birthdays of older adults may prompt particularly genial celebrations by family members, who recognize that the number of future birthday celebrations may be limited. By contrast, a deceased spouse’s birthday may pass unacknowledged by others, provoking a sense of sadness and emptiness among the bereaved survivor.
Implications for Gerontological Practice and Research
Our findings have potentially important implications for gerontological practice. Researchers and clinicians alike have recommended that there be greater outreach and support to the bereaved during potentially emotionally charged times such as anniversaries or holidays (Rabow et al., 2004; Sanders, 1999). For example, Jordan (2003: p. 111) recommended that “anniversary effects need to be addressed preventively, not reactively.” However, these recommendations have been based on anecdotal evidence and small numbers of case studies (Earls & Wolf, 1963; Hertz, 2002; Musaph, 1990). Our results underscore that caregivers and health care professionals should be aware of the distress and sadness experienced during these potentially evocative times, and should develop policies and practices to ameliorate such impacts on the bereaved. Therapists have recommended that the bereaved formally acknowledge special occasions to help lessen the associated sorrow (Sanders, 1999). Likewise, religious traditions such as yahrtzeit in Judaism (i.e., a candle is lit and prayers are said for the decedent) are an annual commemoration of the deaths of loved ones. These practices may be particularly effective during the early stages of loss, when special occasion reactions are most acute.
Empirical studies of bereavement suggest that such practices or interventions, especially during the first year of loss, may be particularly protective for bereaved older adults. For example, Ha (2010) found that positive support from children in the first 6 months of widowhood was associated with reduced depressive symptoms at 18 months. Other forms of social support, especially interventions promoting family conversations and positive commemorations of the deceased loved one’s life, could prove effective. Given that the effects documented here are relatively short lived, we believe that special occasion grief is a normal response to a very recent death. As such, effective interventions may simply provide the bereaved with an opportunity to discuss their memories of their loved one, or to hold a private celebration of the event that may have gone otherwise unnoticed.
Our results also carry potentially important implications for research: The special occasion grief phenomenon, at least in the first year postloss, introduces a potential measurement confound in bereavement studies. Distress symptoms may vary widely depending on the time of assessment. Widowed persons interviewed near special occasions may overstate the feelings of distress that they have experienced in the prior week, if the special event focuses their cognitions on the loss (Loewenstein & Schkade, 1999). Likewise, the tendency of survey researchers to cease conducting interviews during the winter holiday season, as not to inconvenience potential respondents, may inadvertently lead to an underestimation of grief symptoms during these potentially distressing time periods. Future research should incorporate special dates into assessment protocols. Finally, our results may have implications for the ethical conduct of bereavement research (Cook, 2001). If certain time periods elicit strong emotional reactions from older widows and widowers, then researchers may need to identify dates of special significance to the bereaved, and acknowledge that these periods may be a sensitive time for the respondent to be interviewed.
Limitations and Future Directions
Despite these strengths, our study has several methodological and measurement limitations. First, we lack data on the actual month of marriage. June is a reasonable proxy, given that it was the most common month for weddings in the mid-20th century in the United States. However, future studies should explore the direct effects of wedding anniversary month on bereavement outcomes. Second, the effects of “holiday” grief may be specific to the holidays celebrated in one’s faith. For instance, grief symptoms may be particularly acute among Christians during Christmas season, and among Jews during the High Holidays in autumn. Third, the seasonal effects documented here may vary based on the quality of the marriage, especially for wedding anniversary and birthday months. For those with particularly close and loving marriages, wedding anniversary and spousal birthday dates may be particularly difficult days. Fourth, the impact of special occasions and seasons may be conditional upon the nature of the death; some studies suggest that anniversary grief is particularly acute when the death was sudden or traumatic (Jordan, 2003). However, small sample size precluded us from evaluating two-way interactions between date of interview and potential moderators including religious denomination, marital quality, or cause of death.
Finally, we focused on grief symptoms at only two time points within the first 18 months postloss. Emerging literature on “continuing bonds” suggests that in the longer term, bereaved spouses might evidence enhanced psychological well-being on special occasions. As years elapse since the loss and symptoms of sadness fade, positive memories such as a loved one’s birthday, or past celebrations of wedding anniversaries might provide comfort to the bereaved (Carnelley, Wortman, Bolger, & Burke, 2006; Stroebe, Schut, & Boerner, 2006). We look forward to bereavement scholars delving into these issues more fully, in larger data sets.
Funding
The Changing Lives of Older Couples study has its principal support from National Institute on Aging grants AG15948-01 (R. M. Nesse, Principal Investigator), AG610757-01 (C. B. Wortman, Principal Investigator), and AG05561-01 (J. S. House, Principal Investigator).
Acknowledgments
A public use version of the data is available from the Inter-University Consortium for Political and Social Research (ICPSR) at the University of Michigan or via the Web site at http://www.cloc.isr.umich.edu. D. Carr and J. Sonnega co-planned the study, conducted parts of the data analysis, and co-wrote the article. J. S. House and R. M. Nesse co-wrote the article.
References
Author notes
Decision Editor: Gary Lee, PhD
