Association Between Timed Up and Go Test and Future Dementia Onset

Abstract

Background

This study evaluated whether baseline results of the Timed Up and Go (TUG) test is associated with future dementia occurrence.

Methods

Using the Korean National Health Insurance Service-National Health Screening Cohort database, we identified 49,283 subjects without a dementia diagnosis who participated in the National Screening Program for Transitional Ages at 66 years of age during 2007–2012. Gait impairment was defined as taking longer than 10 seconds to perform the TUG test. Dementia occurrence was defined by the first prescription for acetylcholinesterase inhibitors or N-Methyl-D-Aspartate receptor antagonist with an International Classification of Diseases 10th Revision (ICD-10) code for dementia (F00, F01, F02, F03, G30, F051, or G311) during 2007–2013. Cox proportional hazard regression models were used to assess the hazard ratios for dementia occurrence according to baseline TUG test results.

Results

Mean follow-up period was 3.8 years. Incidence rates of dementia were 4.6 and 6.8 cases per 1,000 person-years in the normal and impaired TUG groups, respectively. The impaired TUG group showed a higher risk of total dementia incidence (adjusted hazard ratio [aHR], 1.34; 95% confidence interval [95% CI], 1.14–1.57). Subtype analysis showed that the impaired TUG group had a higher risk of Alzheimer’s disease (aHR, 1.26; 95% CI, 1.06–1.51) and vascular dementia (aHR, 1.65; 95% CI, 1.19–2.30).

Conclusions

The TUG test result was associated with future dementia occurrence. More vigilant follow-up and early intervention to prevent dementia would benefit elderly people with impaired TUG test result.

The aging of populations worldwide has led to a marked increase in chronic, disabling diseases; specifically, the prevalence of dementia is steadily increasing. The worldwide prevalence of dementia is estimated to be 5%–7% for the people ≥ 60 years old and is reported to be 80% for people ≥ 90 years old (1). In addition, worldwide costs of dementia are enormous (2) and family members suffer the burden of caring for dementia patients over a long period of time (3).

Despite extensive research efforts, a cure for dementia has yet to be developed. Until now, the best practice has been to identify individuals who are at a high risk of developing dementia and provide preventive strategies. Therefore, predicting the risk of dementia has been regarded as a research priority, although our prediction methods are still lacking (4).

In this regard, the association between physical frailty and cognitive decline is attracting attention. Some previous studies have suggested that physical frailty is a risk factor of cognitive decline (5). The recently developed concept of “cognitive frailty” includes the presence of both physical frailty and cognitive impairment (6). Because of potential for reversibility of frailty, cognitive frailty is attracting attention as an important target of prevention of dementia (7).

As to association between physical frailty and cognitive decline, some recent studies have investigated the relationship between gait impairment, which is an important indicator of physical frailty, and future cognitive decline (8–11). In one prospective cohort study with 10,308 participants in England, decline in physical activity was observed since 9 years before dementia diagnosis (8). In another study with 1,265 elderly people in France, baseline gait speed independently predicted dementia occurrence after 12 years of follow-up (9). These studies suggested possibility of gait speed as predictor of dementia occurrence. However, in these studies, physical performance was assessed with a questionnaire which entails a risk of memory bias, or 6-m walk test which is not commonly used in clinical setting at least until now (9,10). One study suggested that the Timed Up and Go (TUG) test, which is a simple marker of gait impairment, could predict cognitive decline (11); however, the study focused on cognitive decline rather than clinical dementia occurrence. That study also had a small sample population (n = 111) and a short follow-up period (3 years). Therefore, it is necessary to determine whether the TUG test can serve as a simple predictive marker of dementia occurrence in a larger study population over a longer follow-up period.

In this large, nationwide study using health insurance claims data, we investigated the association between baseline TUG test results and future dementia occurrence. Our underlying hypothesis was that the TUG test could be used as a simple predictive marker of dementia occurrence.

Methods

Study Setting

The Korean National Health Insurance (KNHI) service is a mandatory public health insurance system, which provides universal health coverage to all Koreans (12). The KNHI service offers a biennial National Health Screening Program (NHSP) to all members over 40 years of age. The NHSP includes a questionnaire (past medical history, health behavior), anthropometric exam (body mass index, blood pressure), and laboratory test (blood sugar, cholesterol, etc) (13). For all people of 66 years old, KNHI provides a special health screening program called the National Screening Program for Transitional Ages (NSPTA) (14). Besides the routine items of NHSP, NSPTA also includes a questionnaire regarding cognitive function and depression screening, as well as some tests including TUG test that covers common problems of the elderly people such as frailty (14).

Data Source and Study Population

The KNHI database has been widely used in various epidemiological studies (15) and is described in detail elsewhere (16,17). This study used the National Health Insurance Service-National Health Screening Cohort (NHIS-HEALS) database, which comprises 515 000 people. This figure represents 10% of a random selection from the total Korean population aged 40–79 years who participated in the NHSP at least once during the 2002–2003 index year. The NHIS-HEALS contains demographic factors such as age, sex, death date, and results of the NHSP as well as information on the utilization of medical facilities, including the International Classification of Diseases 10th Revision (ICD-10) codes and prescribed medicines from outpatient clinics and hospitalization. As NHI is mandatory social insurance in Korea, attrition from this cohort occurs only by emigration, which is not common at the age over 65.

For this study, the people at 66 years old who participated in the NSPTA program during 2007–2012 were selected. The initial population was 53,000 subjects. Subjects who were diagnosed with dementia or prescribed acetylcholinesterase inhibitors or N-Methyl-D-Aspartate (NMDA) receptor antagonist before the screening day (n = 1,004) and the subjects with missing records involving the TUG test or other covariables (n = 2,713) were excluded (Figure 1). Ultimately, 49,283 subjects were included in the analysis.

This study was approved by the institutional review board of Seoul National University Hospital (IRB No. E-1611-061-807). Consent from individual subjects was waived because we used publicly open, anonymous data.

Variables

Independent variable

The TUG test was conducted on the examination day of NSPTA program at community hospitals. Subjects were timed while they rose from a chair, walked at a comfortable pace to a line on the floor 3 m away, turned and walked back to the chair, and sat down again. The person wore regular footwear and used his/her customary walking aid. Gait impairment was defined as taking longer than 10 seconds to perform the TUG test (18).

Outcome Variable

Dementia occurrence was defined by the first prescription of acetylcholinesterase inhibitors (donepezil, galantamine, rivastigmine) or NMDA receptor antagonist (memantine) (19) with an ICD-10 dementia code (F00, F01, F02, F03, G30, F051, or G311) after the NSPTA examination (20). We also considered subtypes of dementia; Alzheimer’s disease and vascular dementia. Alzheimer’s disease occurrence was defined by the first prescription of an acetylcholinesterase inhibitor or NMDA receptor antagonist with an ICD-10 code for Alzheimer’s disease (F00, G30); vascular dementia occurrence was defined by the first prescription of the same medication with an ICD-10 code for vascular dementia (19,21).

To file expense claims in Korea, the National Health Insurance Reimbursement Criteria must be fulfilled. When prescribing acetylcholinesterase inhibitors or NMDA receptor antagonist, physicians need to document evidence of cognitive dysfunction according to relatively strict criteria: a Mini-Mental State Examination (MMSE) ≤ 26 and either a Clinical Dementia Rating (CDR) ≥ 1 or a Global Deterioration Scale (GDS) ≥ 3. Thus, all the elderly people of this study who were prescribed the medication were considered to meet these criteria. A similar definition of dementia occurrence has been used in other studies that collected data from administrative databases (22).

Covariates

Comorbid conditions, smoking, depressive symptom, and baseline cognitive function were considered to be related to the future development of dementia and were included as covariates in the multivariate analysis. Information on smoking history and relevant comorbid conditions (stroke, hypertension, diabetes mellitus, and hypercholesterolemia) was obtained from the questionnaire administered on the screening day. Depressive symptom was assessed with a three-item questionnaire extracted from the Geriatric Depression Scale (GDS), including items 2, 17, and 22 that relate to loss of interest, feelings of uselessness, and feelings of hopelessness, respectively (23). A negative response to any of the three items was considered evidence of depressive symptom. Baseline cognitive function was assessed with the Prescreening Korean Dementia Screening Questionnaire (KDSQ-P), which has a score ranging from 0 to 10; a higher KDSQ-P score indicates more severe cognitive decline (24). The KDSQ-P has shown a significant correlation with the Korean version of the MMSE with high validity and reliability (24). Patients with total scores ≥ 4 are expected to receive further evaluation for cognitive function.

Statistical Analyses

Data are summarized as numbers with percentages for categorical variables, and mean values with standard deviations (SD) for continuous variables. For each clinical characteristic by TUG test results, we performed a group-wise comparison using a two-tailed Student’s t test for continuous variables and the chi-squared test for categorical variables (Table 1). Statistical differences between the two groups are represented by p values.

The patients were followed up from the day of screening of the NSPTA program to the occurrence of dementia, death, or the last follow-up day (December 31, 2013), whichever came first. A Kaplan–Meier curve was used to assess the relationship between TUG test results and dementia occurrence (Figure 2). Statistical significance of the Kaplan–Meier curve was tested using a log-rank test. Multivariate analysis was conducted using a Cox proportional hazard regression model with other possible risk factors of dementia, including sex, smoking, stroke, hypertension, diabetes mellitus, hypercholesterolemia, depressive symptom, and baseline cognitive function. The associations between the TUG test results and future dementia occurrence are represented as hazard ratios and 95% confidence intervals. Further, considering death before dementia diagnosis as competing risk of dementia occurrence, we performed competing risk analysis using Fine and Gray model adjusting for the same covariates. From this analysis, adjusted subdistribution hazard ratios were obtained. The same analyses were repeated for the subtypes of dementia; Alzheimer’s disease and vascular dementia (Table 2). In addition, using the same models, subgroup analyses were conducted for the subjects who showed normal baseline cognitive function, as indicated by a KDSQ-P score less than 4 (Supplementary Table 1), and elderly people with no history of stroke or Parkinson’s disease (25) (Supplementary Table 2). In Korea, acetylcholinesterase inhibitors and NMDA receptor antagonist are used for vascular dementia (19,21), however this is not yet a unified practice globally. So, additional analysis was performed using vascular dementia definition only by the diagnosis with ICD-10 code, excluding medication prescription (Supplementary Table 3). STATA software (ver. 14.1; STATA Corp., Houston, TX) was used for all statistical analyses.

Results

Study Population Characteristics

Among the 49,283 subjects who were initially included in the analysis, 6,823 showed gait impairment according to their baseline TUG test. The percentage of women was higher in the impaired TUG group compared with the normal TUG group (57.6% vs 49.4%). The rates of stroke (23.7% vs 15.6%), hypertension (44.4% vs 42.4%), diabetes mellitus (30.5% vs 24.2%), and hypercholesterolemia (25.3% vs 19.3%) were higher in the impaired TUG group than in the normal TUG group. The mean cognitive impairment score, as assessed by the screening questionnaire, was higher in the impaired TUG group (1.8 vs 1.5). The distribution of all the above factors showed significant differences between the two groups (p < .05; Table 1).

Risk of Dementia Occurrence

The mean follow-up period was 3.8 years. Incidence rates of dementia were 4.7 and 6.9 cases per 1,000 person-years in the normal and impaired TUG groups, respectively. The impaired TUG group showed a higher risk of total dementia occurrence than normal TUG group (adjusted hazard ratio [aHR], 1.34; 95% confidence interval [95% CI], 1.14–1.57). Dementia subtype analyses showed that the impaired TUG group had a higher risk for Alzheimer’s disease (aHR, 1.26; 95% CI, 1.06–1.51) and vascular dementia (aHR, 1.65; 95% CI, 1.19–2.30). The competing risk analysis showed similar results (Table 2).

Also for the subjects with normal baseline cognitive function, the impaired TUG group showed a higher risk of total dementia occurrence (aHR, 1.33; 95% CI, 1.10–1.62); it showed a marginally significant increase in the risk of Alzheimer’s disease (aHR, 1.24; 95% CI, 0.99–1.55) or vascular dementia (aHR, 1.51; 95% CI, 0.99–2.31) (Supplementary Table 1). For the elderly people without a history of a disabling disease (stroke or Parkinson’s disease), the impaired TUG group showed a higher risk of total dementia incidence (aHR, 1.26; 95% CI, 1.05–1.50) but not a significant increased risk of Alzheimer’s disease (aHR, 1.22; 95% CI, 0.99–1.48) or vascular dementia (aHR, 1.48; 95% CI, 1.00–2.18) (Supplementary Table 2). When vascular dementia was defined only by ICD-code, impaired TUG group also showed a higher risk of vascular dementia occurrence (aHR, 1.57; 95% CI, 1.22–2.01) (Supplementary Table 3).

Discussion

To the best of our knowledge, this is the first study that showed gait impairment measured by TUG test could predict future clinical dementia in the elderly population. Strengths of this study include use of a large, nationally representative homogenous sample of the elderly population in terms of age (66 years old), and independent examination of dementia occurrence in a clinical setting rather than in a research setting, eliminating potential bias from the investigator. Therefore, our study can be considered a confirmation of the findings from previous small-scale studies conducted in research settings.

To date, many researchers are seeking simple ways to predict dementia occurrence. The TUG test is a simple, much studied test (18,26), and the test protocol is well established. The TUG test takes less than a minute and only requires a 3-m distance and a normal chair. From previous studies, the TUG test predicted fall risks (27), deterioration in activities of daily living (28), and nursing home placement (29). The TUG test also predicted dementia occurrence in this large, longitudinal study.

The mechanisms linking gait impairment and cognitive decline are not well understood. One possibility is the sharing of common risk factors between gait impairment and cognitive decline, especially cardiovascular risk factors, nutrition, and depressive symptom. Cardiovascular risk factors may contribute to gait impairment through impaired blood flow to skeletal muscles (30). It is also known that cardiovascular risk factors play an important role in developing both vascular dementia and Alzheimer’s disease (31). Regarding nutrition, adherence to a Mediterranean diet has been linked to both better physical performance (32) and cognitive function (33). Depression is also a well-established risk factor of both decreased physical performance and cognitive decline (34). In addition, some recent studies have suggested that brain pathophysiologies, such as neurofibrillary tangles and plaques, are common mechanisms for physical frailty and Alzheimer’s disease (35). Such brain deposition might result in gait impairment before clinical dementia occurs.

Our results are robust, as shown by consistent findings, regardless of the covariate adjustment and application of different exclusion criteria. Especially, the subgroup analysis showed normal baseline cognitive function (defined as a KDSQ-P score < 4) with gait impairment in TUG test is a predictive marker of future dementia occurrence (Supplementary Table 1). Furthermore, the self-reporting questionnaire for activities of daily living did not significantly predict dementia (HR, 1.22; 95% CI, 0.93–1.58 and aHR, 1.22; 95% CI, 0.93–1.59; not shown on tables). This result further emphasizes the usefulness of the TUG test for predicting dementia.

Impairment on TUG test result increased the risk of future dementia by 1.34-fold. The risk was comparable or higher than other known risk factors of dementia (female sex: aHR, 1.57; 95% CI, 1.31–1.87, baseline cognitive function: aHR, 1.14; 95% CI, 1.11–1.17; not shown on tables). This suggests that physicians should be more vigilant in looking for signs of cognitive decline if a patient shows gait impairment. If patients show impairment on TUG test results, the physician may perform cognitive tests regularly in order to detect cognitive impairment earlier. This would be valuable for preventing dementia and its adverse consequences, as early detection and management can delay the status of dependency (36). Physicians can also provide patients and family members with strategies for preventing dementia (37).

This study does have some limitations. First, our definition of dementia occurrence is not based on strict criteria using in-depth cognitive testing. Instead, we used a clinical diagnosis and medication prescription from claims data; our method does, however, reflect real clinical practice and is free from the bias of nonblinding. Second, we could not include several well-known risk factors for dementia in the covariate analysis such as the presence of the APOE4 gene and education level. However, it is unlikely that these factors significantly affect the association between gait impairment and dementia.

In conclusion, impairment on TUG test result increased the risk of future dementia occurrence. This suggests that the TUG test might be a useful predictive marker of dementia occurrence. More vigilant follow-up and early intervention to prevent cognitive decline would benefit elderly people with signs of gait impairment.

Supplementary Material

Supplementary data is available at The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences online.

Funding

This study was supported by research grants from Hanmi Pharmaceutical Co.Ltd (grant number NHIS-2016-2-267).

Conflicts of Interest

None reported.

Acknowledgments

The authors appreciate the advice and assistance of Functional Assessment Committee for Elderly (FACE) Committee, The Korean Geriatrics Society.

References