Backgrounds
Frailty is a state of increased vulnerability to poor resolution of homeostasis following a stress. Frailty is commonly observed in older adults, supposed to be a disorder of multiple interrelated physiological systems due to an accelerated decline in physiological reserve with aging [
1]. Frailty increases adverse outcomes including falls, disability, hospitalization, and mortality [
1]. As the world population ages, frailty is an urgent issue.
Exercise-based interventions could delay or improve frailty [
2‐
6], and multicomponent exercise could be especially effective [
7,
8]. Early diagnosis is essential to intervene in frail patients and reduce adverse events.
The Cardiovascular Health Study (CHS) criteria by Fried et al. define frailty as having three or more of the following phenotypes: unintentional weight loss, self-reported exhaustion, weakness, slow walking speed, and low physical activity [
9]. In a Japanese version of the CHS (J-CHS) criteria, the phenotypes are (i) Shrinking: “Have you unintentionally lost 2 or more kg in the past 6 months?” (yes = 1); (ii) Weakness: grip strength < 28 kg in men or < 18 kg in women (yes = 1); (iii) “In the past 2 weeks, have you felt tired without a reason?” (yes = 1); (iv) Gait speed < 1.0 m/s (yes = 1); and (v) “Do you engage in moderate levels of physical exercise or sports aimed at health?” and “Do you engage in low levels of physical exercise aimed at health?” (no to both questions = 1): frailty, prefrailty and robust were defined as having 3–5, 1–2, and 0 points, respectively [
10]. The criteria are supposed to be the standard but require a grip strength tester and a 4–6 m course to measure grip strength and walking speed, consuming time to diagnose.
Diagnosis of frailty using questionnaires has been attempted. One is the Kihon Checklist (KCL), a self-reported questionnaire consisting of 25 items to screen the health and life status of older adults. The English version has been established, and all the items are described elsewhere [
11]. When a KCL score of 4 to 7 points is diagnosed as prefrail and a KCL score of 8 or higher is diagnosed as frail, the best sensitivity and specificity are achieved, and the usefulness of KCL has been validated based on the frailty status diagnosed by the J-CHS criteria [
12]. However, it takes time to complete the 25 items.
The Questionnaire for Medical Checkup of Old-Old (QMCOO) was established by the Ministry of Health, Labour and Welfare in Japan and has been officially used in the medical checkup of older adults in Japan. The QMCOO is self-reported by older adults. The QMCOO is aimed to assess the general health status of older adults, having 15 questions about 10 domains: health condition, mental health, eating behavior, oral function, body weight loss, physical function and falls, cognitive function, smoking, social participation, and social support. All the items of the QMCOO are described elsewhere [
13]. It has been decided that the QMCOO will be used as a platform for frailty checkups for older adults in Japan. However, the QMCOO is not intended to diagnose frailty and no diagnostic criteria using the QMCOO have been established.
The QMCOO has seven questions in common with the KCL (Q4: Do you have any difficulties eating tough foods compared to 6 months ago?; Q5: Have you choked on your tea or soup recently?; Q6: Have you lost 2 kg or more in the past 6 months?; Q8: Have you experienced a fall in the past year?; Q10: Do your family or your friends point out your memory loss? e.g. “You ask the same question over and over again.”; Q11: Do you find yourself not knowing today’s date?; Q13: Do you go out at least once a week?). The QMCOO has several other questions that are not identical but similar to those in the KCL. The QMCOO has fewer items than the KCL, taking less time and burden to complete for older adults. Since the usefulness of the KCL in diagnosing frailty has been validated, the QMCOO could be used to assess frailty, but the evidence is currently insufficient.
In a previous cross-sectional study, we diagnosed frailty in community-dwelling older adults using the QMCOO. The cutoff value of 3/4 points was determined to maximize the Youden index; sensitivity, specificity, and accuracy were 76.3%, 88.1%, and 86.1%, respectively [
14]. However, the number of participants in the study was 223, which is relatively small. To diagnose frailty at the same time as medical checkups using QMCOO would be useful for early intervention, and the cutoff should be validated in another larger cohort for its widespread use. In the present study, therefore, we regarded the participants as the derivation cohort and aimed to validate the cutoff in a newly established validation cohort, establish the QMCOO as a screening tool, and increase options for diagnosing frailty.
We also diagnosed robust and prefrail based on the KCL score and attempted to determine the cutoff for diagnosing prefrail using the QMCOO.
Methods
Study design and the participants
This is a cross-sectional study of community-dwelling older adults. Participants were recruited in the western region of Japan: Yonago City (Tottori Prefecture), Kurayoshi City (Tottori Prefecture), Masuda City (Shimane Prefecture), and Taka Town (Hyogo Prefecture). Candidate participants were those aged 65 or over who had not been certified as requiring support or care by the long-term care insurance. We mailed the candidates a paper survey that included all of the QMCOO and KCL items, and participants answered all of them and returned them. Those who had participated in our previous study [
14] were excluded.
The QMCOO and scoring
The scoring of the QMCOO was conducted as in the previous study [
14]. Each question was scored as 0 or 1, and the total was the score (0–15).
The KCL-based frailty evaluation
Each question of the KCL was scored as 0 or 1, and the total was used as the score (0–25). Based on the previous study [
12], a score of 8 or higher was diagnosed as frail, a score of 4 to 7 as prefrail, and a score of 3 or lower as robust.
Validation of the QMCOO cutoff of 3/4 points
The group of 223 participants analyzed in our previous report [
13] was regarded as the derivation cohort. The group of those who agreed to participate in the present study was set as the validation cohort. The QMCOO cutoff of 3/4 points in our previous report was adopted to the validation cohort, then sensitivity, specificity, and accuracy were calculated. They were also calculated for the “75 years old or over,“ “74 years old or under,“ “males,“ and “females” groups.
The relationship between body weight and the frailty status
We divided the participants into three groups based on the body mass index (BMI): “lean” (BMI < 18.5 kg/m2), “standard” (18.5 ≤ BMI < 25.0 kg/m2), and “obese” (BMI ≥ 25.0 kg/m2). Then we examined the relationship between body weight and the frailty status diagnosed by the QMCOO score. The ratio of frailty was also compared in the male and female groups. Furthermore, the participants were divided into three age groups (74 or under, 75–84, and 85 or over), then the ratio of frailty was compared in the age groups. A logistic regression analysis was performed to examine the relationship between BMI and the frailty status.
Setting a new cutoff for diagnosing prefrail
Robust (the KCL score is three or less) and prefrail (the KCL score is 4–7) participants were extracted from the validation cohort. The cutoff score of the QMCOO for diagnosing prefrail was determined using a receiver operating characteristic (ROC) curve. The point that maximized the Youden index was adopted as the cutoff. Subgroups of age and sex were also tested for QMCOO cutoff values. A logistic regression analysis was performed to examine which of the QMCOO items determined the prefrailty status.
Statistical analysis
A t-test was used to compare the means of two groups, and a one-way analysis of variance (ANOVA) was used to compare the means of multiple groups. Comparisons of proportions were made with a chi-square test. The Pearson test was used to calculate and test the correlation coefficient between KCL and QMCOO.
Logistic regression analysis was used to analyze the factors that affect frailty or prefrailty status. To examine the relationship between body weight and the frailty status, age, sex, and BMI were the explanatory variables, and the frailty status was the outcome. To examine which of the QMCOO items determine the prefrailty status, age, sex, BMI, and QMCOO items were the explanatory variables, and the prefrailty status was the outcome.
P-values < 0.05 were considered significant. All the statistical analyses were performed using R 3.3.3 software (R Foundation for Statistical Computing, Vienna, Austria).
Discussion
In the present study, we have demonstrated the validity of the diagnosis of frailty with a QMCOO score cutoff of 3/4 points.
The QMCOO includes a question about weight loss (Q6), but unlike the KCL, does not include BMI itself. When considering the relationship between BMI and physical function, sarcopenia should also be considered. Sarcopenia is a progressive and generalized skeletal muscle disorder typically observed in older adults, requiring lower appendicular muscle mass or lower muscle quality for diagnosis in the EWGSOP2 criteria [
15]. Lower BMI was related to an increased risk of sarcopenia [
16]. Sarcopenia is associated with functional decline and increased risk of frailty [
17]. Thus it is plausible that lower BMI was associated with frailty in the present study, but higher BMI was also associated with frailty (Fig.
2). The relationship between BMI and the prevalence of frailty is suggested to form a U-shape. A study of British people showed that the BMI range of the lowest prevalence of frailty was 25.0-29.9 kg/m
2 [
18], but in another study, the range was 18.5–24.9 kg/m
2 [
19]. In a study of community-dwelling Japanese older people, the prevalence of frailty was lowest in the BMI range of 21.4–25.7 kg/m
2 [
20]. In the present study, the prevalence of frailty diagnosed using the QMCOO was the lowest in the BMI range of 18.5–25 kg/m
2, compatible with previous findings.
Prefrailty was significantly associated with lower BMI but not with higher BMI (Table
4). The score of Q6 (weight loss) affected the prefrailty status after a logistic regression analysis. Therefore, the experience of body weight loss itself might be the risk of prefrailty, independently of BMI. This suggests that maintaining an appropriate BMI might be important to prevent prefrailty, thus avoiding frailty. However, as little is known about the background of the participants in the study, some diseases (e.g., malignancy, infections, etc.) other than natural aging could result in body weight loss, developing prefrailty or frailty.
We also demonstrated that a QMCOO score cutoff of 2/3 points might help diagnose prefrailty. By picking up patients with a QMCOO score of 3 or more, it might be possible to diagnose and intervene in frailty at an earlier stage. All questions except Q12 (smoking) were significantly associated with the diagnosis (Table
3). In our previous report including 223 participants, only Q1, Q6, Q7, Q10, and Q11 were related to the diagnosis of frailty [
14]. In the present study, the number of participants (n = 6,113) might have sufficient statistical power.
Identifying aspects of frailty and prefrailty is essential to establish their diagnostic methods. Q1 (subjective health status) and Q2 (subjective satisfaction with daily life) are unique to the QMCOO, not included in the J-CHS, the KCL, and the five-item frailty screening index [
21]. The scores of both questions were significantly related to prefrailty status after the multiple linear regression analysis (Table
4). These straightforward questions about subjective health status and satisfaction could be considered to be included in a new questionnaire. Furthermore, other QMCOO items, such as Q6 (body weight loss), Q7 (loss of walking speed), and Q13 (habits of walking), significantly affected prefrailty and frailty status. By picking appropriate items from the QMCOO, a new frailty questionnaire could be developed.
An important limitation of our study is that we had very limited information about the participants. We used only data about the participants’ age, sex, height, body weight, and answers to the questionnaires, but other data were missing. We included age, sex, and BMI as the explanatory variables in the logistic regression analysis but could not consider other confounding factors that might affect the frailty/prefrailty status. Only those who had not been certified as requiring support or care by the long-term care insurance were recruited. However, older adults in general tend to have multiple comorbidities even if they are independent. As stated earlier, sarcopenia and diseases could cause body weight loss and lower gait speed, which are characteristics of frailty/prefrailty. Furthermore, other factors (medication, past medical history, protein and calorie intake, exercise habits, social status, etc.) should also be considered as explanatory variables in the analysis.
Since the QMCOO will be used as a platform for frailty checkups for older adults in Japan, diagnosing frailty at the same time as medical checkups can contribute to medical care for older adults. The QMCOO could be used for screening, then older adults would be formally diagnosed as frail according to the J-CHS, which is supposed to be the standard. However, the present study has limitations. To establish the QMCOO as a diagnostic tool, further studies are needed on older adults with more information about their background. In addition, the KCL was used instead of the J-CHS criteria for the diagnosis of frail and prefrail in the present study, but further research using the J-CHS is needed. Furthermore, this is a cross-sectional study in four limited areas, and the QMCOO should be validated in other regions. Thus by accumulating evidence, the QMCOO might contribute to early diagnosis and intervention of frailty and prefrailty in the future.
Acknowledgements
We thank Mr. Shinya Masuda, Mr. Yoshimasa Taniguchi, Mr. Hiroto Takashita, Ms. Mieko Yamane, and Mr. Shinji Umehara (Columbus Co., Ltd., Tokyo, Japan) for providing the participants’ data and supporting our study. We also thank Ms. Yoko Ishida (Yonago City), Ms. Chizue Kawamoto (Kurayoshi City), Ms. Atsuko Hirahara (Masuda City), and Ms. Tsubasa Shimohara (Taka Town) for recruiting the participants.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.