- Split View
-
Views
-
Cite
Cite
Lars Bäckman, Sari Jones, Brent J. Small, Hedda Agüero-Torres, Laura Fratiglioni, Rate of Cognitive Decline in Preclinical Alzheimer's Disease: The Role of Comorbidity, The Journals of Gerontology: Series B, Volume 58, Issue 4, July 2003, Pages P228–P236, https://doi.org/10.1093/geronb/58.4.P228
- Share Icon Share
Abstract
We investigated the influence of individual-difference variables implicated as risk factors for Alzheimer's disease (AD) or known to be related to cognitive performance in normal aging (e.g., age, sex, years of education, previous and recent diseases, apolipoprotein E status, social network, and substance use) on rate of cognitive change from preclinical to clinical AD. With the use of data from a population-based study, 230 persons who were nondemented at baseline and diagnosed with AD at a 3-year follow-up were examined with the Mini-Mental State Examination (MMSE). Of all predictor variables examined, only number of diseases resulting in hospital admission during the follow-up period made an independent contribution to rate of MMSE change. These results suggest that many variables affecting the onset of the degenerative process as well as cognitive functioning in normal aging exert little influence on rate of cognitive change in preclinical AD. This may reflect the fact that the emerging dementia disease overshadows the role of these variables for cognitive functioning. A possible exception to this pattern is that an increasing number of concomitant health conditions may exacerbate the rate of cognitive decline during the final portion of the preclinical phase in AD.
A characteristic feature of normal cognitive aging is the large interindividual variability in performance. Although many cognitive functions decline in normal aging, there is systematic variation across multiple demographic (e.g., age, sex, and years of education), social (e.g., activity patterns and social network), health-related (e.g., depression and vitamin status), and genetic (e.g., apolipoprotein E, or APOE, genotype) domains, with regard to the size of the age-related cognitive deficit (for reviews, see Bäckman, Small, Wahlin, & Larsson, 1999; Hultsch, Hertzog, Dixon, & Small, 1998).
Several studies indicate that the importance of individual-difference variables for cognitive performance is reduced in Alzheimer's disease (AD). For example, considerable research effort has been directed at examining whether onset age influences performance at a given time as well as rate of cognitive decline in clinical AD. Although some research suggests that early-onset patients may be more impaired and exhibit a faster rate of decline than late-onset patients (Burns, Jacoby, & Levy, 1991; Wilson, Gilley, Bennett, Beckett, & Evans, 2000), other studies show the opposite pattern (Huff, Growdon, Corkin, & Rosen, 1987), and still others demonstrate negligible effects of onset age on cognitive functioning in AD (Bäckman, Hill, Herlitz, Fratiglioni, & Winblad, 1994; Rubin et al., 1993; Small, Viitanen, Winblad, & Bäckman, 1997). In addition, a number of other variables known to influence cognitive functioning in normal aging have been found to have limited effects in AD. These include education (Bäckman et al., 1994; Katzman et al., 1983), sex (Buckwalter, Sobel, Dunn, Diaz, & Henderson, 1993; Teri, Hughes, & Larson, 1990), and depression (Bäckman, Hassing, Forsell, & Viitanen, 1996; Fahlander, Berger, Wahlin, & Bäckman, 1999), as well as a variety of biological variables such as blood pressure, thyroid function, and vitamin B12 level (Bäckman et al., 1994; Hill, Bäckman, Wahlin, & Winblad, 1995; Small, Viitanen, Winblad, et al., 1997). A possible reason for the weak association between these variables and cognitive functioning in AD is that the influence of various subject characteristics is overshadowed by the pathogenesis itself. In other words, the dementing process may make people more cognitively similar.
Converging evidence indicates that there is a preclinical phase in AD during which cognitive deficits are detectable (e.g., Andel et al., 2001; Masur, Sliwinski, Lipton, Blau, & Crystal, 1994; Tierney et al., 1996). Such preclinical deficits have been documented for global indices of cognitive performance (e.g., Fabrigoule et al., 1998; Small, Viitanen, & Bäckman, 1997), such as the Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975), as well as for specific tasks assessing memory (e.g., Linn et al., 1995), verbal (Jacobs et al., 1995), and visuospatial (e.g., Small, Herlitz, Fratiglioni, Almkvist, & Bäckman, 1997) functioning. In most of these studies, the retest interval has been approximately 3 years, although preclinical cognitive deficits in AD have been documented by using follow-up periods of more than 10 years (Elias et al., 2000; La Rue & Jarvik, 1987).
Although cognitive deficits are observed many years before clinical diagnosis, evidence is mixed regarding the exact time point at which accelerated cognitive decline begins in preclinical AD. Hall and colleagues (Hall, Lipton, Sliwinski, & Stewart, 2000; Hall et al., 2001) observed accelerated decline in measures of memory more than 5 years before diagnosis, although speeded measures of performance IQ exhibited accelerated decline approximately 2 years before diagnosis. However, Bäckman, Small, and Fratiglioni (2001) found clear preclinical memory deficits 6 years before the AD diagnosis, although precipitous decline did not occur until the last 3 years preceding diagnosis (see also Small, Fratiglioni, Viitanen, Winblad, & Bäckman, 2000). Accelerated cognitive decline during the final portion of the preclinical phase in AD has been observed in several other studies (e.g., Chen et al., 2001; Fox, Warrington, Sieffer, Agnew, & Rossor, 1998).
Knowledge pertaining to whether various individual-difference variables are systematically related to rate of cognitive decline during the final years of the preclinical phase of AD should be of both theoretical and applied value. Theoretically, such knowledge may shed further light on the development of the disease process. Further, if potentially modifiable factors (e.g., concomitant diseases or social network) are related to progression rate, this should prove useful from a prevention perspective. Although attempts to link different subject characteristics to cognitive functioning in clinical AD have met with limited success, such associations may be more likely to be present when the disease has not progressed to a point at which the pathogenic process absorbs the influence of other variables. This important issue was addressed in a recent population-based study (Wilson, Beckett, Bennett, Albert, & Evans, 1999). In this study, persons aged 65 and older were assessed annually across an average period of 3.5 years on measures of memory, language, and visuoconstruction. The results showed that age had no effect on rate of cognitive decline in preclinical AD.
The purpose of the present study was to provide further knowledge concerning potential predictors of cognitive change during the preclinical phase of AD. Using data from a population-based sample of older adults, we examined 3-year changes in MMSE performance in a group of persons who were nondemented at baseline but who were diagnosed with AD at follow-up. Thus, a unique feature of this study is that we focused exclusively on individual differences in rate of cognitive change among incident AD patients. Including a comparison group of nondemented persons was not considered meaningful, given the minimal changes in MMSE performance across 3 years in normal aging (e.g., Small, Viitanen, & Bäckman, 1997; Small et al., 2000).
The examined predictor variables included demographic factors, such as age, education, and sex; specific health conditions, such as depression, vitamin deficiency, and high blood pressure; composite measures of previous and recent diseases as determined from hospital registers; APOE genotype; social network; and substance use. The rationale for selecting these variables was that they have been implicated in normal cognitive aging or as risk factors for AD.
Increasing age and low education are related to both lower cognitive performance among nondemented older adults (e.g., Bäckman et al., 1999; Nilsson et al., 1997) and to an increased risk of AD (e.g., Gatz et al., 2001; Nielsen, Lolk, Andersen, Andersen, & Kragh-Sorensen, 1999). With regard to sex, there is emerging evidence of an increased risk of late-onset AD in women compared with men (e.g., Fratiglioni et al., 1997; Fratiglioni, Launer, et al., 2000), although sex differences in global cognitive ability are small or nonexistent in normal aging (e.g., Halpern & LaMay, 2000; Hill & Bäckman, 1995). Deficits in cognitive performance as well as an increased risk of AD have also been linked to depression (e.g., Devanand et al., 1996; Kindermann & Brown, 1997), vitamin deficiency (e.g., Hassing, Wahlin, Winblad, & Bäckman, 1999; Wang et al., 2001), and high blood pressure (e.g., Elias, Wolf, D'Agostino, Cobb, & White, 1993; Skoog et al., 1996). In addition, there is strong evidence that presence of the APOE ϵ-4 allele increases the risk of AD (e.g., Petersen et al., 1995), although it remains unclear whether APOE status influences cognitive performance among nondemented older adults (Bondi, Salmon, Galasko, Thomas, & Thal, 1999; Wilson et al., 2002). Recent evidence suggests that a poor social network is associated with an increased risk of AD (Fratiglioni, Wang, Ericsson, Maytan, & Winblad, 2000), and numerous studies indicate that participation in social, cognitive, and physical activities may promote good cognitive performance in normal aging (Bäckman et al., 1999; Hultsch et al., 1998). Finally, although it has been suggested that substance use (i.e., smoking and alcohol use) may be related to lower cognitive performance in normal aging and increase the risk of AD, the available evidence is somewhat mixed (Bäckman et al., 1999; Fratiglioni & Rocca, 2001).
Of chief interest here was whether these variables are related to rate of cognitive change during the 3 years preceding clinical diagnosis of AD. Although numerous studies have examined individual differences in cognitive performance and change among normal old adults as well as among clinical AD patients, to our knowledge no such research has systematically examined the role of individual-difference variables in cognitive change with a focus on preclinical AD cases.
Methods
Participants
The study sample was taken from a population-based study, including all inhabitants aged 75 years and older in the Kungsholmen parish of Stockholm, Sweden. More detailed information about the study and methods used has been provided elsewhere (Fratiglioni et al., 1991, 1997). Time 1 involved two phases. In the first phase, 1,810 individuals were administered a questionnaire, which included the MMSE, to detect suspected dementia cases. In the second phase, all those with an MMSE score below 24 (n = 314) and a control sample, matched on age and sex distribution, of those with an MMSE score above 23 (n = 354) were invited back for further investigation. These participants were assessed with extensive medical, neurological, and psychiatric examinations; social and family interviews; laboratory blood analyses; and a comprehensive cognitive test battery. All participants were invited back to be reexamined 3 and 6 years later. The same examinations as in the second phase of Time 1 were administered to all participants at these follow-ups. All participants were tested individually at the Stockholm Gerontology Research Center or in their homes.
The diagnostic criteria and procedures to reach the clinical diagnosis of dementia, as well as type of dementia, at baseline and follow-up, were as follows: In Step 1, a preliminary diagnosis was made after a common discussion among the geriatricians who had examined the participant and reviewed his or her social and family history. Step 2 involved a second preliminary diagnosis of all participants by a physician expert in dementia. In Step 3, the two preliminary diagnoses were compared, and cases with discordant diagnoses were reviewed again to ascertain causes of agreement and disagreement. This eliminated most of the discordant diagnoses. However, in those cases in which disagreement persisted, a supervising physician made the final diagnosis. This process yielded a diagnosis of the presence or absence of dementia according to the criteria in the Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev.; DSM–III–R; American Psychiatric Association, 1987). Type of dementia was further defined according to standardized criteria (Fratiglioni, Grut, Forsell, Viitanen, & Winblad, 1992). The AD diagnosis corresponds to probable AD, according to criteria set forth by the National Institute of Neurological and Communicative Disorders and Stroke—Alzheimer's Disease and Related Disorders Association (McKhann et al., 1984).
The present study includes those persons who were nondemented at Time 1 and received a diagnosis of AD at Time 2 (n = 164). Of these, 9 (5.49%) had died and 5 (3.05%) refused MMSE testing at Time 2. For those who had died, the AD diagnosis was confirmed from clinical records. The mean retest interval between Time 1 and Time 2 was 3.27 years (SD = 0.57). In addition, those who were nondemented at Time 2 and received a diagnosis of AD at Time 3 were included (n = 93). Of these, 10 (10.75%) had died and 3 (3.22%) refused MMSE testing at Time 3. The mean retest interval between Time 2 and Time 3 was 3.32 years (SD = 0.55). The length of the two retest intervals was indistinguishable (p >.50). Thus, the final study sample comprised 150 incident AD cases at Time 2 and 80 incident AD cases at Time 3. Table 1 shows participant characteristics at baseline for the two groups of incident AD cases.
The incident AD persons at Time 2 were younger, that is, F(1,228) = 8.61, MSE = 20.07, and p <.01, and had lower baseline performance on the MMSE, that is, F(1,228) = 13.72, MSE = 7.78, and p <.001, compared with the incident AD persons at Time 3. In addition, there was a larger proportion of women among the incident AD persons at Time 2: χ2(1, N = 230) = 4.36, p <.05. These differences were all expected, given the 3-year interval between Time 2 and Time 3. To maximize the sample size and the variability in the predictor variables, we aggregated the two samples of incident AD cases in subsequent analyses. Baseline characteristics of the aggregated sample are shown in the rightmost column of Table 1. In the aggregated sample, baseline was defined as the measurement point preceding diagnosis for all participants; hence, the mean follow-up period was 3.29 years (SD = 0.56) for the total sample of 230 participants.
Measures and Procedure
Outcome measure
The primary outcome measure was the change from baseline to follow-up on the MMSE, a measure of global cognitive functioning (Folstein et al., 1975), with a maximum score of 30. A Swedish version of the MMSE was administered according to the same standardized procedure at baseline and follow-up. The Swedish MMSE version has been found to possess good psychometric properties (Fratiglioni et al., 1992; Hill & Bäckman, 1995).
Predictor Variables
Demographic variables
The demographic variables included age at baseline, sex, and years of formal education.
Disease variables
Two disease measures were calculated by using information from a register of all inpatients from the hospitals in the Stockholm area. This register includes main diagnoses and up to five secondary diagnoses for all hospital admissions since April 1969. Both main and secondary diagnoses were included in the present analyses.
Previous disease indicates whether a person had been admitted to a hospital with any disease except for dementia before the date of the baseline assessment in this study. The diseases were categorized according to the International Coding of Diseases (ICD)-8 (hospital admissions before 1987) or ICD-9 (hospital admissions between 1987 and 1996) Coding Standards, and the maximum number of disease categories was 15. These were infectious disease, cancer, endocrine disease, blood disease, mental disease (excluding dementia), nervous system disease, cardiovascular disease, cerebrovascular disease, respiratory disease, digestive disease, genitourinary disease, skin disease, muscular disease, unspecified symptoms and signs, and injuries. In the present sample, the mean number of previous diseases was 2.35 (SD = 1.99). The number of previous diseases ranged from 0 to 9, and 49 persons (21.30%) had no previous diseases.
Recent diseases were calculated by determining how many of the aforementioned diseases resulted in hospital admission between baseline and follow-up. The mean number of recent diseases was 1.73 (SD = 1.94), ranging from 0 to 8. Ninety-six persons (41.74%) had no recent diseases.
An alternative way of operationalizing the two disease variables is to examine the number of hospital admissions resulting from these diseases. We examined this variable and found that it exerted nearly identical effects as the number of disease variables. Thus, the analyses reported are restricted to the number of disease variables.
In addition to these register-based disease variables, we included three specific conditions that were clinically assessed in the Kungsholmen project: high blood pressure, depression, and vitamin B12 deficiency. For each of these variables, both baseline scores and change scores (between baseline and follow-up) were examined.
High blood pressure was defined as systolic blood pressure of 160 or above and diastolic blood pressure of 90 or above. Six persons (2.61%) had missing blood pressure data either at baseline or follow-up. At baseline, 55 persons (25.23%) were classified as having high blood pressure, whereas 169 (75.45%) had normal blood pressure. At follow-up, 43 persons (19.20%) had changed from having high blood pressure to normal blood pressure, 152 (67.85%) stayed the same, and 29 (12.95%) previously classified as having normal blood pressure had changed to high blood pressure.
Depression comprises both participants who were diagnosed with major depression and those with dysthymia according to DSM–III–R criteria. Major depression and dysthymia differ in severity and duration but share the same basic depressive symptoms. Dysthymia is a somewhat less pronounced depressive disorder, but with a longer duration. Diagnostic information on depression was available on 147 persons (63.91%). Of these, 17 (11.56%) received a diagnosis of depression at baseline. At follow-up, 10 persons (6.80%) were no longer depressed, 121 persons (82.31%) stayed the same, and 16 persons (10.88%) had developed depression.
Vitamin B12 values were obtained from blood samples collected in the morning before the cognitive testing. Vitamin data were available on 126 persons (54.78%) at both baseline and follow-up. For these persons, the mean B12 level was 364.45 pmol/L (SD = 367.28). The vitamin B12 variable was dichotomized into those that had deficiency (n = 35; 25.40%) and those who did not (n = 107; 74.60%). Following previous research on vitamin B12 in very old adults, a cutoff value was set at 200 pmol/L (Hassing et al., 1999; Wang et al., 2001). At follow-up, 11 persons (8.73%) were no longer classified as vitamin deficient, 97 (76.98%) stayed the same, and 18 (14.29%) were deficient.
APOE status
APOE genotyping was carried out by DNA microsequencing (AffiGen APOE, Sangtec Medical, Bromma, Sweden). The procedure has been described in detail elsewhere (Basun et al., 1996; Small, Basun, & Bäckman, 1998). Genotyping was done blind to the clinical information. APOE data were available on 187 persons (81.30%) and categorized on the basis of the presence (n = 64; 34.22%) or absence (n = 123; 65.78%) of at least one ϵ4 allele.
Social network
Social network was coded into four levels by using the following three criteria: being married or living with a spouse or partner, having children, and having a friend in whom the participant can confide (Fratiglioni, Wang, et al., 2000). If none of the criteria were met (0), the subject was classified as having a poor social network (n = 34; 14.78%). If one criterion was met (1), the person was classified as having a limited social network (n = 92; 40.00%). Meeting two criteria (2) was classified as a moderate social network (n = 79; 34.35%), and meeting all three criteria (3) was classified as having an extensive social network (n = 25; 10.87%). Change in social network from baseline to follow-up was also examined. The social network could either become poorer for two criteria (n = 1; 0.43%) or one criterion (n = 20; 8.70%), stay the same (n =154; 66.96%), improve for one criterion (n = 52; 22.61%), or improve for two criteria (n = 3; 1.30%).
Substance use
Smoking and alcohol habits at baseline were calculated as dichotomous variables, that is, those who drank alcohol or smoked versus those who did not (Wang, Fratiglioni, Frisoni, Viitanen, & Winblad, 1999). Data were available for 142 persons (61.74%) on the smoking variable. Out of these, 14 (9.86%) smoked at baseline. For the alcohol variable, the corresponding figures were 145 (63.04%) and 92 (63.45%), respectively. No data on substance use were available at the second follow-up, and hence no change scores were calculated for these variables.
Data Analysis
The basic analytic strategy involved residualized change regressions whereby MMSE score at follow-up served as the dependent variable and MMSE at baseline was entered in the first step. Using this strategy, we first examined the various blocks of variables separately. In subsequent analyses, we focused on those predictor variables that were significantly related to rate of MMSE change when they were examined individually.
Results
Overall Change in MMSE Score
There was a significant overall change in MMSE score between baseline and follow-up, that is, t(229) = 18.28 and p <.001. The mean MMSE score was 25.20 (SD = 2.87) at baseline and 19.32 (SD = 5.27) at follow-up, yielding an average decline of 5.88 MMSE points over the follow-up period. The mean annual decline was 1.81 points (SD = 1.57) in the present sample of incident AD cases.
Predictors of Cognitive Change
As expected, MMSE score at baseline was related to MMSE score at follow-up (R2 =.16, ß =.40, and p <.001). Table 2 depicts the amount of variance in MMSE change accounted for by each block of predictor variables when they are entered one at a time, and when MMSE score is controlled for at baseline. Of all categories of predictor variables examined, only the demographic and disease blocks were reliably related to MMSE change. In the block of demographic variables, age was the only significant predictor, with increasing age being associated with more cognitive decline from baseline to follow-up.
In the disease block, number of recent diseases was the only significant predictor, indicating that an increasing number of diseases between baseline and follow-up was associated with more decline on the MMSE. This effect is illustrated in Figure 1, in which the incident AD cases are grouped according to whether they had no recent diseases (n = 96), one or two recent diseases (n = 61), or three or more recent diseases (n = 73).
As expected, age and recent diseases were related (r =.19 and p <.01), with increasing age being associated with more recent diseases. In subsequent analyses, we sought to determine the relative importance of age and recent diseases for rate of MMSE change in preclinical AD. This was accomplished by conducting hierarchical regression analyses in which the order of entry between age and recent diseases was varied and MMSE at baseline was controlled for. The outcome from these analyses is shown in Table 3. When age was entered before recent diseases, both of these variables made independent contributions to rate of MMSE change. Importantly, however, when the order of entry was reversed, recent diseases alone was reliably related to rate of decline on the MMSE. There was no significant interaction between age and recent diseases.
Further, to determine whether the effect of recent diseases on MMSE change generalized across the aggregated samples used in these analyses, we computed the interaction term between number of recent diseases and time of onset of dementia (Time 2 vs. Time 3). The interaction term fell far short of significance (p >.50), indicating that time of dementia onset did not influence the effect of recent diseases on MMSE change.
By including both baseline and change scores in the analyses for some of the predictor variables (i.e., high blood pressure, depression, vitamin deficiency, and social network), we run the risk for colinearity, which may explain the lack of relationship between these variables and MMSE change. Hence, we conducted separate analyses for baseline and change scores for these variables. However, the lack of significant associations remained when we analyzed baseline and change scores separately.
All these analyses were repeated by excluding individuals with a baseline MMSE score lower than 24 to control for the possibility that the results were affected by missing some persons who may have already been clinically demented at baseline (Sliwinski, Lipton, Buschke, & Stewart, 1996). However, the outcome from these analyses was virtually identical to that of the original analyses.
Finally, we reanalyzed the data by using change scores (Time 2–Time 1) rather than residual scores as outcome measures. The results of these analyses were the same as those reported for residual scores. In these analyses, we also assessed the influence of initial level of MMSE performance on rate of change. In so doing, we classified the sample into those who scored relatively high (M = 27.06, SD = 0.97, and n = 126) versus those who performed at a relatively low level (M = 22.96, SD = 2.80, and n = 104) on the MMSE at baseline. Results indicated no differential rate of change as a function of initial MMSE performance (p >.30). Specifically, the high-scoring group dropped to a mean of 20.49 (SD = 4.65) and the low-scoring group averaged 17.89 (SD = 5.63) at follow-up. We also examined the influence of initial MMSE status on the effect of number of recent diseases on MMSE change. As with the corresponding analysis on time of dementia onset, this analysis revealed that the effect of recent diseases generalized across initial MMSE status (p >.30).
Discussion
This study demonstrated significant change in a global measure of cognitive functioning (i.e., MMSE score) during the 3 years preceding the clinical diagnosis of AD. The rate of cognitive change was similar among persons who scored relatively high versus those that scored relatively low on the MMSE at baseline. This finding of robust cognitive decline in the final part of the preclinical phase of AD is consistent with previous research (e.g., Fox et al., 1998; Small, Viitanen, & Bäckman, 1997). The mean annual decline on the MMSE was 1.81 points. This figure is somewhat lower than corresponding figures in clinical AD, in which mean annual decline estimates around 3 points are typically reported (e.g., Clark et al., 1999; Wilson et al., 2000). This difference may reflect the fact that the rate of cognitive decline is slower in preclinical compared with clinical AD. Another possibility is that once precipitous decline occurs in preclinical AD, it is as rapid as in manifest AD, but in this sample all participants did not decline over the entire retest interval. The design of this population-based study does not allow for disentangling these possibilities.
The present results also demonstrated that few of the variables examined were related to rate of MMSE change from baseline to follow-up. Years of education, sex, APOE status, previous disease history as determined from hospital records, baseline status on substance use, blood pressure, depression, vitamin B12 level, and social network were unrelated to MMSE change. Note that for the latter four variables, change from baseline to follow-up was also unrelated to rate of cognitive decline. The lack of relationship between these variables and rate of MMSE change is noteworthy, as these variables have been associated with cognitive performance in normal aging or have been implicated as risk factors for AD in numerous studies from the Kungsholmen project (e.g., Bäckman & Forsell, 1994; Fahlander et al., 2000; Fratiglioni et al., 1997; Fratiglioni, Wang, et al., 2000; Guo, Viitanen, Winblad, & Fratiglioni, 1999; Hassing et al., 1999; Hill, Wahlin, Winblad, & Bäckman, 1995; Qiu, Bäckman, Winblad, Agüero-Torres, & Fratiglioni, 2001; Small et al., 1998; Wahlin et al., 1993) as well as in related research (e.g., Bäckman et al., 1999; Hultsch et al., 1998; Letenneur et al., 2000; Petersen et al., 1995).
From a different perspective, these findings suggest that rate of cognitive decline in preclinical AD is invariant across multiple demographic, social, health-related, and genetic factors. Thus, the overall pattern of data is consistent with both cross-sectional and longitudinal reports in clinical AD, suggesting that the influence of many individual-difference variables on cognitive functioning may be overshadowed by the dementing process (Bäckman et al., 1994, 1996; Buckwalter et al., 1993; Fahlander et al., 1999; Hill et al., 1995; Small, Viitanen, Winblad, et al., 1997; Teri et al., 1990). The current results extend these observations to the preclinical phase of AD. An important point to note is that although these factors have limited effects on the progression once the disease process starts accelerating, they may be active in promoting the development of the degenerative process. In other words, these variables may play a role as risk factors, but not as precipitating or prognostic factors.
There were, however, two exceptions to this pattern of findings: Increasing age and number of diseases resulting in hospital admission between baseline and follow-up were associated with a more rapid rate of cognitive decline. As expected, age and number of recent diseases were related such that increasing age was associated with an increasing number of conditions. Perhaps the most important finding from this study was that number of recent diseases was related to rate of cognitive decline independent of age, whereas the reverse was not true. When controlling for number of recent diseases, the effect of age on rate of cognitive decline was eliminated. Thus, number of recent diseases was the stronger predictor of cognitive decline and mediated the effect of age observed in the univariate analysis. The finding of no independent effect of age on rate of cognitive decline in preclinical AD is in agreement with recent findings from a population-based study by Wilson and colleagues (2000). Note also that there was no interaction effect between age and recent diseases, indicating that the effect of recent diseases on rate of MMSE change generalized across the age range examined. Of further note is that neither time of dementia onset (Time 2 vs. Time 3) nor initial level of MMSE performance influenced the effect of recent diseases on MMSE change.
The important question, then, is how we should account for the influence of number of recent diseases on the magnitude of cognitive decline in preclinical AD. One possibility is that an increasing number of conditions between baseline and follow-up may be related to the dementing process. Although all participants in the sample were diagnosed with AD at follow-up, the 3-year period elapsing between assessment occasions does not make it possible to determine at which specific time point during the retest interval a dementia diagnosis could have been rendered. Thus, it may be that be that those persons who already were clinically demented early on during the follow-up interval were more likely to develop other conditions than those who would be diagnosed closer to the follow-up assessment. If this were to be the case, the relationship between number of recent diseases and rate of change in preclinical AD would simply reflect the fact that AD is associated with an increased risk of developing other diseases.
Although the present study design cannot rule out this possibility, several observations indicate that it is not particularly likely. First, the effects of AD on the development of other health conditions remain unclear, with some studies reporting an increased risk (Burke et al., 1994; Chandra, Bharucha, & Schoenberg, 1986), others a decreased risk (Holstein, Chatellier, Piette, & Moulias, 1994; Wolf-Klein et al., 1988), and still others no effect (Zubenko et al., 1997). In addition, in those studies reporting an AD-related increase in the risk of developing other diseases, the focus has been on clinically demented patients relatively late in the disease process. Conceivably, an increased risk of developing other conditions should be more likely to occur late rather than early in the pathogenesis.
Second, the effects of number of recent conditions on rate of MMSE change were observed after MMSE performance at baseline was controlled for. Indeed, Figure 1 shows that persons with 0, 1–2, or 3+ recent diseases had virtually identical MMSE scores at baseline. Nevertheless, there was a systematic increase in rate of decline as a function of number of recent diseases. Finally, there is pervasive evidence that health conditions are associated with cognitive performance and decline among nondemented older adults (for a review, see Bäckman et al., 1999). Conceivably, serious health conditions have a stronger impact on cognitive functioning in normal aging compared with the other predictor variables examined (e.g., sex, education, vitamin B12 level, and social network). As a result, although the influence of individual-difference variables on cognitive functioning is reduced in preclinical AD, residual effects may be more easily observed for a variable in which the effects are sizable among nondemented elderly adults.
If one considers that the present sample showed was nondemented at baseline assessment, the fact that number of recent diseases was related to rate of cognitive change is consistent with the bulk of research on health and cognition in normal aging. In this respect, the present finding of a relationship between disease and cognitive decline in preclinical AD suggests an interesting difference between preclinical and clinical AD with regard to factors that may modify cognitive functioning: Whereas the importance of other diseases to cognitive performance may be overshadowed by the dementing process in clinical AD (Agüero-Torres, Fratiglioni, Guo, Viitanen, & Winblad, 1998; Bäckman et al., 1994, 1996; Fahlander et al., 1999; Hill et al., 1995; Small, Viitanen, et al., 1997), there still appears to be room for comorbidity to exert an influence on cognition in preclinical AD. Collectively, these observations suggest that the effects of recent diseases on rate of cognitive decline in preclinical AD observed in this study may be independent of the emerging dementia disease.
Even though this research has yielded novel results with regard to the cognitive transition from preclinical to clinical AD, several limitations should be acknowledged. First, only the MMSE was available to assess cognitive performance in this sample. The MMSE is a global assessment instrument that lacks specificity with regard to the underlying cognitive processes. Although MMSE performance is strongly related to performance on specific tasks assessing memory, verbal, and spatial functioning in preclinical AD (e.g., Bäckman et al., 2001; Small et al., 2000; Small, Herlitz, et al., 1997), the present findings have to be replicated and extended to cognitive tasks that specifically assesses those functions most strongly affected in preclinical AD such as episodic memory (Jacobs et al., 1995; Linn et al., 1995; Tierney et al., 1996). Second, in order to maximize sample size and the number of predictor variables included, we aggregated data from two follow-up assessments. Because the present analyses are confined to two measurement points, we cannot address the qualitative nature of the decline function in preclinical AD. However, research suggests that decline is most pronounced during the final years preceding diagnosis (Bäckman et al., 2001; Chen et al., 2001; Small et al., 2000). Thus, in examining individual differences in rate of cognitive change, we find that it seems justifiable to focus on this portion of the preclinical phase. Another limitation has to do with the fact that comorbidity was assessed globally in terms of number of diseases. An analysis focusing on disease categories was not considered meaningful, given the low number of diseases within specific categories. As a result, we cannot determine whether there are differences among various disease categories with regard to the influence on rate of cognitive change. A final caveat is that we were unable to determine the extent to which medications given to treat certain illnesses may have affected rate of MMSE change. It is well established that many medications (e.g., anticholinergic drugs, antihypertensives, and digitalis) may have deleterious effects on cognitive performance among older adults (e.g., Blass & Plum, 1983; Foy et al., 1995). Thus, disentangling the influence of medications from those of the diseases themselves on rate of cognitive change in preclinical AD remains an important task for future research.
In summary, the current results indicate that many variables related to cognitive functioning in normal aging as well as to the occurrence of AD play a limited role as precipitating factors during the transition to clinically manifest dementia. However, given that pharmacological and other treatment in AD may profit from targeting persons in a preclinical phase of the disease (Flicker, 1999; Peterson et al., 1999; Schenk et al., 1999), knowledge that rate of cognitive decline in this phase is largely invariant across major subject characteristics should prove useful. Another important implication is that a careful assessment of health conditions may be essential in delaying the rapid progression to clinical dementia during the final portion of the preclinical phase in AD.
. | Incident AD . | . | . | |
---|---|---|---|---|
Variable . | Time 2 (n = 150) . | Time 3 (n = 80) . | All (n = 230) . | |
Age at baseline | ||||
M | 83.38 | 85.20 | 84.01 | |
SD | 4.79 | 3.82 | 4.55 | |
Sex (% female) | 88.00 | 77.50 | 84.30 | |
Years of education | ||||
M | 7.83 | 8.60 | 8.10 | |
SD | 2.51 | 2.60 | 2.56 | |
MMSE at baseline | ||||
M | 24.71 | 26.14 | 25.20 | |
SD | 3.16 | 1.90 | 2.87 | |
MMSE at follow-up | ||||
M | 19.17 | 19.59 | 19.32 | |
SD | 5.32 | 5.18 | 5.27 |
. | Incident AD . | . | . | |
---|---|---|---|---|
Variable . | Time 2 (n = 150) . | Time 3 (n = 80) . | All (n = 230) . | |
Age at baseline | ||||
M | 83.38 | 85.20 | 84.01 | |
SD | 4.79 | 3.82 | 4.55 | |
Sex (% female) | 88.00 | 77.50 | 84.30 | |
Years of education | ||||
M | 7.83 | 8.60 | 8.10 | |
SD | 2.51 | 2.60 | 2.56 | |
MMSE at baseline | ||||
M | 24.71 | 26.14 | 25.20 | |
SD | 3.16 | 1.90 | 2.87 | |
MMSE at follow-up | ||||
M | 19.17 | 19.59 | 19.32 | |
SD | 5.32 | 5.18 | 5.27 |
Notes: For the Time 2 incident AD persons, baseline = Time 1; for the Time 3 incident AD persons, baseline = Time 2. MMSE = Mini-Mental State Examination; AD = Alzheimer's disease.
. | Incident AD . | . | . | |
---|---|---|---|---|
Variable . | Time 2 (n = 150) . | Time 3 (n = 80) . | All (n = 230) . | |
Age at baseline | ||||
M | 83.38 | 85.20 | 84.01 | |
SD | 4.79 | 3.82 | 4.55 | |
Sex (% female) | 88.00 | 77.50 | 84.30 | |
Years of education | ||||
M | 7.83 | 8.60 | 8.10 | |
SD | 2.51 | 2.60 | 2.56 | |
MMSE at baseline | ||||
M | 24.71 | 26.14 | 25.20 | |
SD | 3.16 | 1.90 | 2.87 | |
MMSE at follow-up | ||||
M | 19.17 | 19.59 | 19.32 | |
SD | 5.32 | 5.18 | 5.27 |
. | Incident AD . | . | . | |
---|---|---|---|---|
Variable . | Time 2 (n = 150) . | Time 3 (n = 80) . | All (n = 230) . | |
Age at baseline | ||||
M | 83.38 | 85.20 | 84.01 | |
SD | 4.79 | 3.82 | 4.55 | |
Sex (% female) | 88.00 | 77.50 | 84.30 | |
Years of education | ||||
M | 7.83 | 8.60 | 8.10 | |
SD | 2.51 | 2.60 | 2.56 | |
MMSE at baseline | ||||
M | 24.71 | 26.14 | 25.20 | |
SD | 3.16 | 1.90 | 2.87 | |
MMSE at follow-up | ||||
M | 19.17 | 19.59 | 19.32 | |
SD | 5.32 | 5.18 | 5.27 |
Notes: For the Time 2 incident AD persons, baseline = Time 1; for the Time 3 incident AD persons, baseline = Time 2. MMSE = Mini-Mental State Examination; AD = Alzheimer's disease.
Predictor . | n . | ΔR2 . | ß . | p . |
---|---|---|---|---|
Demographics | 230 | .02 | ||
Age | −.14 | .03 | ||
Sexa | −.03 | .67 | ||
Education | .03 | .60 | ||
Number of diseases | 230 | .08 | ||
Previous diseases | .03 | .59 | ||
Recent diseases | −.30 | <.001 | ||
High blood pressureb | 224 | .00 | ||
Baseline status | .05 | .55 | ||
Change | .06 | .53 | ||
Depressionc | 147 | .01 | ||
Baseline status | −.08 | .43 | ||
Change | .04 | .66 | ||
Vitamin B12 deficiencyd | 126 | .01 | ||
Baseline status | .01 | .90 | ||
Change | .09 | .37 | ||
Genetic factorse | 187 | .00 | ||
APOE status | .02 | .75 | ||
Social network | 230 | .01 | ||
Baseline status | .10 | .15 | ||
Change | .02 | .81 | ||
Substance usef | 142 | .01 | ||
Smoking at baseline | .07 | .37 | ||
Alcohol intake at baseline | .01 | .90 |
Predictor . | n . | ΔR2 . | ß . | p . |
---|---|---|---|---|
Demographics | 230 | .02 | ||
Age | −.14 | .03 | ||
Sexa | −.03 | .67 | ||
Education | .03 | .60 | ||
Number of diseases | 230 | .08 | ||
Previous diseases | .03 | .59 | ||
Recent diseases | −.30 | <.001 | ||
High blood pressureb | 224 | .00 | ||
Baseline status | .05 | .55 | ||
Change | .06 | .53 | ||
Depressionc | 147 | .01 | ||
Baseline status | −.08 | .43 | ||
Change | .04 | .66 | ||
Vitamin B12 deficiencyd | 126 | .01 | ||
Baseline status | .01 | .90 | ||
Change | .09 | .37 | ||
Genetic factorse | 187 | .00 | ||
APOE status | .02 | .75 | ||
Social network | 230 | .01 | ||
Baseline status | .10 | .15 | ||
Change | .02 | .81 | ||
Substance usef | 142 | .01 | ||
Smoking at baseline | .07 | .37 | ||
Alcohol intake at baseline | .01 | .90 |
Notes: All blocks of variables were entered separately. Mini-Mental State Examination (MMSE) at baseline was partialed out by entering it before each block (baseline MMSE: R2 =.16, ß =.40, and p <.001). APOE = apolipoprotein E.
aMale coded as 1; female coded as 2.
bNormal blood pressure coded as 0; high blood pressure coded as 1.
cNot depressed coded as 0; depressed coded as 1.
dNo vitamin deficiency coded as 0; vitamin deficiency coded as 1.
eNo APOE-ϵ4 allele coded as 0; at least one APOE-ϵ4 allele coded as 1.
fNot smoking–not drinking alcohol coded as 0; smoking–drinking alcohol coded as 1.
Predictor . | n . | ΔR2 . | ß . | p . |
---|---|---|---|---|
Demographics | 230 | .02 | ||
Age | −.14 | .03 | ||
Sexa | −.03 | .67 | ||
Education | .03 | .60 | ||
Number of diseases | 230 | .08 | ||
Previous diseases | .03 | .59 | ||
Recent diseases | −.30 | <.001 | ||
High blood pressureb | 224 | .00 | ||
Baseline status | .05 | .55 | ||
Change | .06 | .53 | ||
Depressionc | 147 | .01 | ||
Baseline status | −.08 | .43 | ||
Change | .04 | .66 | ||
Vitamin B12 deficiencyd | 126 | .01 | ||
Baseline status | .01 | .90 | ||
Change | .09 | .37 | ||
Genetic factorse | 187 | .00 | ||
APOE status | .02 | .75 | ||
Social network | 230 | .01 | ||
Baseline status | .10 | .15 | ||
Change | .02 | .81 | ||
Substance usef | 142 | .01 | ||
Smoking at baseline | .07 | .37 | ||
Alcohol intake at baseline | .01 | .90 |
Predictor . | n . | ΔR2 . | ß . | p . |
---|---|---|---|---|
Demographics | 230 | .02 | ||
Age | −.14 | .03 | ||
Sexa | −.03 | .67 | ||
Education | .03 | .60 | ||
Number of diseases | 230 | .08 | ||
Previous diseases | .03 | .59 | ||
Recent diseases | −.30 | <.001 | ||
High blood pressureb | 224 | .00 | ||
Baseline status | .05 | .55 | ||
Change | .06 | .53 | ||
Depressionc | 147 | .01 | ||
Baseline status | −.08 | .43 | ||
Change | .04 | .66 | ||
Vitamin B12 deficiencyd | 126 | .01 | ||
Baseline status | .01 | .90 | ||
Change | .09 | .37 | ||
Genetic factorse | 187 | .00 | ||
APOE status | .02 | .75 | ||
Social network | 230 | .01 | ||
Baseline status | .10 | .15 | ||
Change | .02 | .81 | ||
Substance usef | 142 | .01 | ||
Smoking at baseline | .07 | .37 | ||
Alcohol intake at baseline | .01 | .90 |
Notes: All blocks of variables were entered separately. Mini-Mental State Examination (MMSE) at baseline was partialed out by entering it before each block (baseline MMSE: R2 =.16, ß =.40, and p <.001). APOE = apolipoprotein E.
aMale coded as 1; female coded as 2.
bNormal blood pressure coded as 0; high blood pressure coded as 1.
cNot depressed coded as 0; depressed coded as 1.
dNo vitamin deficiency coded as 0; vitamin deficiency coded as 1.
eNo APOE-ϵ4 allele coded as 0; at least one APOE-ϵ4 allele coded as 1.
fNot smoking–not drinking alcohol coded as 0; smoking–drinking alcohol coded as 1.
Predictor . | R . | ΔR2 . | Cum R2 . | ß . | p . |
---|---|---|---|---|---|
MMSE | .40 | .16 | .16 | .40 | <.001 |
Age entered first | |||||
Age | .42 | .02 | .18 | −.14 | .03 |
Recent diseases | .50 | .07 | .25 | −.27 | <.001 |
Age × Recent Dis. | .50 | .00 | .25 | −.41 | .71 |
Recent dis. entered first | |||||
Recent diseases | .49 | .08 | .24 | −.29 | <.001 |
Age | .50 | .01 | .25 | −.08 | .17 |
Age × Recent Dis. | .50 | .00 | .25 | −.41 | .71 |
Predictor . | R . | ΔR2 . | Cum R2 . | ß . | p . |
---|---|---|---|---|---|
MMSE | .40 | .16 | .16 | .40 | <.001 |
Age entered first | |||||
Age | .42 | .02 | .18 | −.14 | .03 |
Recent diseases | .50 | .07 | .25 | −.27 | <.001 |
Age × Recent Dis. | .50 | .00 | .25 | −.41 | .71 |
Recent dis. entered first | |||||
Recent diseases | .49 | .08 | .24 | −.29 | <.001 |
Age | .50 | .01 | .25 | −.08 | .17 |
Age × Recent Dis. | .50 | .00 | .25 | −.41 | .71 |
Notes: MMSE = Mini-Mental State Examination; dis. = diseases.
Predictor . | R . | ΔR2 . | Cum R2 . | ß . | p . |
---|---|---|---|---|---|
MMSE | .40 | .16 | .16 | .40 | <.001 |
Age entered first | |||||
Age | .42 | .02 | .18 | −.14 | .03 |
Recent diseases | .50 | .07 | .25 | −.27 | <.001 |
Age × Recent Dis. | .50 | .00 | .25 | −.41 | .71 |
Recent dis. entered first | |||||
Recent diseases | .49 | .08 | .24 | −.29 | <.001 |
Age | .50 | .01 | .25 | −.08 | .17 |
Age × Recent Dis. | .50 | .00 | .25 | −.41 | .71 |
Predictor . | R . | ΔR2 . | Cum R2 . | ß . | p . |
---|---|---|---|---|---|
MMSE | .40 | .16 | .16 | .40 | <.001 |
Age entered first | |||||
Age | .42 | .02 | .18 | −.14 | .03 |
Recent diseases | .50 | .07 | .25 | −.27 | <.001 |
Age × Recent Dis. | .50 | .00 | .25 | −.41 | .71 |
Recent dis. entered first | |||||
Recent diseases | .49 | .08 | .24 | −.29 | <.001 |
Age | .50 | .01 | .25 | −.08 | .17 |
Age × Recent Dis. | .50 | .00 | .25 | −.41 | .71 |
Notes: MMSE = Mini-Mental State Examination; dis. = diseases.
This research was supported by grants from the Swedish Science Council to Lars Bäckman, the Swedish Council for Working Life and Social Research to Lars Bäckman and Laura Fratiglioni, and the Swedish Science Council to Laura Fratiglioni.
The cooperation of our colleagues in the Kungsholmen Project in data collection and management is gratefully acknowledged.
References
Agüero-Torres, H., Fratiglioni, L., Guo, Z., Viitanen, M., Winblad, B.
American Psychiatric Association.
Andel, R., Gatz, M., Pedersen, N. L., Reynolds, C. A., Johansson, B., Berg, S.
Bäckman, L., Forsell, Y.
Bäckman, L., Hassing, L., Forsell, Y., Viitanen, M.
Bäckman, L., Hill, R. D., Herlitz, A., Fratiglioni, L., Winblad, B.
Bäckman, L., Small, B. J., Fratiglioni, L.
Bäckman, L., Small, B. J., Wahlin. Å., Larsson, M.
Basun, H., Corder, E. H., Guo, Z., Lannfelt, L., Corder, L. S., Manton, K. G, et al.
Blass, J. P., Plum, F.
Bondi, M. W., Salmon, D. P., Galasko, D., Thomas, R. G., Thal, L. J.
Buckwalter, J. G., Sobel, E., Dunn, M. E., Diaz, M. M., Henderson, V. W.
Burke, W. J., McLaughlin, G. T., Chung, H. D., Gillespie, K. N., Grossberg, G. T., Luque, F. A., et al.
Burns, A., Jacoby, R., Levy, R.
Chandra, V., Bharucha, N. E., Schoenberg, B. S.
Chen, P., Ratcliff, G., Belle, S. H., Cauley, J. A., DeKosky, S. T., Ganguli, M.
Clark, C. M., Sheppard, L., Fillenbaum, G. G., Galasko, D., Morris, J. C., Koss, E., et al.
Devanand, D. P., Sano, M., Tang, M. X., Taylor, S., Gurland, B. J., Wilder, D., et al.
Elias, M. F., Beiser, A., Wolf, P. A., Au, R., White, R. F., D'Agostino, R. B.
Elias, M. F., Wolf, P. A., D'Agostino, R. B., Cobb, J., White, L.
Fabrigoule, C., Rouch, I., Taberly, T., Letenneur, L., Commenges, D., Mazaux, J. M., et al.
Fahlander, K., Berger, A. K., Wahlin, Å., Bäckman, L.
Fahlander, K., Wahlin, Å., Fastbom, J., Grut, M., Forsell, Y., Hill, R. D., et al.
Flicker, L.
Folstein, M. F., Folstein, S. E., McHugh, P. R.
Fox, N. C., Warrington, E. K., Sieffer, A. L., Agnew, S. K., Rossor, M. N.
Foy, A., O'Connell, D., Henry, D., Kelly, J., Cooking, S., Halliday, J.
Fratiglioni, L., Grut, M., Forsell, Y., Viitanen, M., Grafström, M., Holmén, K., et al.
Fratiglioni, L., Grut, M., Forsell, Y., Viitanen, M., Winblad, B.
Fratiglioni, L., Launer, L., Andersen, K., Breteler, M. M. B., Copeland, J. R. M., Dartigues, J. F., et al.
Fratiglioni, L., Rocca, W.
Fratiglioni, L., Viitanen, M., von Strauss, E., Tonadoanti, V., Herlitz, A., Winblad, B.
Fratiglioni, L., Wang, H., Ericsson, K., Maytan, M., Winblad, B.
Gatz, M., Svedberg, P., Pedersen, N. L., Mortimer, J. A., Berg, S., Johansson, B.
Guo, Z., Viitanen, M., Winblad, B., Fratiglioni, L.
Hall, C. B., Lipton, R. B., Sliwinski, M., Stewart, W. F.
Hall, C. B., Ying, J., Kuo, L., Sliwinski, M., Buschke, H., Katz, M., et al.
Halpern, D. F., LaMay, M. L.
Hassing, L., Wahlin, Å., Winblad, B., Bäckman, L.
Hill, R. D., Bäckman, L.
Hill, R. D., Bäckman, L., Wahlin, Å., Winblad, B.
Hill, R. D., Wahlin, Å., Winblad, B., Bäckman, L.
Holstein, J., Chatellier, G., Piette, F., Moulias, R.
Huff, F. J., Growdon, J. H., Corkin, S., Rosen, T. J.
Hultsch, D. F., Hertzog, C., Dixon, R. A., Small, B. J.
Jacobs, D. M., Sano, M., Dooneif, G., Marder, K., Bell, K. L., Stern, Y.
Katzman, R., Brown, T., Thal, L. J., Fuld, P., Aronson, M., Butters, N., et al.
Kindermann, S. S., Brown, G. G.
La Rue, A., Jarvik, L. F.
Letenneur, L., Launer, L. J., Andersen, K., Dewey, M. E., Ott, A., Copeland, J. R. M., et al.
Linn, R. T., Wolf, P. A., Bachman, D. L., Knoefel, J. E., Cobb, J. L., Belanger, A. J., et al.
Masur, D. M., Sliwinski, M., Lipton, R. B., Blau, A. D., Crystal, H. A.
McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., Stadlan, E. M.
Nielsen, H., Lolk, A., Andersen, K., Andersen, J., Kragh-Sorensen, P.
Nilsson, L.-G., Bäckman, L., Erngrund, K., Nyberg, L., Adolfsson, R., Bucht, G., et al.
Petersen, R. C., Smith, G. E., Ivnik, R. J., Tangalos, E. G., Schaid, D. J., Thibodeau, S. N., et al.
Petersen, R. C., Smith, G. E., Waring, S. C., Ivnik, R. J., Tangalos, E., Kokmen, E.
Qiu, C., Bäckman, L., Winblad, B., Agüero-Torres, H., Fratiglioni, L.
Rubin, E. H., Storandt, M., Miller, J. P., Grant, E. A., Kinscherf, D. A., Morris, J. C., et al.
Schenk, D., Barbour, R., Dunn, W., Gordon, G., Grajeda, H., Guido, T., et al.
Skoog, I., Lernfelt, B., Landahl, S., Palmertz, B., Andreasson, L. A., Nilsson, L., et al.
Sliwinski, M., Lipton, R. B., Buschke, H., Stewart, W.
Small, B. J., Basun, H., Bäckman, L.
Small, B. J., Fratiglioni, L., Viitanen, M., Winblad, B., Bäckman, L.
Small, B. J., Herlitz, A., Fratiglioni, L., Almkvist, O., Bäckman, L.
Small, B. J., Viitanen, M., Bäckman, L.
Small, B. J., Viitanen, M., Winblad, B., Bäckman, L.
Teri, L., Hughes, J. P., Larson, E. B.
Tierney, M. C., Szalai, J. P., Snow, W. G., Fisher, R. H., Nores, A., Nadon, G., et al.
Wahlin, Å., Bäckman, L., Mäntylä, T., Herlitz, A., Viitanen, M., Winblad, B.
Wang, H.-X., Fratiglioni, L., Frisoni, G. B., Viitanen, M., Winblad, B.
Wang, H.-X., Wahlin, Å., Basun, H., Fastbom, J., Winblad, B., Fratiglioni, L.
Wilson, R. S., Beckett, L. A., Bennett, D. A., Albert, M. S., Evans, D. A.
Wilson, R. S., Gilley, D. W., Bennett, D. A., Beckett, L. A., Evans, D. A.
Wilson, R. S., Schneider, J. A., Barnes, L. L., Beckett, L. A., Aggarwal, N. T., Cochran, E. J., et al.
Wolf-Klein, G. P., Silverstone, F. A., Brod, M. S., Levy, A., Foley, C. J., Termotto, V., et al.