Relationship Between Non-Cognitive Intrinsic Capacity and Activities of Daily Living According to Alzheimer’s Disease Stage

Subjects, cognitive function assessments, and diagnosis

A total of 6,969 patients aged 65 years or older attended an initial visit at our memory clinic at the National Center for Geriatrics and Gerontology from July 2010 to June 2021. We excluded 572 patients with conditions other than amnestic mild cognitive impairment (aMCI) or AD, such as dementia with Lewy bodies and frontotemporal dementia. Diagnoses were based on findings from the initial medical examination and neuropsychological tests. This study did not include re-examined patients, and there were no duplicate subjects. After the exclusions, 6,397 outpatients were left for analysis in this study: 1,036 with normal cognition (NC), 1,712 with aMCI, 1,837 with mild AD (MMSE≥19) [20], and 1,812 of moderate to severe AD (MMSE≤18). All patients were diagnosed based on detailed neuropsychological tests and examinations as described in our previous report [21]. The diagnosis of MCI was based on the Petersen criteria. Patients with MCI were further categorized as having aMCI if their education-adjusted score on the Wechsler Memory Scale-Revised (WMS-R) Logical Memory II was≥1.5 standard deviation (SD) below the age-adjusted norm. AD was diagnosed based on the U.S. National Institute on Aging-Alzheimer’s Association (NIA-AA) guidelines [21]. Since completing all neuropsychological tests was difficult for AD patients, 3,980 of the 6,397 subjects had data available for all the neuropsychological tests. We grouped the AD patients into those with mild AD and those with moderate or severe AD based on their MMSE score as in previous papers because grouping based on these neuropsychological tests would significantly reduce the number of subjects for group analysis.

IC and ADL assessments

IC was assessed in five domains, namely, cognition, locomotion, sensory, vitality, and psychological. For each IC subdomain, a bivariate scale was transformed to 0 or 1 according to a cutoff, and when an IC domain has two subdomains, their mean was calculated. With each domain assigned a score of 0 to 1, we summed the locomotion, sensory, vitality, and psychological domains, omitting the cognition domain, to derive a composite score for non-cognitive IC ranging from 0 to 4, with higher scores representing greater IC. The cognitive domain was not incorporated into the total IC score used in this study. Only MMSE score was used as a cutoff because detailed neuropsychological tests could not performed for all patients, as noted above. The top 25% of MMSE scores was regarded as relatively good MMSE, and the following cutoffs were set: 30 for NC, 27 for aMCI, 23 for mild AD, and 17 for moderate to severe AD. Locomotion was assessed using Timed Up and Go (TUG) [22], and one leg standing (OLS) tests. A time of 11 s or longer TUG test were considered to be show impairment. The OLS test measures the time that the subject can stand unassisted on one leg with eyes open up to a maximum of 60 s. Each leg was tested, and we used the average of the left and right OLS times as the OLS time. A time of less than 15 s in the OLS test was considered to show impairment [23]. The sensory domain was assessed using yes-or-no questions about the presence of self-reported vision impairment and hearing problems. For vitality, handgrip strength (GS) and body mass index (BMI) were assessed. GS was assessed on the same day using a digital force gauge (ZP 500 N; Imada, Toyohashi, Japan) [24]. GS of < 28 kg for men and < 18 kg was considered weak. BMI < 20 was considered as underweight. Psychological status was evaluated by Geriatric Depression Scale-15 (GDS) [25]. GDS≥5 was classified as having depressed mood.

Functional assessment was done by measuring BADL performance using the Barthel Index (BI) [26] and IADL performance using Lawton’s IADL [27]. Total BI score ranges from 0 (completely dependent) to 100 (completely independent) (Supplementary Table 1). Both 80% and 90% were used of the cutoff for BI based on a study in stroke patients, where the optimal cutoff scores for the BI were 95 formodified Rankin Scale (mRS) 1, 90 for mRS2, and 75 for mRS3 [28]. Lawton’s IADL scale has 8 items, with a summary score ranging from 0 (low function) to 8 (high function) (Supplementary Table 2). From these 8 items, “food preparation”, “housekeeping”, and “laundry” were examined among only women in this study. An IADL score of≤4 of 5 points for men and≤6 of 8 points for women were considered to indicate impairment, corresponding to a Lawton’s IADL score of≤80% [29]. When the sub-items were used as objective variables in regression analysis, a value of zero was used for less than full marks and a value of one was used for full marks for BI and raw score was used for Lawton’s IADL (Supplementary Tables 1 and 2).

Covariates and others

Age, sex, years of education, number of comorbidities, needing assistance to go out or not, and exercise frequency (twice a week or less) were obtained from medication records and interviews with patients and their families. The number of comorbidities were counted, including liver disease, heart disease, cancer, hypertension, lung disease, stroke, dyslipidemia, kidney disease, and diabetes.

Statistical analysis

The Kruskal-Wallis test was used to compare patient characteristics between groups. Chi-square analysis was used to compare the frequency of events in each group. The Kruskal-Wallis test followed by a post hoc test was performed to compare the total non-cognitive IC score across the groups. Logistic and linear regression models were used to test for trends of change in characteristics across AD stages and to determine the correlation between IC and both BADL and IADL performance for each AD stage. Age, sex, educational years, and number of comorbidities were included as covariates in the analysis. Results with MMSE or year of diagnosis added as a covariate are presented as a sub-analysis in the supplemental data. Statistical analysis was performed using SPSS software (version 29.0, SPSS, Chicago). A p-value of less than 0.05 was considered statistically significant.

Characteristics

Patients with a more advanced AD stage were older and more frequently male and had fewer years of education, fewer comorbidities, and a lower frequency of exercise at least twice a week. The proportion of patients who needed assistance to go out and the rates of BI≤80% or≤90% and IADL≤80% increased with more advanced AD stage (Table 1).

IC decline in the course of AD

As shown in Table 1, the total non-cognitive IC score decreased in the aMCI stage through later AD stages compared with the NC group and decreased as the AD stage advanced. The proportions of patients with TUG≥11 s, OLS < 15 s, weak GS, BMI < 20, and GDS≥5 increased with increasing AD stage. However, the rate of having both self-reported vision and hearing impairments decreased.

As shown in Table 2, the aMCI group was significantly more likely than the NC group to have worse TUG (odds ratio [OR], 1.4; 95% confidence interval [CI], 1.2–1.6; p < 0.001), worse OLS (OR, 1.3; 95% CI, 1.1–1.6; p = 0.008), weak GS (OR, 1.4; 95% CI, 1.1–1.7; p = 0.002), and BMI < 20 (OR, 1.3; 95% CI, 1.1–1.6; p = 0.007). However, the aMCI group was significantly less likely to have self-reported vision impairments (OR, 0.8; 95% CI, 0.7–0.9; p = 0.002) and hearing impairments (OR, 0.5; 95% CI, 0.4–0.6; p < 0.001). For patients with a more advanced AD stage, ORs were higher for worse TUG, worse OLS, weak GS, and BMI < 20. All four IC domains were worse among patients with moderate to severe AD compared with the NC group. The OR for vision or hearing impairment decreased for more severe AD.

ADL decline in the course of AD

As shown in Table 2, no BADL item was impaired in the aMCI stage. For mild AD, compared with NC group, the number of patients who did not have a full score on the “feeding”, “bathing”, “stairs”, “dressing”, “bowel control”, and “bladder control” items decreased to 40%, and the number of patients who did not have a full score on the “transfer” and “grooming” items decreased to 30%. In the moderate to severe AD stages, all BADL items were significantly impaired compared with the NC group (70–90% decreased).

In the aMCI stage, IADL was significantly impaired for shopping, using transportation, taking medication, and food preparation in comparison with the NC group. In the mild AD stage, all IADL items except housekeeping were significantly impaired compared with the NC group (60–95% decreased). In the moderate to severe AD stage, all IADL items were significantly impaired compared with the NC group (80–98% decreased).

Correlation between non-cognitive IC and ADL performance

As shown in Table 3, for each 1 point increase in non-cognitive total IC score, patients with NC were 0.3 times as likely to have worse BADL performance (≤80%) (p = 0.03) and 0.4 times as likely to have worse IADL (≤80%) (p < 0.001), patients with aMCI were 0.4 times as likely to have worse BADL performance (≤80%) (p = 0.007) and 0.6 times as likely to have worse IADL performance (≤80%) (p < 0.001), patients with mild AD subjects were 0.5 times as likely to have worse BADL (p = 0.004) and 0.6 times as likely to have worse IADL performance (≤80%) (p < 0.001), and patients with moderate to severe AD were 0.7 times as likely to have worse BADL performance (≤80%) (p = 0.03) and IADL performance (p = 0.01). The results were essentially the same when using 90% as the BADL cutoff (Table 3) and when MMSE or year of diagnosis was included as a covariate (Supplementary Table 3).

Correlation between IC domains and BADL performance

As shown in Table 4, after adjusting for covariates, impaired BADL (≤80%) was more likely in NC subjects with TUG≥11 s (OR, 7.6; 95% CI, 1.7–35.3), vision problems (OR, 5.3; 95% CI, 1.2–23.4), hearing problems (OR, 4.6; 95% CI, 1.0–20.4), or weak grip strength (OR, 5.7 95% CI, 1.2–26.8) than in those without these problems. In the same way, among patients with aMCI, those with MMSE scores in the top 25% (≥27) were 0.3 times as likely to have worse BADL (≤80%) than those with MMSE scores in the bottom 75%. BADL performance (≤80%) was more likely to be worse patients with weak grip strength (OR, 11.7; 95% CI, 1.5–88.7) than in those with stronger grip strength. Patients with mild AD who had TUG≥11 s (OR, 3.3; 95% CI, 1.7–6.7), hearing problem (OR, 2.2; 95% CI, 1.4–3.5), weak grip (OR, 4.9; 95% CI, 1.9–12.5), or GDS≥5 (OR, 2.0; 95% CI, 1.3–3.1) tended to have worse BADL performance (≤80%) than those without these risk factors. Among patients with moderate to severe AD, those with MMSE scores in the top 25% (≥17) were 0.4 times as likely to have worse BADL performance (≤80%) than those with MMSE in the bottom 75%. Patients with TUG≥11 s (OR, 1.7; 95% CI, 1.2–2.5), weak grip strength (OR, 3.6; 95% CI, 1.9–7.0), or GDS≥5 (OR, 1.5; 95% CI, 1.1–1.9) tended to have worse BADL performance (≤80%) than those without these risk factors. When cutoff was set to 90% for worse BADL performance, the number of IC items that were correlated with worse BADL performance increased. However, the trend was the same as at the 80% cutoff. When MMSE or year of diagnosis was included as a covariate, the results were essentially the same (Supplementary Tables 4 and 5).

Correlation between IC domains and IADL performance

As shown in Table 4, after adjusting for covariates, patients with NC who had TUG≥11 s (OR, 3.4; 95% CI, 2.2–5.3), OLS < 15 s (OR, 2.4; 95% CI, 1.4–4.1), vision problems (OR, 1.8; 95% CI, 1.2–2.8), hearing problems (OR, 1.9; 95% CI, 1.2–3.0), or weak grip strength (OR, 2.1 95% CI, 1.3–4.0) were more likely to have worse IADL (≤80%) than those without these problem. In the same way, among patients with aMCI, those with MMSE scores in the top 25% (≥27) were 0.7 times as likely to have worse IADL (≤80%). Those with any risk factors were likely to have worse IADL than those with no risk factors among patients with aMCI. Among patients with mild AD, those with MMSE scores in the top 25% (≥23) were 0.8 times as likely to have worse IADL (≤80%) than those with MMSE in the bottom 75%. Those with any risk factors except for hearing problems tended to have worse IADL than those with no risk factors. Among patients with moderate to severe AD, those with MMSE scores in the top 25% (≥17) 0.4 times as likely to have worse IADL. Patients with TUG≥11 s (OR, 1.6; 95% CI, 1.2–2.2), or weak grip strength (OR, 2.3; 95% CI, 1.5–3.4) tended to have worse IADL (≤80%) than those without these risk factors. When MMSE or year of diagnosis was included as a covariate, the results were essentially the same (Supplementary Tables 4 and 5).

Correlation between non-cognitive IC total score and ADL subdomains

As shown in Table 5, when each BI subitem was set as the objective variable, non-cognitive total IC score was correlated with 4 subitems in the NC group (mobility, dressing, bowel control, and bladder control), 5 subitems in the aMCI group (feeding, mobility, stairs, bowel control, and bladder control), all subitems in the mild AD group, and 6 subitems in the moderate to severe AD group (toilet, mobility, stairs, depression, bowel control, and bladder control). When each IADL subitem was set as the objective variable, total IC score was correlated with 5 subitems in the NC and aMCI groups (shopping, transportation, medication, finances, and food preparation), 3 subitems in the mild AD group (shopping, transportation, and medication), and 3 subitems in the moderate to severe AD group (using phone, shopping, and food preparation). There were fewer patients with NC or moderate to severe AD than patients with aMCI or mild AD. When MMSE or year of diagnosis was included as a covariate, the results were essentially the same (Supplementary Tables 6 and 7).

Non-cognitive IC decline in the course of AD

Non-cognitive total IC score decreased in the course of AD, as seen from the results in Table 1. The score declined even in the aMCI group compared with the NC group. Meanwhile, the locomotion and vitality domains were impaired from the aMCI stage and the proportion of subjects with impairment in the locomotion, sensory, and psychological domains increased with increasing AD stage, as seen from the results in Table 2. However, the rate of having both self-reported vision and hearing impairments decreased as AD stage progressed.

Looking at the individual domains in Table 2, the locomotion domain was impaired from the aMCI stage. Declines in physical function in the locomotion domain have been reported previous in the aMCI stage [21]. Our results are consistent with previous studies and suggest that there is a need to focus on declines in physical function from the early stages of AD.

Our results in Table 2 show that the prevalence of sensory impairment was significantly increased in the aMCI stage compared with the NC group. As the AD stage advanced, the prevalence of having both vision and hearing impairments decreased. The emergence of AD pathology in peripheral and central visual systems (i.e., retinal thinning, contract sensitivity, and pupillary response) is well established in patients with AD. For hearing problems, on the other hand, regional loss of right temporal lobe volume and social isolation in patients with dementia are secondary to hearing impairments [30]. Our results suggest that the patients may gradually lose the ability to report sensory impairments as AD advances. Hearing aid use is reported to have a protective effect against dementia [30]. It is important to take into account the possibility that sensory impairments may be present from the beginning of the disease and to consider appropriate assistance for sensory impairments. It is possible that patients with more severe illness may no longer be able to make outpatients visits. More objective evaluation will be necessary in future study.

For the vitality domain, it is reported that BMI decreases as AD advances and that AD with lower BMI is associated with significant cognitive decline during the course of the disease [31]. In addition, the presence and burden of cerebral amyloid and tau are in vivo biomarkers associated with lower BMI in people with MCI [32]. Poor nutritional status is also reported to increase the Behavioral and Psychological Symptoms of Dementia score among patients with MCI and early-stage AD [33]. Our results suggest that AD patients should be examined for possible undernutrition from the early stages.

In the psychological domain, the prevalence of depression is reported to be about 30% in aMCI and AD, and depression is known to promote further cognitive decline. The prevalence of depression is reported to increase from mild to moderate AD and to decrease in more advanced AD as those with advanced AD cannot describe depression and these patients have increased risk of death from depression [34]. It is possible that the prevalence of depression in patients with aMCI and mild AD in this study was not significantly different from that in the NC group because the patients were relatively healthy and able to attend an outpatient clinic.

ADL decline in the course of AD

Among the patients in this study, BADL performance was not impaired in those with aMCI, as shown in Table 2. A previous review found that BADL performance was slightly impaired in the MCI stage [3], and many previous reports suggest that patients with more advanced AD stage exhibit more difficulties in BADL performance as their cognitive function declines. Our results were for only outpatients with aMCI, not patients with MCI more broadly, because we excluded other conditions such as dementia with Lewy bodies and vascular dementia. Our results show that most aspects of BADL except for toilet use and mobility were impaired in mild AD. It is possible that these patients may be less likely to notice problems in bowel or bladder control or less likely to report it, and the simple act of walking may be relatively preserved.

On the other hand, our results show that IADL performance was impaired from the aMCI stage. IADL performance require more complex neuropsychological processing capacity than BADL performance and therefore are more prone to deterioration due to cognitive or physical decline. In terms of the subdomains of IADL performance, our results here are consistent in some ways with those of previous studies, such as inability to shop and manage medications, and different in other ways, such as preparing food and using transportation [35]. These differences from previous studies may be due to differences in time period and culture. The fact that only women were surveyed for item food preparation, housekeeping, and laundry may have influenced the results.

Correlation between non-cognitive IC and ADL performance

As shown in Table 3, non-cognitive IC was significantly correlated with ADL performance from aMCI stage across all AD stages. Although the results differ somewhat depending on whether the cutoff for BADL is set at 80% or 90%, our results shown in Table 4 indicate that multiple factors including physical function of the legs, sensory impairments, grip strength, and depression are broadly related to BADL performance from the early stages of AD. The relationship became stronger for mild and moderate to severe AD stages compared with the aMCI stage. Our results suggest that even in later stages of AD, non-cognitive factors are correlated with BADL performance. These results are consistent with those of studies showing that exercise and other interventions were effective in preventing ADL performance decline in advanced AD. The fact that more of our patients had impaired BADL performance in the mild AD group and later stages may have affected the results. The relationship between non-cognitive factors and BADL performance may have been weak at the aMCI stage because BADL performance was still largely preserved. It is also possible that the small number of patients with moderate to severe AD relative to patients with aMCI and mild AD affected the results. As shown in Table 5, when the four IC domains were analyzed in the same way as the total score, the non-cognitive total IC score was related to almost all BADL subitems individually, but the relationship with IC differed by BADL subitem. It is interesting that mobility, bowel control, and bladder control, which are expected to have the heaviest care burden, are relatively strongly related to non-cognitive total IC score in all AD stages. In future studies, we would like to longitudinally evaluate how interventions for non-cognitive IC affect these ADL performance subitems in terms of the heavy caregiving burden.

Many previous reports have suggested that patients with early-stage AD, including MCI, have greater difficulty in IADL performance that requires higher cognitive effort [4]. However, our results suggest that not only higher cognitive function but also a wide range of other factors, such as locomotion, sensory status, vitality, and psychological factors, are involved in IADL performance across all AD stages, particularly from the aMCI stage. As the AD stage advanced, the patients’ range of activities may become narrower if they have decreased non-cognitive IC. Patients in the aMCI stage still engage in a wide range of activities, which may indicate that many of these activities are not possible without preservation of physical and mental status.

When non-cognitive IC was analyzed as the total score (Table 5), it was found to be related to 5 of the 8 IADL subitems in the aMCI stage. Moreover, as AD advanced beyond the aMCI stage, the number of IADL performance subdomains correlated with non-cognitive IC decreased. Although non-cognitive total IC was broadly related to IADL subitems, the specific subitems of using the phone, managing finances, housekeeping, and doing laundry were less related to non-cognitive total IC than the other IADL performance subitems. The fact that food preparation, housekeeping, and laundry were examined in women but not men could be a factor contributing to our results. The fact that many older adults do not so often use the phone or make only routine financial transactions may also have played a role. It is interesting that non-cognitive total IC score was well correlated with items other than housework that are essential to living independently, such as the items on transportation and medication. Patients who cannot adhere to their medication may tend to lose non-cognitive function. Future research on the impact of IADL performance on AD should also focus on these subitems. The small number of patients in the moderate to severe AD group compared with the aMCI group could also be one of the reasons for the results, as could differences in the proportion of patients with impaired IADL performance according to AD stage. Multiple other factors such as comorbidities and social factors should be correlated with decline in IADL performance. These many factors may contribute to the complexity of IC and IADL relationship.

Future considerations regarding the correlation between non-cognitive IC and ADL

A key issue to consider in preventive interventions for ADL performance decline in AD patients is that multiple risks usually co-occur and interact across the stages of AD. As is often the case, a “one-size-fits-all” preventive approach in the course of AD might not be effective. Instead, an approach tailored to the disease course targeting multiple risk factors is likely needed to effectively prevent ADL performance decline in AD even from the aMCI stage [5]. Many previous studies have suggested that these multiple factors are intricately intertwined and interrelated with ADL decline in AD patients. Based on the results of this study, we would like to propose that such intervention methods should be considered in the future, depending on the AD stage and ADL type, and even depending on ADL subitems if possible.

Our study could suggest the following points. 1) Non-cognitive functional decline from an early stage of AD should be noted in addition to cognitive decline, as non-cognitive IC was found to decline from the aMCI stage. 2) Even in advanced AD, multiple interventions for BADL performance may be effective in view of the result that the association between non-cognitive IC and BADL performance was observed even in moderate to severe AD. 3) It might be better to consider multiple interventions for IADL performance from the aMCI stage based on our results that the association between non-cognitive impairment IC and IADL performance was stronger in the aMCI stage than in the later AD stages. 4) To evaluate changes in ADL performance in response to multiple interventions, the type of ADL performance, including subitems, should be considered because as shown in Table 5, the relationship between non-cognitive IC and each subitem varied. Thus, for further studies, it could be clinically relevant to consider that the association between non-cognitive factors that decline from the aMCI stage and ADL performance differs depending on AD stage, ADL type, and ADL subitems.

Limitations

A limitation of this study is that it had a smaller number of patients in the NC and moderate to severe AD groups compared with the aMCI and mild AD groups. The patients in this study were attending a memory clinic; therefore, as is often the case, the number of patients with NC was smaller than the number of patients with dementia. Furthermore, the patients were all outpatients, which means that ADL performance impairment might be underestimated because this study included only patients who were well enough to attend an outpatient clinic, given that the number of comorbidities was observed to decrease as AD advanced. In addition, this study included many cognitively impaired subjects and the number of older adults living alone is increasing in Japan. Although the patients’ families were interviewed, it is unclear to what extent ADL could be accurately accessed by interview. In addition, in this study, only MMSE was used as a cutoff between mild AD and moderate to severe AD. Further studies using other cutoffs such as Clinical Dementia Rating, Functional Assessment Staging Tool for Alzheimer’s disease, and other neuropsychological tests are needed. Lastly, this was a cross-sectional study, which has a risk of ecological fallacy where differences at the group level may not be representative of differences at the individual level. Therefore, longitudinal studies are warranted to observe change in ADL performance over the course of AD.

Conclusion

Most domains of non-cognitive IC were impaired from the aMCI stage. Non-cognitive function decline from the early stage of AD should be noted in addition to cognitive decline. Non-cognitive IC was well correlated with both BADL and IADL in almost all stages of AD from the initial stage. The relationship between non-cognitive IC and BADL performance was stronger in mild AD and later stages than in the earlier stages. Even in advanced AD, multiple interventions for BADL performances may be effective. On the other hand, the relationship between IC and IADL was stronger in the aMCI stage than in the later AD stages. Thus, it may be better to consider multiple interventions for IADL performance from the aMCI stage. The relationship between non-cognitive IC and ADL performance differed depending on AD stage, ADL type, and subitem; thus, more in-depth research on preventing ADL decline may require assessment and intervention according to ADL type, ADL subitem, and AD stage from the initial AD stage.