Reported Memory Failures

Self-Reported Memory Failures: Associations with Future Dementia in a Population-Based Study with Long-Term Follow-Up Michael R€ onnlund, PhD,* Anna Sundstr€ om, PhD,*† Rolf Adolfsson, MD,‡ and Lars-G€ oran Nilsson, §k PhD

OBJECTIVES: To examine the association between selfreported memory failures and incident dementia in individuals aged 60 and older. DESIGN: Longitudinal, community based. SETTING: Betula Prospective Cohort Study, a population-based study in Ume a, Sweden. PARTICIPANTS: Individuals with a mean age of 71.5  8.8 (range 60–90) (N = 1,547). MEASUREMENTS: Participants rated the frequency of everyday memory failures using the 16-item Prospective and Retrospective Memory Questionnaire (PRMQ) and underwent objective memory testing at baseline. Participant self-reports of complaints of poor memory by family and friends were evaluated. Dementia status was followedup for 10 to 12 years. RESULTS: Over the study period, 225 participants developed dementia (132 with Alzheimer0 s disease (AD)). In Cox proportional hazard regression models adjusted for demographic factors, PRMQ z-scores predicted incident dementia (hazard ratio (HR) = 1.21 for all-cause dementia; HR = 1.25 for AD, Ps < .01). The significant associations remained when depressive symptoms and objective memory performance were adjusted for, when low performers on objective memory (≥1 standard deviations below the age group mean) were excluded, and in analyses with delayed entry (survival time ≥ 5 years). Similar patterns were observed for the prospective and retrospective subscales, although including how often participants selfreported that others complained about their poor memory eliminated the association between PRMQ scores and dementia and itself emerged as a significant predictor. From the *Department of Psychology, Ume a University; †Centre for Demographic and Ageing Research, Ume a University; ‡Division of Psychiatry, Department of Clinical Sciences, Ume a University; §Ume a Center for Functional Brain Imaging, Ume a University, Ume a; and kAging Research Center, Karolinska Institutet, Stockholm, Sweden. Address correspondence to Michael R€ onnlund, Department of Psychology, Ume a University, S-90187 Ume a, Sweden. E-mail: michael.ronnlund@ umu.se DOI: 10.1111/jgs.13611

JAGS 63:1766–1773, 2015 © 2015, Copyright the Authors Journal compilation © 2015, The American Geriatrics Society

CONCLUSION: Self-reported memory failure predicted future dementia or AD independent of objective memory performance. Subjective reports of complaints by family and friends appear to be an even more-important indicator of preclinical impairments, and physicians should not ignore them, even in the absence of objective memory deficits. J Am Geriatr Soc 63:1766–1773, 2015.

Key words: subjective memory; objective memory; dementia; Alzheimer’s disease

E

arly identification of signs of dementia disorders such as Alzheimer’s disease (AD) is important and will likely increase in importance as more-effective medical treatments designed to slow the disease process are developed. A subjective memory assessment could potentially be informative of early signs of impending disease, because memory (episodic memory in particular) is known to be affected in preclinical dementia and AD.1 However, in older adults, subjective memory assessments typically show no or only a small association with objective memory performance according to a recent meta-analysis that found a correlation coefficient of 0.06 across studies.2 Subjective memory impairments have consistently been related to depressive symptoms3–5 and neuroticism.6,7 Based on such patterns of findings, the validity of subjective memory measures as indicators of memory functioning has been questioned. Nevertheless, studies indicate that subjective memory impairment is predictive of incident dementia. A review8 noted that four of six community-based longitudinal studies observed an association with all-cause dementia or AD, and more-recent studies provide further support of this link.9–13 Most of these studies involved a short follow-up period (≤4 years), but a recent study confirmed that subjective assessment made in response to general questions of memory functioning (e.g., “How do you perceive your own memory compared with 5 years ago?”) was predictive of

0002-8614/15/$15.00

JAGS

SEPTEMBER 2015–VOL. 63, NO. 9

incident dementia and AD in a population-based sample with long-term (up to 17 years) follow-up.14 The objective of the present study was to further investigate the relationship between subjective memory and incident dementia (all-cause dementia and AD). This study was based on the same study population15,16 as a previous study14 but considered a different, more-extensive measure of memory functioning, the Prospective and Retrospective Memory Questionnaire (PRMQ),17 which involves a frequency assessment of specific everyday memory failures rather than global judgments of memory functioning. Unlike most subjective memory measures included in prior studies of dementia, the PRMQ has been validated psychometrically.18,19 Thus, to the extent that some subjective measures used in past studies may have had lessthan-ideal measurement properties, such as suboptimal reliability, the PRMQ is as a promising tool for identifying subtle self-experienced deficits in memory functioning. It also allows for a comparison of prospective and retrospective memory failures as preclinical markers of dementia. Prospective memory failure has been considered to be a sensitive, clinically relevant way to evaluate early-stage dementia,20 and a recent study indicated that informant assessment of PRMQ prospective memory failure has predictive value of AD severity beyond that of the retrospective scale.21 A comparison of the prospective and prospective subscales in regard to the prediction of incident dementia was therefore of interest. In addition to potentially confounding variables (e.g., demographic factors, depressive symptoms), a composite measure of episodic memory was included. The purpose was twofold. First, it allowed for a direct comparison of the magnitude of potential associations between dementia and AD and the subjective and objective memory measures. Second, it allowed further examination of the finding that subjective reports may convey information useful to predict incident dementia in addition to concurrent memory performance.14 The subjective assessment that emerged as the most predictive of dementia in a previous study14 (how often someone close to the participant complained of her or his poor memory) was also included to examine whether potential associations between PRMQ scores and incident dementia would persist when this assessment was considered in the analyses.

METHODS Participants Data were drawn from the Betula Prospective Cohort Study,15,16 and was stratified according to age and sex. The sample was collected using randomized sampling from the population register in Ume a municipality, Sweden. Data have been collected on six test occasions; 1988–90 (T1), 1993–95 (T2), 1998–2000 (T3), 2003–05 (T4), 2008–10 (T5), and 2013–14 (T6), with separate samples ranging in age from 25 to 90 at inclusion. On each test occasion, participants were invited to two sessions. The first session mainly concerned assessment of health status, and the second involved tests of cognitive functioning. Participants provided informed consent, and the regional board of the ethics committee in Ume a, Sweden approved the Betula study.

SELF-REPORTED MEMORY FAILURES AND FUTURE DEMENTIA

1767

Participants aged 60, 65, 70, 75, 80, 85, or 90 at baseline were included in the present study. T3 served as the baseline because the PRMQ was first administered on this test occasion. Four subsamples (S1–S4) were involved at T3. Participants in S1 had been assessed on two prior test occasions and those in S2 and S3 on one prior occasion, and participants in S4 took part in the study for the first time. Participants in all samples were followed with regard to dementia status until 2010 (10–12 years after the baseline assessment). One thousand five hundred seventy-nine participants without dementia had taken part in the two sessions devoted to assessment of health and cognitive status at T3. Twenty-seven participants without dementia who died shortly (1 means that poorer memory performance is associated with greater dementia risk). In Model 4, HRs are also reported for selfreport of complaints of poor memory by family and friends.

RESULTS Over the study period, 225 participants developed dementia, including 132 cases of AD; 1,322 participants remained dementia free until loss to follow-up (n = 33; e.g., moved from the catchment area), death (n = 398), or end of study period (n = 891). Mean follow-up time was 8.8 years overall, and mean time to onset of dementia was 4.6 years (4.3 years for AD). The incidence of all-cause dementia, of frequency of loss to follow-up, and of frequency of death (excluding participants in the former two categories) over the course of the study (per year, from Year 1 to Year 12) is presented in Table 1. A majority of the participants who remained dementia free but who were unable to stay in the study until final follow-up due to the fact that they moved from the catchment area or died still contributed substantial information in terms of time at risk. Background characteristics for participants who remained dementia free and those who developed dementia (all-cause dementia and AD) are presented in Table 2. Those who developed dementia were older at baseline, more often female (significant for AD), and more likely to be unmarried (significant for all-cause dementia) and had fewer years of formal schooling, slightly lower MMSE

JAGS

SEPTEMBER 2015–VOL. 63, NO. 9

SELF-REPORTED MEMORY FAILURES AND FUTURE DEMENTIA

1769

Table 1. Frequency Distribution of Participants with Incident Dementia, Participants Lost to Follow-Up, and Participants Who Died Over the Study Period Incident Dementia Year

1 2 3 4 5 6 7 8 9 10 11 12

Lost to Follow-Up

Deceased

n (Cumulative n)

28 25 32 33 27 28 20 14 11 7 0 0

(53) (85) (118) (145) (173) (193) (207) (218) (225) (225) (225)

5 7 3 3 3 2 2 4 2 0 1 1

35 33 38 37 37 38 34 35 35 38 27 11

(12) (15) (18) (21) (23) (25) (29) (31) (31) (32) (33)

(68) (106) (143) (180) (218) (252) (287) (322) (360) (387) (398)

Table 2. Baseline Characteristics of Participants Who Remained Dementia-Free or Developed Dementia (All Types) or Alzheimer’s Disease (AD) Over the Study Period

Characteristic

Age, mean  SD Female, % Married, % Education, years, mean  SD Mini-Mental State Examination29 score, mean  SD Number of depressive symptoms, mean  SD

No Dementia, n = 1,322

All-Cause Dementia, n = 225

Figure 1. Mean Prospective and Retrospective Memory Questionnaire (PRMQ) scores (prospective scale, retrospective scale, total scale) in z-score units as a function of frequency of self-reported complaints of poor memory by family and friends.

AD, n = 132

70.5  8.7 54.5 68.4 9.0  3.6

76.9  6.9c 60.0 56.1a 8.1  3.3a

76.9  6.9c 68.9c 57.1 7.9  3.1c

27.4  1.8

26.3  2.3c

26.2  2.3c

6.7  5.7

7.9  6.0b

8.1  6.5a

SD = standard deviation. P < a.05, b.01, c.001.

scores, and more depressive symptoms than participants who remained dementia free. Cox proportional regression analysis of the association between PRMQ scores and dementia were conducted next. Results for the PRMQ total scale are summarized in Table 2, and the analyses of the subscales (prospective, retrospective) are reported in the text. In the basic model, adjusted for age, sex, marital status, and education (Model 1), the PRMQ score was significantly associated with all-cause dementia (HR = 1.22, 95% CI = 1.07–1.38) and AD (HR = 1.25, 95% CI = 1.06–1.47). Additional adjustment for depressive symptoms (Model 2) had little effect (all-cause dementia: HR = 1.19, 95% CI = 1.05–1.35; AD: HR = 1.22, 95% CI = 1.04–1.44). Adjusting for objective memory score (Model 3) did not alter the results either (HR = 1.18, 95% CI = 1.04–1.34 for all-cause dementia). As a final critical step, participant self-reports of complaints of poor memory by family and friends were added to the model (Model 4). A graphical illustration of mean

Figure 2. Cumulative hazard of all-cause dementia for participants who reported that family or friends often complained of their poor memory and for participants who reported no such complaints (never), obtained from Kaplan–Meier analysis.

PRMQ scores (in z scores) as a function of frequency of self-reported complaints from others in Figure 1 show that the assessments covary, with average PRMQ scores for participants in the extreme category (often) exceeding the levels of self-reported everyday memory failures in those who indicated “never” by approximately 1.5 z-scores. Figure 1 further illustrates that prospective and retrospective subscales yield highly similar outcomes with regard to the association with the single-question self-report of complaints by family and friends. In the final model (Model 4), PRMQ score was no longer associated with dementia (HR = 1.05, 95% CI = 0.91–1.21 for all-cause dementia; HR = 1.05, 95% CI = 0.88–1.26 for AD). Instead, self-report of complaints often or usually by others of participants’ poor memory

1770

€ RONNLUND ET AL.

SEPTEMBER 2015–VOL. 63, NO. 9

was highly significant (HR = 2.98, 95% CI = 1.60–5.54 for all-cause dementia; HR = 3.86, 95% CI = 1.79–8.35 for AD). Figure 2 depicts the cumulative hazard of all-cause dementia over the study period, based on Kaplan-Meier analyses, for participants who reported that family and friends often or usually complained of their poor memory and for participants who reported no such complaints. A subdivision into prospective and retrospective scores yielded a pattern similar to that of the PRMQ total score; that is, there was a significant relationship between the prospective and retrospective scores and dementia (Model 1, HR = 1.18, 95% CI = 1.04–1.34 for prospective score; HR = 1.22, 95% CI = 1.08–1.38 for retrospective score for all-cause dementia, with similar values for AD), an association that persisted when additional covariates were added, except self-report of complaints from family and friends in Model 4 (HR = 1.03, 95% CI = 0.89–1.18 for prospective score; HR = 1.07, 95% CI = 0.93–1.22 for retrospective score). It could be that a minority of participants with low memory performance had a substantial influence on the results. To examine this possibility, all participants with memory performance 1 SD or more below the mean of their age group were excluded, resulting in a sample of 1,303 participants (165 participants with all-cause dementia, including 92 participants with AD). As in the previous analyses, PRMQ total score was a significant predictor of all-cause dementia (HR = 1.21, 95% CI = 1.04–1.40, P = .01) and AD (HR = 1.25, 95% CI = 1.03–1.52, P = .03). Adjustment for depressive symptoms and objective memory (Models 2 and 3) did not alter the results. In Model 4, the associations between self-reports pertaining to complaints of poor memory and AD were significant,

JAGS

whereas PRMQ scores were not associated with future AD diagnoses. As before, separate analyses of the prospective and retrospective yielded similar patterns. Finally, to examine the possibility that cases with near-onset dementia in terms of years to diagnosis drove the associations between PRMQ scores and dementia (Models 1–3), the analyses were rerun with a delayed entry. More specifically, the analysis was restricted to participants with a survival time of 5 years or more (n = 1,161 for no dementia, n = 107 for all-cause dementia, including 56 cases of AD). Mean time to dementia onset for the included participants was 6.8 years for allcause dementia and 6.7 years for AD. The results are summarized in Table 3. PRMQ score was a significant predictor of all-cause dementia in Model 1 (HR = 1.27, 95% CI = 1.05–1.52). Additional adjustment for depressive symptoms (Model 2) and objective memory (Model 3) had little effect on the HR. Unlike the PRMQ scores, the HR for objective memory was lower with a delayed entry (1.73 in Model 3) than with the analysis of the entire time period (2.43; Table 2). Inclusion of self-report of complaints of participants’ poor memory (Model 4) diminished the predictive value of PRMQ scores, and instead, these self-reports were significant. Analysis of the prospective and retrospective subscales yielded comparable values for Models 1 to 4 (Model 1: HR = 1.27, 95% CI = 1.06–1.52 for prospective scores and HR = 1.22, 95% CI = 1.01–1.45 for retrospective score). In the corresponding analysis with AD as the dependent variable, no association with PRMQ was observed (Model 1–4), although the association between AD and self-reports of complaints of poor memory by family and friends were significant in the final model (Model 4). Once more, patterns of results pertaining to the

Table 3. Summary of Cox Proportional Hazard Regression Analyses with All-Cause Dementia or Alzheimer’s Disease (AD) as the Dependent Variable Model 1 Variable

Dementia PRMQ total score (z-score) Objective memory (z-score) Complaints by others (reference never) Rarely Sometimes Often AD PRMQ total score (z-score) Objective memory (z-score) Complaints by others (reference never) Rarely Sometimes Often

Model 2

Model 3

Model 4

Hazard Ratio (95% Confidence Interval)

1.22 (1.07–1.38)b

1.19 (1.05–1.35)b

1.18 (1.04–1.34)b 2.43 (2.01–2.94)c

1.05 (0.91–1.21) 2.37 (1.95–2.87)c 1.65 (1.11–2.45)a 2.04 (1.38–3.00)c 2.98 (1.60–5.54)c

1.25 (1.06–1.47)b

1.22 (1.04–1.44)

1.22 (1.04–1.43)a 2.93 (2.28–3.77)c

1.06 (0.88–1.26) 2.80 (2.17–3.62)c 1.78 (1.07–2.97)a 2.06 (1.24–3.42)b 3.86 (1.79–8.35)c

PRMQ = Prospective and Retrospective Memory Questionnaire. Model 1, age, sex, marital status, and education as covariates. Model 2, age, sex, marital status, education, and depressive symptoms as covariates. Model 3, age, sex, marital status, education, depressive symptoms, and objective memory score as covariates. Model 4, age, sex, education, marital status, depressive symptoms, objective memory, and self-reported complaints of poor memory by others (friends and family) as covariates. P < a.05, b.01, c.001.

JAGS

SEPTEMBER 2015–VOL. 63, NO. 9

SELF-REPORTED MEMORY FAILURES AND FUTURE DEMENTIA

1771

Table 4. Summary of Cox Proportional Hazard Regressions for Participants with Time in Study of 5 Years or More with All-Cause Dementia or Alzheimer’s Disease (AD) as the Dependent Variable Model 1 Variable

Dementia PRMQ total score (z-score) Objective memory (z-score) Complaints by others (reference never) Rarely Sometimes Often AD PRMQ total score (z-score) Objective memory (z-score) Complaints by others (reference never) Rarely Sometimes Often

Model 2

Model 3

Model 4

Hazard Ratio (95% Confidence Interval)

1.27 (1.05–1.52)a

1.23 (1.02–1.49)a

1.23 (1.02–1.49)a 1.74 (1.33–2.29)b

1.11 (0.90–1.36) 1.68 (1.28–2.21)b 1.15 1.76 (1.01–3.05)a 2.93 (1.15–7.47)a

1.25 (0.97–1.62)

1.21 (0.93–1.58)

1.22 (0.94–1.58) 1.62 (1.10–2.38)a

1.05 (0.78–1.39 1.58 (1.08–2.31)a 1.44 (0.63–3.30) 2.24 (1.02–4.94)a 4.89 (1.36–17.60)a

PRMQ = Prospective and Retrospective Memory Questionnaire. Model 1, age, sex, marital status, and education as covariates. Model 2, age, sex, marital status, education, and depressive symptoms as covariates. Model 3, age, sex, marital status, education, depressive symptoms, and objective memory score as covariates. Model 4, age, sex, education, marital status, depressive symptoms, objective memory, and self-reported complaints of poor memory by others (friends and family) as covariates. P < a.05, b.001.

prospective and retrospective subscales were similar, and each mirrored the patterns observed for the PRMQ total scale (Table 4).

DISCUSSION A significant association was found between self-reported memory failure and incident dementia, consistent with a growing number of studies showing that older adults with subjective memory impairment are at greater risk of future dementia.8–13 The association held for all-cause dementia and for AD in basic models adjusting for demographic factors and persisted after additional adjustment for depressive symptoms.14,30 Thus, preclinical alterations in mood state are an unlikely source of the link between subjective memory assessment and incident dementia. Moreover, removal of participants with below-normal (≥1 SDs below the age group mean) memory performance did not alter the results, nor did analyses with a delayed entry to 5 years, in allcause dementia. Thus, the predictive value of subjective memory assessment persisted even when low performers were excluded and when the assessments were made long in advance of dementia onset (6.8 years on average). Similar patterns of results were observed for the prospective and retrospective subscales, suggesting that self-reported prospective and retrospective memory failures are equally sensitive to early signs of dementia or AD. Subjective assessment of memory failures was predictive of incident dementia also when objective memory performance was adjusted for statistically. Whereas a previous study10 found that self-reported memory was no longer associated with incident dementia when objective memory performance (and processing speed) was controlled for, the results of the present study are in line with

those of a previous study using different subjective memory measures.14 A more-precise account of these findings is needed. One explanation is that subjective assessments have validity as markers of intraindividual change, whereas objective tests (taken once) mainly provide information concerning levels of performance. Thus, a first possibility is that, even though objective and subjective memory deficits in preclinical dementia co-occur in time, objective impairments are harder to detect, unless the individual is subject to repeated longitudinal follow-up measurement. Moreover, some studies suggest a neural base of an independent contribution of subjective memory impairment to predicting dementia. For example, two studies showed that older adults with subjective memory impairment, but normal objective memory performance, had smaller entorhinal cortex volume bilaterally than controls.31,32 In the first study, no significant difference in hippocampus volume from that of controls was observed. The authors suggested that, in the earliest detectable stage of dementia (AD), a reduction of entorhinal cortex can be compensated for, and deficits in memory test performance are not necessarily apparent before the hippocampus is significantly affected. Similar reasoning could be applied to findings that white matter lesions33 and a measure of amyloid burden34 were related to subjective memory in the absence of objective impairment. Thus, a second possibility is that subjective memory deficits precede objective memory deficits in the preclinical stages of dementia or AD. Longitudinal analyses of data involving repeated measurement of subjective and objective memory will have to be conducted to provide evidence of the timing of subjective and objective memory deficits before dementia onset. Notwithstanding the persistence of the association between PRMQ scores and incident dementia after

1772

€ RONNLUND ET AL.

adjustment for important factors, the effect size must be regarded as modest, in particular in comparison with the effect size pertaining to the objective memory measure as a predictor of dementia (HR > 2 for the objective memory measure vs 1.22–1.25 for the PRMQ total score). Thus, alone and as a substitute for an objective memory assessment, the PRMQ might not perform well as a screening tool for dementia. Of more fundamental concern, simultaneous entry of the subjective assessment of how often family and friends complained of the participants’ poor memory, which was found to be most predictive of dementia in a previous study,14 eliminated the predictive value of the PRMQ scores and emerged as a highly significant predictor of dementia. An explanation of the finding that the single-question assessment concerning family’s and friends’ complaints of participants’ poor memory was more predictive of future dementia than a full questionnaire targeting memory failures is that it concerned feedback from other people, with several studies indicating that informant reports are a particularly valid indicator of preclinical memory impairment.10,35 Thus, the current results suggest that, at least in an early phase of the disease process, individuals have the ability to perceive and accurately report feedback from family and friends concerning subtle memory impairment, changes that may go unnoticed in objective tests administered at a single time. It is tempting to assume that selfreported memory failures and self-reported complaints of memory deficits by people close to the individual are independent markers of subtle objective deficits, but it is likely that self-reported observations and feedback from others are, at least partly, dependent or co-constructive in nature. More specifically, subjective complaints and self-reported everyday failures, presumably reflected by the PRMQ scores, are likely to be one basis for judgments of and feedback from others. It could also be that part of the predictive value of PRMQ scores was that feedback from family members and friends may influence self-report. Complaints of other people is perhaps less likely to have a substantial influence on assessment of specific everyday failures of the type included as part of the PRMQ but are a factor to consider in future studies involving subjective assessment of participants’ memory. An additional factor to consider to account for the better predictive value of the present single-question assessment is that certain extraneous factors, such as activity level (e.g., “forgetting appointments” will rarely occur if you have few appointments), may influence global memory judgments (such as the type used in the question pertaining to complaints from others) less than assessment of specific memory failures, which may account for the lack of age-related increase, or even a decrease, in PRMQ scores.19,36 At any rate, these results are important clinically, suggesting that physicians must take subjective reports of memory deficits by an older individual seriously, even in the absence of objective memory deficits, particularly when people close to the individual provide evidence in support of this. Whereas strengths of the present study deserve to be highlighted, including a large population-based sample, a comprehensive memory assessment, and long follow-up, limitations should also be noted. In particular, even if major demographic factors and mood state were both

SEPTEMBER 2015–VOL. 63, NO. 9

JAGS

adjusted for, other factors, possibly related to subjective memory and dementia, were not considered, for example quality of social relationships37 and neuroticism,38 which could have influenced the results. Additionally, the questionnaire was introduced at a time in the study when there had already been some attrition (e.g., mortality, refusal to participate), probably resulting in a final sample that may be slightly biased, with better memory performance and higher education level than the target population.39 In summary, self-reported memory failures were associated with incident dementia, even after adjustment for depressive symptoms and baseline memory performance, and the results persisted when memory assessments were made 5 or more years in advance of the diagnosis. Sensitivity to intraindividual change of subjective assessments may explain the independent contribution of subjective reports to prediction of dementia. Furthermore, suggestions of a neural basis for and independent contribution of subjective reports are available in the literature, including sensitivity to deterioration of the entorhinal cortex and white mater lesions. Finally, in spite of a persistent association between PRMQ and dementia, subjective reports of complaints by family and friends of poor memory appeared to be more useful in identifying individuals at risk of developing dementia. Thus, subjective complaint of poor memory, in particular complaints by family or friends, must not be ignored even in the absence of objective signs of impairment. In the light of the present findings, efforts should be made to construct subjective memory questionnaires that maximize their predictive validity,40,41 and questions concerning feedback on memory from friends and family are important to consider at this point.

ACKNOWLEDGMENTS Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Author Contributions: R€ onnlund: statistical analyses, writing first draft of manuscript. Sundstr€ om: data interpretation, revision of manuscript. Nilsson: design of Betula Study, data interpretation, revision of manuscript. Adolfsson: diagnosis of dementia, revision of original draft. All authors read and approved the final version of paper. Sponsor’s Role: The Betula Study is supported by the Bank of Sweden Tercentenary Foundation (Grants 1988– 0082:17, J2001–0682), Swedish Council for Planning and Coordination of Research (Grants D1988–0092, D1989– 0115, D1990–0074, D1991–0258, D1992–0143, D1997– 0756, D1997–1841, D1999–0739, B1999–474), Swedish Council for Research in the Humanities and Social Sciences (Grants F377/1988–2000), the Swedish Council for Social Research (Grants 1988–1990: 88–0082, 311/1991– 2000), and the Swedish Research Council (Grants 345– 2003–3883, 315–2004–6977).

REFERENCES 1. B€ackman L, Jones S, Berger A-K et al. Cognitive impairment in preclinical Alzheimer’s disease: A meta-analysis. Neuropsychology 2005;19:520–531. 2. Crumley JJ, Steller CA, Horhota M. Examining the relationship between subjective and objective memory performance in older adults: A metaanalysis. Psychol Aging 2014;29:250–263.

JAGS

SEPTEMBER 2015–VOL. 63, NO. 9

3. Balash Y, Mordechovich M, Shabtai H et al. Subjective memory complaints in elders: Depression, anxiety, or cognitive decline? Acta Neurol Scand 2013;127:344–450. 4. Bolla KI, Lindgren KN, Bonaccorsy ML et al. Memory complaints in older adults: Fact or fiction? Arch Neurol 1991;48:61–64. 5. Reid M, Parkinson L, Gibson R et al. Memory complaint questionnaire performed poorly as a screening tool: Validation against psychometric tests and affective measures. J Clin Epidemiol 2012;65:199–205. 6. Hanninen T, Reinikainen KJ, Helkala EL et al. Subjective memory complaints and personality traits in normal elderly subjects. J Am Geriatr Soc 1994;42:1–4. 7. Pearman A, Storandt M. Predictors of subjective memory in older adults. J Gerontol B Psychol Sci Soc Sci 2004;59B:P4–P6. 8. Jonker C, Geerlings MI, Schmand B. Are memory complaints predictive of dementia? A review of clinical and population-based studies. Int J Geriatr Psychiatry 2000;15:983–991. 9. St John P, Montgomery P. Are cognitively intact seniors with subjective memory loss more likely to develop dementia? Int J Geriatr Psychiatry 2002;17:814–820. 10. Rabin LA, Wang C, Katz MJ et al. Predicting Alzheimer’s disease: Neuropsychological tests, self-reports, and informant reports of cognitive difficulties. J Am Geriatr Soc 2012;60:1128–1134. 11. Waldorff FB, Siersma V, Vogel A et al. Subjective memory complaints in general practice predicts future dementia: A four year follow-up study. Int J Geriatr Psychiatry 2012;27:1180–1188. 12. Wang L, van Belle G, Crane PK et al. Subjective memory deterioration and future dementia in people age 65 and older. J Am Geriatr Soc 2004;52:2045–2051. 13. Mitchell AJ, Beaumont H, Ferguson D et al. Risk of dementia and mild cognitive impairment in older people with memory complaints: A metaanalysis. Acta Psychiatr Scand 2014;130:439–451. 14. R€ onnlund M, Sundstr€ om A, Adolfsson R et al. Subjective memory impairment in older adult predicts future dementia independent of memory performance: Evidence from the Betula Prospective Cohort Study. Alzheimers Dement 2015. doi: http://dx.doi.org/10.1016/j.jalz.2014.11.006 15. Nilsson LG, B€ ackman L, Erngrund K et al. The Betula Prospective Cohort Study: Memory, health and aging. Aging Neuropsychol Cogn 1997;4:1–32. 16. Nilsson LG, Adolfsson R, B€ackman L et al. Betula: A prospective cohort study on memory, health and aging. Aging Neuropsychol Cogn 2004;11:134–148. 17. Smith GV, Della Sala S, Logie RH et al. Prospective and retrospective memory in normal ageing and dementia: A questionnaire study. Memory 2000;8:311–321. 18. Crawford JR, Smith GS, Maylor EA et al. The Prospective and Retrospective Questionnaire (PRMQ): Normative data and latent structure in a large non-clinical sample. Memory 2003;11:261–275. 19. R€ onnlund M, M€ antyl€a T, Nilsson L-G. The Prospective and Retrospective Memory Questionnaire (PRMQ): Factorial structure, relations to global subjective memory ratings, and Swedish norms. Scand J Psychol 2008;49:11–18. 20. Karantzoulis S, Troyer AK, Rich JB. Prospective memory in amnestic mild cognitive impairment. J Int Neuropsychol Soc 2009;15:407–415. 21. Hsu Y-H, Huang C-F, Tu M-C et al. The clinical utility of informants’ appraisals on prospective and retrospective memory in patients with early Alzheimer’s disease. PLoS One 2014;9:1–8. 22. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 4th Ed, Technical Revision. Washington, DC: American Psychiatric Association, 2000.

SELF-REPORTED MEMORY FAILURES AND FUTURE DEMENTIA

1773

23. McHann GM, Knopman DS, Chertkow H et al. The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association work groups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:263–269. 24. Gorelick PB, Scuteri A, Black SE et al. Vascular contributions to cognitive impairment and dementia: A statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2011;42:2672–2713. 25. Dixon RA, Hultsch DF, Hertzog C. The Metamemory in Adulthood (MIA) Questionnaire. Psychopharmacol Bull 1988;24:671–688. 26. Kliegel M, J€ager T. Can the Prospective and Retrospective Memory Questionnaire (PRMQ) predict actual prospective memory performance? Curr Psychol 2006;25:182–191. 27. Radloff LS. The CES-D scale: A self-report depression scale for research in the general population. Appl Psych Meas 1977;1:385–401. 28. R€ onnlund M, Nilsson LG. The Betula study: Reliabilities and long-term stabilities. Baltic J Psychol 2006;7:6–14. 29. Folstein MF, Folstein SE, McHugh PR. ‘Mini-mental state’. A practical method for grading the cognitive status of patients for the clinician. J Psychiat Res 1975;12:189–198. 30. Schmand B, Jonker C, Geerlings MI et al. Subjective memory complaints in the elderly: Depressive symptoms and future dementia. Br J Psychiatry 1997;171:373–376. 31. Jessen F, Feyen L, Freymann K et al. Volume reduction of the entorhinal cortex in subjective memory impairment. Neurobiol Aging 2006;27:1751– 1756. 32. Striepens N, Scheef L, Wind A et al. Volume loss of the medial temporal lobe structures in subjective memory impairment. Dement Geriatr Cogn 2010;29:75–81. 33. Minett TS, Dean JL, Firbank M et al. Subjective memory complaints, white-matter lesions, depressive symptoms and cognition in elderly patients. Am Geriatr Psychiatry 2005;13:665–671. 34. Amariglio RE, Becker JA, Carmasin J et al. Subjective cognitive complaints and amyloid burden in cognitively normal older individuals. Neuropsychologia 2012;50:2880–2886. 35. Carr B, Gray S, Baty J et al. The value of informant versus individual’s complaints of memory impairment in early dementia. Neurology 2000;55:124–126. 36. R€ onnlund M, Vestergren P, M€antyl€a T et al. Predictors of self-reported prospective and retrospective memory in a population-based sample of older adults. J Genet Psychol 2011;172:266–284. 37. Amieva H, Stoykova R, Matharan F et al. What aspects of social network are protective for dementia? Not the quantity but the quality of social interactions is protective up to 15 years later. Psychosom Med 2010;72:905–911. 38. Johansson L, Guo X, Duberstein PR et al. Midlife personality and risk of Alzheimer disease and distress. Neurology 2014;83:1538–1544. 39. R€ onnlund M, Nyberg L, B€ackman L et al. Stability, growth, and decline in adult life-span development of declarative memory: Cross-sectional and longitudinal data from a population-based study. Psychol Aging 2005;20:3–18. 40. Burmester B, Leathem J, Merrick P. Assessing subjective memory complaints: A comparison of spontaneous reports and structured questionnaires. Int Psychogeriatr 2014;27:61–77. 41. Vestergren P, R€ onnlund M, Nyberg L et al. Multigroup confirmatory factor analysis of the Cogntive Dysfunction Questionnaire: Instrument refinement and measurement invariance across age and sex. Scand J Psychol 2012;53:390–400.