impact of offspring death on cognitive health in late life: the cache county study

Impact of Offspring Death on CognitiveHealth in Late Life: The Cache

County Study

Daylee Greene, Ph.D., JoAnn T. Tschanz, Ph.D., Ken R. Smith, Ph.D.,Truls Østbye, M.D., Ph.D., Chris Corcoran, Sc.D., Kathleen A. Welsh-Bohmer, Ph.D.,

Maria C. Norton, Ph.D. for the Cache County Investigators

Received July 30(DG, MCN), Depfor EpidemiologHuntsman CancBryan AlzheimeSingapore. PreseSend correspondUniversity, 2905

� 2013 Amehttp://dx.d

Am J Geriatr Ps

Objective: Experiencing the death of a child is associated with negative short-term

mental health consequences, but less is known about cognitive outcomes and

whether such associations extend to late life. We tested the hypothesis that expe-

riencing an offspring death (OD) is associated with an increased rate of cognitive

decline in late life. Methods: This population-based longitudinal study observed

four cognitive statuses spaced 3e4 years apart, linked to an extensive database

containing objective genealogic and vital statistics data. Home visits were con-

ducted with 3,174 residents of a rural county in northern Utah, initially without

dementia, aged 65e105. Cognitive status was measured with the Modified Mini-

Mental State Exam at baseline and at 3-, 7-, and 10-year follow-ups. OD was ob-

tained from the Utah Population Database, which contains statewide birth and

death records. Results: In linear mixed models, controlling for age, gender,

education, and apolipoprotein E status, subjects who experienced OD while

younger than age 31 years experienced a significantly faster rate of cognitive

decline in late life, but only if they had an ε4 allele. Reclassifying all OD (regardless

of age) according to subsequent birth of another child, OD was only related to faster

cognitive decline when there were no subsequent births. Conclusion: ExperiencingOD in early adulthood has a long-term association with cognitive functioning in

late life, with a geneeenvironment interaction at the apolipoprotein E locus.

Subsequent birth of another child attenuates this association. (Am J GeriatrPsychiatry 2013; -:-e-)

Key Words: Cognitive decline, psychosocial stress, apoE

, 2012; revised April 21, 2013; accepted May 9, 2013. From the Department of Family Consumer and Human Developmentartment of Psychology, Center for Epidemiologic Studies (JTT, MCN), and Department of Mathematics and Statistics, Centeric Studies (CC), Utah State University, Logan, UT; the Department of Family and Consumer Studies, Population Sciences,er Institute (KRS), University of Utah, Salt Lake City, UT; Duke Global Health Institute (TØ) and The Joseph and Kathleenr’s Disease Research Center (KAW-B), Duke University, Durham, NC; and DukeeNUS Graduate Medical School (TØ),nted in part at the 63rd annual conference of the Gerontological Society of America, New Orleans, November 19-23, 2010.ence and reprint request to Dr. Maria C. Norton, Department of Family Consumer and Human Development, Utah StateOld Main Hill, Logan, UT 84322-2905. e-mail: [email protected] Association for Geriatric Psychiatryoi.org/10.1016/j.jagp.2013.05.002

ychiatry -:-, - 2013 1

mailto:[email protected]

http://dx.doi.org/10.1016/j.jagp.2013.05.002

2

Impact of Offspring Death on Dementia Risk in Late Life

“There’s no tragedy in life like the death of a child. Thingsnever get back to the way they were.”

—Dwight D. Eisenhower

INTRODUCTION

The death of any family member is difficult, butlosing a child may be the most traumatic, because it isan “off-timing” event that violates the natural orderof life. Further, it is often not the number of stressfulevents but rather the timing of these events thatdetermines the amount of stress individuals experi-ence.1 Experiencing the death of a child may leadparents to have lasting psychological distress,2 espe-cially because it is associated with longer and moreintense periods of grief than the loss of a spouse orparent.3 Some parents experience sadness not onlyfor the loss of their child but also feel as though a partof themselves has died with their child.4

Bereavement-related stress after offspring death(OD) may be compounded by concomitant experi-ences. Nearly half of parents who experience the deathof a child report strained marital relationships, lack ofcommunication between spouses, or distancing in therelationship, often resulting in divorce.5 This addedstress may result in negative lifestyle behaviors, alsoraising the risk of adverse health outcomes andmortality.6,7 Loss of a child may increase the risk ofseveral negative health outcomes, such as myocardialinfarction,8 diabetes,9 and multiple sclerosis.10 Thus,experiencing the death of a child disrupts parentalhealth and well-being, with effects that may be man-ifested short and long term over the course of parents’lives.11

Less well studied is the potential impact of thedeath of a child or other close family member on late-life dementia risk. Death of a spouse has been asso-ciated with more rapid cognitive decline in later life12

and death of a father during childhood with higherdementia risk,13 with a net of adjustment for socio-economic status. Because most bereaved parents feelas though their grief is a continual process that doesnot have a definitive end,7 they experience stressfrom the loss of their child for a number of years aftertheir child’s death.

A possible explanation for the association betweenchronic psychological stress and dementia risk derivesfrom observations showing that chronic exposure to

psychological stress leads to a sustained increase instress hormone production, which in turn can produceexcess free radicals that may damage tissues andorgans, including the brain.14 The RCAN1 gene isthought to play a role in the development of tau-pathies such as Alzheimer disease (AD), andRCAN1 proteins can be induced by psychosocial andemotional stress.15 In addition, aged individuals havean impaired ability to discontinue the production ofstress hormones (glucocorticoids); this may paradox-ically lead to the so-called glucocorticoid cascade cyclethat can have serious neurologic consequences.16

The objective of the current study was to assess theimpact of a child’s death on cognitive decline in latelife, hypothesizing that such adversity would beassociated with chronic psychological stress andresult in faster cognitive decline. This study useda large population-based sample of older adults,using objective records to define exposure to OD,thereby eliminating recall bias. Cognitive functioninggenerally declines over time as a natural function ofage, and animal studies also suggest that neuronalresilience to stress may be lost with age, given thatthe brains of older rats self-repaired less readily afterstress than younger rats.17 These observationssuggest that the older the individual is when expe-riencing the death of a child, the greater the impacton cognitive health. However, young adults aremuch more likely than older adults to use copingstrategies that suggest lower levels of impulse controland self-awareness, which are indicative of theirlower levels of maturity.18 Younger individuals maytherefore be less successful at coping with the trau-matic event of losing a child. Although someevidence suggests that age of a child at death caninfluence the grieving process for parents,19 thefindings are mixed regarding the link between ODand parental outcomes when using child’s age atdeath as an independent variable.20 The currentstudy therefore examined whether associationsbetween OD and cognitive decline varied by timingof OD and age of the offspring at death. Furthermore,because women are 1.5e2 times as likely to experi-ence depression (potential marker of psychologicalstress) than men,21 we hypothesized that this asso-ciation would be stronger among women. Addition-ally, because apolipoprotein E (APOE) genotypehas been shown to moderate associations betweenother psychosocial stressors and dementia risk, for

Am J Geriatr Psychiatry -:-, - 2013

Greene et al.

example, higher dementia risk among APOE ε4carriers who also experienced low childhood socio-economic status22 or a large family size,23 wehypothesized that individuals who were ε4 positivewould have a stronger association.

METHODS

Study Design

The Cache County Memory Study (CCMS) isa prospective, population-based, epidemiologicstudy of dementia. Four waves of ascertainment wereconducted between 1995 and 2008 (baseline andfollow-up interviews at 3, 7, and 10 years). Thepresent study used data on cognitive status (acrossall four study waves) and linkage to an objective datasource to define OD exposure.

Subjects

The original CCMS cohort included 5,092 partici-pants (90% of county residents aged 65 years or olderin January 1995). We excluded 359 participants withprevalent dementia24 (see Breitner et al.24 for detaileddescription of dementia ascertainment protocol),leaving 4,733. Thirty confirmed nulliparous subjectswere excluded, leaving 1,529 with missing (1,049with no offspring records and 480 with offspringnoted on genealogy records but lacking all dates) and3,174 with nonmissing offspring data.

Cognitive Decline Measurement

The Modified Mini-Mental State Exam25,26 wasused to assess cognitive status at baseline and each ofthe three follow-up assessments. The Modified Mini-Mental State Exam is a 100-point adaptation of theMini-Mental State Exam27 and assesses attention,concentration, orientation to time and place, memory,constructional praxis, and expressive and receptivelanguage. It was administered by interviewerstrained by a study neuropsychologist as to stan-dardized administration and scoring.

OD Assessment

OD was operationalized objectively throughlinkage between the Utah Population Database(UPDB) and CCMS records. The UPDB compiles


demographic, genetic, and epidemiologic informationon Utah residents28 and contains information on 7million individuals, representing over 14 millionrecords. UPDB records are already linkedwith CCMS;5,091 of 5,092 CCMS subjects appear in the UPDB.

The UPDB provided information about the numberof children born to each participant as well as thedate of each child’s birth and death. ODs were codedalong two dimensions: age of subject (i.e., the parent)at OD and age of offspring at OD. OD dates wereignored if occurring after subject’s death date ordementia onset (if diagnosed with dementia). Sepa-rate variables were derived, with age ranges chosento approximate young/emerging adulthood (ages17e30 years), younger middle age (ages 31e44years), and later middle age (ages 45e64 years).These were each coded trichotomously as OD duringage 17e30 years versus OD during some other stageof life versus no OD experience (similarly for theother two life stages). We further considered exami-nation of the manner of death (accidental, homicide,or suicide versus natural), but 82% of the cohortmissed data on this variable, precluding its analysis.

We defined childbearing years according to themaximum gender-specific parental age at child birth(49 years for women and 60 years for men) and thencoded OD without replacement (during childbearingyears), OD without replacement (after childbearingyears), OD with replacement birth, and no OD. Inexploratory analyses, we hypothesized that ODwithout the subsequent birth of another child (duringchildbearing years) would be associated with fastercognitive decline than OD with a “replacement child”birth.

Covariates

APOE genotype was determined from buccal DNAusing polymerase chain reaction amplification anda restriction isotyping method described by Saunderset al.29 Subjects were coded as carriers or noncarriersof at least one 34 allele. Age at baseline interview andeducational attainment were measured in years.Total number of children born to each subject wasderived through birth records in the UPDB.

Statistical Analyses

Descriptive analyses. One-way analyses of vari-ance were computed on continuous variables and

3


c2 tests were computed on categorical variables todetermine bivariate associations between key cova-riates and the experience of OD to explore possibleconfounding. To assess potential nonresponse bias,participants in the final sample were compared withthose who were removed due to missing offspringdata, using t tests for continuous variables and c2

tests for categorical variables.Rate of cognitive decline. General linear models

were used that accommodate repeated measures,incorporating a quadratic (nonlinear) term for time.This approach models the association between fixedfactors (OD) and trajectory of an outcome (cognition)while accounting for within-subject correlation incognitive scores across time. Models were parame-terized to allow examination of differences in base-line cognitive score and differences in rate of declineas a function of OD exposure. When a significantinteraction between OD and time was observed,additional models examined moderation of thisassociation by APOE and gender. If the interactionbetween OD � APOE � time was significant atp <0.10, we conducted stratified models by APOEgenotype (similarly for gender interaction). Ina separate model, the “OD with/without replace-ment” exposure variable was tested for associationwith cognitive decline.

We first studied OD exposure before the baselineinterview, with separate variables to focus on eachlife stage. Then we studied OD exposure after base-line (among unexposed before baseline), coded 0 asof baseline, and then set to 1 at the first cognitivemeasurement after the incident OD (and thereafter).

Statistical analyses were conducted using SPSSversion 21 software (IBM, Armonk, NY). In all models,the difference in the log likelihood between nestedmodels was tested; only if this test was significant wasthemore complexmodel retained.Allmodels includedage, gender, education, andAPOE status as covariates.

RESULTS

Bivariate Associations Between OD andDemographic Variables

In the final analysis sample of 3,174 subjects, 68.0%were women, 99% were white, and 30.4% had at leastone Ɛ4 allele at APOE. Subjects had a mean age of

4

74.8 years (standard deviation [SD]: 7.0) age anda mean education level of 13.1 years (SD: 2.7). In thesample, 420 persons (13.2%) developed incidentdementia. The number of children born ranged from1 to 14 (mean: 3.42, SD: 2.10). OD occurred in 209subjects during ages 17e30 years, in 175 during ages31e44 years, and in 122 during ages 45e64 years(overlap consisted of 19 during the first two intervals,7 during the second and third intervals, 7 during thefirst and third intervals, and none with OD in allthree intervals). Another 127 subjects experienced ODonly after baseline, and 2,481 were unexposed to OD.

Participants with missing offspring data were morelikely to be men (65.3%) compared with the finalsample (32%; c2 ¼ 474, df ¼ 1, p <0.001) and werealso more highly educated (mean: 13.43, SD: 3.3) thanthe final sample (mean: 13.1, SD: 2.7; t ¼ 4.09, df ¼4719, p <0.001). Subjects missing offspring data were,on average, older (mean: 76.3 years, SD: 7.1) thansubjects with nonmissing data (mean: 74.8 years, SD:7.0; t ¼ 6.99, df ¼ 4731, p <0.001). The two groupswere not significantly different on APOE status(Fisher’s exact test c2 ¼ 0.10, df ¼ 1, p ¼ 0.756;Table 1).

In bivariate tests of association between socio-demographic variables and whether or not a subjecthad ever experienced OD (removing the missing ODgroup), APOE was not significant (c2 ¼ 0.158, df ¼ 1,p ¼ 0.691; Table 1). There was a significant associa-tion between gender and OD: 24.0% of women and17.1% of men experienced OD (c2 ¼ 20.2, df ¼ 1,p <0.001). Education was significantly associatedwith OD (t ¼ e4.783, df ¼ 1141, p <0.001), withparticipants who had experienced an OD havinglower average education (mean: 12.7 years, SD: 2.6)than those without OD exposure (mean: 13.2 years,SD: 2.7). Age was significantly associated with OD(t ¼ 7.22, df ¼ 989, p <0.001), with the average age ofOD exposed participants (mean: 76.6 years, SD: 7.8)lower than average age of unexposed participants(mean: 74.3 years, SD: 6.7).

Subject (Parent) Age at OD

Experiencing OD while the subject was under age31 years was significantly related to faster cognitivedecline, and presence of the Ɛ4 allele at APOEmoderated this association. Results are displayed inTable 2 showing statistical significance for the


TABLE 1. Demographic Summary of Participants by Prevalent OD

Final Sample Excluded from Final Sample

No OD (N [ 2,564)(58%)

1D OD (N [ 610)(14%) pa

Missing Offspring Information(N [ 1,196) (27%) pa

Education, y (mean, SD) 13.2, 2.7 12.7, 2.6 <0.001 13.7, 3.4 <0.001Baseline age, y (mean, SD) 74.3, 6.7 76.6, 7.8 <0.001 75.1, 6.5 0.220Total number of offspring (mean, SD) 3.2, 2.0 4.5, 2.2 <0.001 — <0.001Gender: male, N (%) 879 (45.9) 136 (7.1) <0.001 875 (73.2) <0.001APOE ε4 carrier, N (%) 777 (57.6) 187 (13.9) 0.866 372 (27.6) 0.641

ap Values are from independent group t tests for comparisons of age, education, and number of offspring; p values are from c2 tests forcomparisons of gender and APOE status. Groups compared in the first column of p values are participants with no OD versus those with oneor more OD (all had at least one offspring born) to address potential confounding; Groups compared in the second column of p values areparticipants in the final sample (combining no OD and 1þ OD groups) versus those excluded from final sample due to missing offspringinformation to address potential nonresponse bias.

Greene et al.

omnibus test and single-df interaction terms betweenOD and time and time2 demonstrating associationbetween OD and rate and acceleration of cognitivedecline, respectively. Additionally (not shown), therewere nonsignificant differences in baseline cognitivescores between exposure groups (p values rangedfrom 0.543 to 0.586 across all models for omnibus

TABLE 2. Linear Mixed Models of Modified Mini-Mental State Exam oof OD Experience (N [ 3,157)

Model No. EffectP

1: OD during subject age of 17e30 years: F(2,1456) ¼ 3.41, p ¼ 0.033 (liOD at subject age other than 17e30 yearsOD at subject age 17e30 years



4: OD during subject age of 17e30 years: OD � APOE F(2,1540) ¼ 2.93,

5: (ε4 negative) subject age of 17e30 years: F(2,1073) ¼ 0.61, p ¼ 0.546

6: (ε4 positive) subject age of 17e30 years: F(2,420) ¼ 5.15, p ¼ 0.006 (lOD at subject age other than 17e30 yearsOD at subject age 17e30 years

7: OD during subject age of 17e30 years: OD � gender F(2,1422) ¼ 0.04

8: OD with/without “replacement child” F(3,1561) ¼ 3.11, p ¼ 0.025 (linOD without replacement (after childbearing years)OD with replacementOD without replacement (during childbearing years)

9: OD with/without “replacement child”: OD � APOE F(3,1582) ¼ 0.28,

10: OD with/without “replacement child”: OD � gender F(3,1570) ¼ 0.92,

11: Post-baseline ODb: F(1,1073) ¼ 2.31, p ¼ 0.129 (linear); F(1,896) ¼ 3.1Incident OD

aType III F tests are for omnibus test of overall effect of the given OD vafreedom parameter estimates in the first/second column indicate how macceleration of cognitive decline is, on average, for subjects experiencing

bAmong subjects without OD before baseline, the number of subjects exand 70 subjects occurring in the first, second, and third postbaseline inte


F test of OD main effect in the various models). Inseparate models, subjects with no Ɛ4 allele had nosignificant association with OD, whereas subjectswith one or more copies of the Ɛ4 allele had signifi-cant linear and quadratic association with OD, witha 1.24 points per year faster decline than those withno OD exposure.

ver 10 Years of Observation, as a Function of Various Measures

arameter Estimatea (p value)OD 3 Time

Parameter Estimate (p value)OD 3 Time2

near); F(2,1137) ¼ 3.84, p ¼ 0.022 (quadratic)0.13 (p ¼ 0.433) �0.03 (p ¼ 0.056)

�0.50 (p ¼ 0.016) 0.04 (p ¼ 0.067)

near); F(2,1118) ¼ 0.09, p ¼ 0.912 (quadratic)�0.02 (p ¼ 0.938) �0.004 (p ¼ 0.881)�0.14 (p ¼ 0.365) �0.007 (p ¼ 0.679)

near); F(2,1147) ¼ 0.09, p ¼ 0.919 (quadratic)�0.23 (p ¼ 0.368) �0.001 (p ¼ 0.956)�0.07 (p ¼ 0.634) �0.01 (p ¼ 0.681)

p ¼ 0.054 (linear); F(2,1219) ¼ 3.40, p ¼ 0.034 (quadratic)

(linear); F(2,829) ¼ 1.08, p ¼ 0.342 (quadratic)

inear); F(2,331) ¼ 4.05, p ¼ 0.018 (quadratic)0.24 (p ¼ 0.393) �0.04 (p ¼ 0.252)

�1.24 (p ¼ 0.003) 0.12 (p ¼ 0.012)

, p ¼ 0.961 (linear); F(2,1094) ¼ 0.24, p ¼ 0.784 (quadratic)

ear); F(3,1242) ¼ 1.59, p ¼ 0.191 (quadratic)0.15 (p ¼ 0.289) �0.014 (p ¼ 0.315)0.36 (p ¼ 0.309) �0.04 (p ¼ 0.289)

�0.77 (p ¼ 0.009) 0.05 (p ¼ 0.118)



5, p ¼ 0.076 (quadratic); N ¼ 2,594�1.38 (p ¼ 0.129) .117 (p ¼ 0.076)

riable on rate and acceleration of cognitive decline. Single degree ofuch slower (positive values) or faster (negative values) the rate/the given OD stressor.periencing OD after baseline was as follows: 73 subjects, 73 subjects,rval, respectively.

5

FIGURE 2. Cognitive trajectory of the Modified Mini-Mental


Figure 1 depicts the different cognitive trajectoriesas a function of OD exposure and APOE status,where effect sizes can be discerned at various pointsalong the time axis by visual comparison of “OD”

and “no OD” curves, separately for “Ɛ4 positive” and“Ɛ4 negative” subgroups. Gender did not moderatethis association. When the total number of childrenborn to a subject was included as a covariate (notshown), the model results were virtually unchanged.

To explore the impact of having excluded subjectswith missing offspring data, we conducted additionalmodels (not shown) adding these subjects back intothe sample as another exposure group. This ambig-uous group did not differ on rate of cognitive declinecompared with subjects with no OD experience (B ¼0.0006, t ¼ 0.063, df ¼ 1,947, p ¼ 0.950), but results forthe group with OD exposure during 17e30 years(B ¼ e0.51, t ¼ e2.47, df ¼ 2007, p ¼ 0.014) werenearly identical to analyses without inclusion of themissing group.

Experiencing OD when subjects were ages 31e44years or 45e64 years was not significantly related torate of cognitive decline. (Additional analyses wereconducted examining association between OD andcognitive decline, according to offspring age at death;these variables were highly correlated with ODaccording to subject’s age at time of OD and aretherefore not included in this report, e.g., c2 ¼ 1,574,df ¼ 4, p <0.001).

In the model examining OD occurrence withversus without subsequent birth of a “replacementchild,” subjects experiencing OD (during childbearingyears) without replacement (N ¼ 99) had an average

FIGURE 1. Cognitive trajectory of the Modified Mini-MentalState Exam (3MS) for persons experiencingoffspring death under the age of 31 years versus atsome later age versus not at all, stratified bypresence versus absence of ε4 allele at APOE.

6

0.77 points per year faster cognitive decline thanthose without OD exposure (N ¼ 2,572), consistentwith our hypothesis. OD with replacement (N ¼ 77)and OD (after childbearing years) without replace-ment (N ¼ 427) were not significantly related to rateof cognitive decline. Figure 2 depicts the differentcognitive trajectories as a function of OD exposurewhen categorized into these four groups, clearlyshowing that subjects with “OD (during childbearingyears) without replacement” experienced more rapidcognitive decline.

Time-Varying Exposure to OD

A separate model tested the association betweenOD exposure after baseline as a time-varying covar-iate and rate of cognitive decline, including onlythose subjects who had not been exposed to ODbefore baseline. Individuals exposed to OD afterbaseline exhibited a trend for faster decline, with anaverage 1.38 points per year faster rate of declinethan those without OD exposure after baseline.

DISCUSSION

The key finding of this investigation is that ODduring young parental age (<31 years) predictsa faster rate of cognitive decline in late life, net ofseveral known predictors. This is consistent withother studies that describe shorter-term effects of

State Exam (3MS) as a function of offspring deathexperience (with or without subsequent birth ofa “replacement child”).


Greene et al.

parental bereavement,2,7,9,11 with the present studyextending this to include cognition in late life. Theresults are also in line with previous studies that havedemonstrated longer-term associations betweenother “loss event” stressors such as paternal deathduring ages 0e4 years,30 maternal death duringadolescence,14 and the death of a spouse13 and withlate-life cognitive health.

Finding that OD was only associated with fastercognitive decline if experienced during the teens andtwenties suggests that individuals in earlier devel-opmental stages of life may have fewer resources(financial, social, emotional) to help them deal withthe stress of an OD. Because the brain continues todevelop throughout adolescence and early adult-hood, individuals in this stage may have an increasedvulnerability to stress-related physiologic processesthat lead to negative consequences in the brain.31

Also, young adults display a tendency to chooseless helpful coping mechanisms compared with olderadults.19 This increased vulnerability may thereforeexacerbate the negative physiologic effect of stressand raise the risk for negative cognitive outcomes inlate life. Experiencing adversities in the earlier stagesof life may also prevent the brain from reaching fullmaturity, placing these individuals at higher risk ofAD in late life.32 It has also been shown that early lifeadversity can alter hypothalamo-pituitary-adrenalaxis (HPA) stress responses, leading to negativeoutcomes on brain functioning later in life.33 TheHPA also plays a role in cognition in late life; mal-functions in the HPA are observed in the early stagesof AD and result in increased glucocorticoid levels,34

which can lead to hippocampal damage as well aslearning and memory deficits.35

Importantly, the association between early-life ODand cognitive decline was significant only amongpersons with at least one Ɛ4 allele at APOE. Somestudies,24 but not all,33 have similarly found thatAPOE is a strong moderator of early-life adversity onAD risk, suggesting that heightened genetic vulner-ability for AD may define a subpopulation that ismore vulnerable to the effects of early-life psycho-logical stress. Neurodegenerative processes maybegin decades before the onset of dementia, and thecombination of OD as a stressor in the presence ofgenetic risk may accelerate this process. Support forthis assertion can be found in research concludingthat prolonged exposure to stress results in memory


decline in older, nondemented individuals who havean ε4 allele.36

We also found that only those individuals experi-encing OD without the subsequent birth of anotherchild had significantly faster cognitive decline. Thebirth of a “replacement child” may have provided thegrieving parent with a new target for their parentalaffection, thusbuffering the effects of their earlier loss.37

A limitation to the current study is the highpercentage of missing offspring data, particularly formen, suggesting that our findings may be somewhatmore generalizable to women. Even so, our principalfinding of faster late-life cognitive decline amongindividuals experiencing OD in young adult life wasrobust when analyses included those with missingoffspring data in a separate category. To the extentthat this “missing data” subgroup contained indi-viduals who had experienced OD, our results wouldbe a conservative estimate of effects. Another limi-tation is ethnic homogeneity, because the studysample was composed almost exclusively of whiteindividuals. A strength is that the study used a pop-ulation-based sample, with results that are morerepresentative of the community than a clinicalsample alone, such as one comprised of persons whoseek medical evaluation or intervention in a memorydisorders clinic. Another strength is the long periodof follow-up, spanning up to 12 years. Finally, weused objective data to measure exposure to OD ratherthan reliance on retrospective self-reports, known tohave problems with recall bias.

In summary, this study provides evidence extend-ing the adverse physical and mental health outcomesassociated with parental bereavement to includefaster cognitive decline in late life, a strong risk factorfor eventual dementia. Additional research is war-ranted to examine whether the mechanism(s) is viachronic exposure to stress-related hormones, height-ened risk for depression, less healthy lifestyle choicesin bereaved parents, or other factors. Moreover, situ-ational, intrapersonal, and interpersonal factors, aswell as coping styles, interact to determine the ulti-mate outcome of a bereavement experience.38 Addi-tionally, each of these children died for a reason,whether after a long illness, a sudden traumatic event,or other circumstance. The OD experience may alsohave coincided with other chronic stressors at the timeof OD, possibly with weak social support system orother adverse contextual factors. Thus, awide range of

7


circumstances surrounding the death may explain thecognitive decline of the participant in later life. Thesefactors likely interact in complex ways to impactbereavement outcomes. Further research should focuson manner of death, the social support network of theparent, and other circumstances surrounding thedeath of the child. Ultimately, such studies mayinform interventions targeted at bereaved parents andparents of terminally ill children to provide appro-priate support to help buffer the stress associated withthe death of a child.

This research was supported by the National Insti-tutes of Health grants R01-AG031272, R01-AG011380,

8

and AG-022095. Partial support for all data sets withinthe UPDB was provided by the University of UtahHuntsman Cancer Institute and the Huntsman CancerInstitute Cancer Center support grant (P30 CA42014)from the National Cancer Institute.

We acknowledge the contributions of the followingindividuals whose activities have helped to ensure thesuccess of the project: Cara Brewer, B.A., John C.S.Breitner, M.D., M.P.H., Tony Calvert, B.S., Carol Leslie,M.S., Michelle McCart, Ronald G. Munger, Ph.D.,M.P.H., Roxane Pfister, M.S., Georgiann Sanborn, M.S.,Nancy Sassano, Ph.D., Sarah Schwartz, M.S., MartinToohill, Ph.D., Heidi Wengreen, Ph.D., R.D., JamesWyatt, and Peter P. Zandi, Ph.D., M.P.H.

References

1. Goodhart D, Zautra A: Off-time events and life quality of olderadults. Paper presented at the annual meeting of the WesternPsychological Association, San Diego, CA, April 5-8, 1979.

2. McCarthy MC, Clarke NE, Ting CL, et al: Prevalence and predic-tors of parental grief and depression after the death of a childfrom cancer. J Palliat Med 2010; 13:1321e1326

3. Middleton W, Raphael B, Burnett P, et al: A longitudinal studycomparing bereavement phenomena in recently bereaved spou-ses, adult children and parents. Aust NZ J Psychiatry 1998; 32:235e241

4. Malkinson R, Bar-Tur L: Long term bereavement processes ofolder parents: the three phases of grief. OMEGA J Death Dying2005; 50:103e129

5. Arnold J, Gemma PB, Cushman PF: Exploring parental grief:combining quantitative and qualitative measures. Arch PsychiatrNurs 2005; 19:245e255

6. Vassalis CA, Morgan HG: Suicide in Avon. Life stress, alcoholmisuse, and use of services. Br J Psychiatry 1997; 170:453e455

7. Li J, Precht DH, Mortensen PB, et al: Mortality of parents afterdeath of a child in Denmark: a nationwide follow-up study.Lancet 2003; 361:364e367

8. Li J, Hansen D, Mortensen PB, et al: Myocardial infarction inparents who lost a child: a nationwide prospective cohort studyin Denmark. J Am Heart Assoc 2002; 106:1634e1639

9. Olsen J, Li J, Precht DH: Hospitalization because of diabetes andbereavement: a national cohort study of parents who lost a child.Diabetic Med 2005; 22:1338e1342

10. Li J, Johansen C, Bronnum-Hansen H, et al: The risk of multiplesclerosis in bereaved parents: a nationwide cohort study inDenmark. Neurology 2004; 62:726e729

11. Song J, Floyd FJ, Seltzer MM, et al: Long-term effects of childdeath on parents’ health-related quality of life: a dyadic analysis.Interdisc J Appl Fam Stud 2010; 59:269e282

12. Rosnick CB, Small BJ, Burton AM: The effect of spousalbereavement on cognitive functioning in a sample of older adults.Neuropsychol Dev Cogn B Aging Neuropsychol Cogn 2010; 17:257e269

13. Norton MC, Smith KR, Østbye T, et al: Early parental death andremarriage of widowed parents as risk factors for Alzheimer’sdisease. Am J Geriatr Psychiatry 2011; 19:814e824

14. McEwen BS: Sex, stress, and the hippocampus: allostasis, allo-static load, and the aging process. Neurbiol Aging 2002; 23:921e939

15. Ermak G, Pritchard MA, Dronjak S, et al: Do RCAN1 proteins linkchronic stress with neurodegeneration? Fed Am Soc Exp Biol2011; 25:3306e3311

16. Sapolsky RM, Krey LC, McEwen BS: The neuroendocrinology ofstress and aging: the glucocorticoid cascade hypothesis. EndocrRev 1986; 7:284e301

17. Bloss EB, Janssen WG, McEwen BS, et al: Interactive effects ofstress and aging on structural plasticity in the prefrontal cortex.J Neurosci 2010; 30:6726e6731

18. Diehl M, Coyle N, Labouvie-Vief G: Age and sex differences instrategies of coping and defense across the lifespan. PsycholAging 1996; 11:127e139

19. Keesee NJ, Currier JM, Neimeyer R: Predictors of grief followingthe death of one’s child: the contribution of finding meaning.J Clin Psychol 2008; 64:1145e1163

20. Rostilla M, Saarela J, Kawachi I: Mortality in parents following thedeath of a child: a nationwide follow-up study from Sweden.J Epidemiol Commun Health 2012; 66:927e933

21. Sex Differences in Depression. Edited by Nolen-Hoeksema S.Stanford, CA, Stanford University Press, 1990

22. Moceri VM, Kukull WA, Emanuel I, et al: Using census data andbirth certificates to reconstruct the early-life socioeconomicenvironment and the relation to the development of Alzheimer’sdisease. Epidemiology 2001; 12:383e389

23. Borenstein AR, Wu Y, Mortimer JA, et al: Developmental andvascular risk factors for Alzheimer’s disease. Neurobiol Aging2005; 26:325e334

24. Breitner JC, Wyse BW, Anthony JC, et al: APOE-epsilon4 countpredicts age when prevalence of AD increases, then declines: theCache County Study. Neurology 1999; 52:321e331

25. Teng EL, Chui HC: The Modified Mini-Mental State (3MS) exam-ination. J Clin Psychiatry 1987; 48:314e318

26. Tschanz JT, Welsh-Bohmer KA, Plassman BL, et al: An adaptationof the Modified Mini-Mental State Examination: analysis ofdemographic influences and normative data: the cache countystudy. Neuropsychiatry Neuropsychol Behav Neurol 2002; 12:270e276


http://refhub.elsevier.com/S1064-7481(13)00241-8/sref1





















































http://refhub.elsevier.com/S1064-7481(13)00241-8/sref17a
















































Greene et al.

27. Folstein MF, Folstein SE, McHugh PR: Mini-Mental State. A prac-tical method for grading the cognitive state of patients for theclinician. J Psychiatr Res 1975; 12:189e198

28. Smith KR. University of Utah Pedigree and Population ResourceUtah Population Database: Retrieved from http://www.huntsmancancer.org/research/shared-resources/utah-population-database/overview. Accessed July 16, 2013

29. Saunders AM, Strittmatter WJ, Schmechel D, et al: Association ofapolipoprotein E allele Ɛ4 with late-onset familial and sporadicAlzheimer’s disease. Neurology 1993; 43:1467e1472

30. Norton MC, Ostbye T, Smith KR, et al: Early parental death andlate-life dementia risk: findings from the Cache County Study. AgeAgeing 2009; 40:340e343

31. Brown SA, McGue M, Maggs J, et al: A developmental perspective onalcoholandyouths16to20yearsofage.Pediatrics2008;121:S290eS310

32. Moceri VM, Kukull WA, Emanuel I, et al: Early-life risk factors andthe development of Alzheimer’s disease. Neurology 2000; 54:415


33. McGowan PO, Sasaki A, D’Alessio AC, et al: Epigenetic regulationof the glucocorticoid receptor in human brain associates withchildhood abuse. Nat Neurosci 2002; 12:342e348

34. Green KN, Billings LM, Roozendaal B, et al: Glucocorticoidsincrease amyloid-b and tau pathology in a mouse model of Alz-heimer’s disease. J Neurosci 2006; 26:9047e9056

35. Lupien SJ, de Leon M, de Santi S, et al: Cortisol levels duringhuman aging predict hippocampal atrophy and memory deficits.Nat Neurosci 1998; 1:69e73

36. Peavy GM, Lange KL, Salmon DP, et al: The effects of prolongedstress and APOE genotype on memory and cortisol in olderadults. Biol Psychiatry 2007; 62:472e478

37. A Life Course Approach to Chronic Diseases Epidemiology.Edited by Kuh D, Ben-Shlomo Y. New York, Oxford UniversityPress, 2004

38. Stroebe M, Schut H, Stroebe W: Health outcomes of bereave-ment. Lancet 2007; 370:1960e1973

9





http://www.huntsmancancer.org/research/shared-resources/utah-population-database/overview






































impact of offspring death on cognitive health in late life: the cache county study

Documents