Article Text

Download PDFPDF

Original research
Association between the frailty index and readmission risk in hospitalised elderly Chinese patients: a retrospective cohort study
  1. Lina Wang,
  2. Xiaolin Zhang,
  3. Xinmin Liu
  1. Geriatric Department, Peking University First Hospital, Beijing, China
  1. Correspondence to Dr Xinmin Liu; lxm2128{at}


Objectives Frailty is a common and important concern of the ageing population. This study examined the association between the frailty index and negative outcomes of hospitalised elderly Chinese patients.

Design Retrospective cohort study.

Setting Geriatrics Department of Peking University First Hospital.

Participants 470 hospitalised elderly patients.

Main outcomes and measures Frailty was measured using a 30-item deficit-accumulation frailty index. The outcomes were the hospitalisation duration and readmission.

Results The frailty index was available for 470 patients: 72 (15.32%) were categorised as robust, 272 (57.87%) as prefrail and 126 (26.81%) as frail. The frail group had a longer hospital stay than the robust and prefrail groups. After adjustment for age, sex and cause of hospitalisation at baseline, frailty remained a strong independent risk factor for all-cause readmission and cardiocerebrovascular disease readmission (HR 2.41, 95% CI 1.49 to 3.91, p<0.001; HR 4.92, 95% CI 1.47 to 6.31, p<0.001, respectively).

Conclusions The frailty index predicted a longer length of stay and higher all-cause and cardiocerebrovascular disease readmission risk in hospitalised elderly patients.

  • Aged
  • Hospitalization
  • Risk management
  • Frailty

Data availability statement

Data are available on reasonable request.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • This was the first study that applied the frailty index (FI) to evaluate the health status and readmission risk of hospitalised Chinese patients.

  • The FI evaluated frailty from multiple dimensions (physical measurement, activities of daily living, function disorder and medical history).

  • The population included in the study was hospitalised patients admitted to the geriatric department of a large tertiary hospital rather than the community population.

  • This was a single-centre retrospective cohort study, and the sample size was relatively small.


Frailty is one of the most prevalent geriatric syndromes1 and is expected to rise in the ageing population. For elderly people in the community, frailty might be a useful predictor of negative health outcomes, including mortality, rehospitalisation, loss of activities of daily living (ADL) and physical limitations.2 Therefore, identifying the frailty status is necessary and might improve the prognosis of elderly people. Comprehensive geriatric assessment is beneficial for managing frail elderly patients, which increases the possibility of remaining alive and returning home after emergency admission to the hospital.3

There is no gold standard for measuring frailty in hospitalised elderly patients. Frailty in older people is typically evaluated using the frailty phenotype,4 the Clinical Frailty Scale (CFS),5 the Hospital Frailty Risk Score (HFRS)6 and the frailty index (FI). The frailty phenotype is a well-validated and frequently used measurement in research. However, its limited scope7 and lack of frailty grading have been criticised. The CFS was validated to stratify older adults according to the frailty status level and reliably predict poor outcomes in critically ill patients8; however, the assessment can be subjective. Neither the frailty phenotype nor the CFS can be conducted retrospectively, which limits their application in clinical studies. The HFRS was created for hospitalised patients based on the 10th revision of the International Statistical Classification of Diseases and Related Health Problems coding system and involves a complicated evaluation process.9 The concept of the FI is that frailty is a state caused by the accumulation of health deficits during life.10 The FI aims to assess the health status of patients from multiple dimensions, such as comorbidities, function independence, physiological indicators and polypharmacy,11 which provide quantitative and relatively easy variables for implementation.12 Numerous epidemiological studies involving non-hospitalised older people demonstrated that the FI is a valid measure of frailty.2 13 An increased FI was independently associated with negative health outcomes in community settings, even in younger individuals (age <65 years).14

Frailty prevalence is high in hospitalised elderly patients.15 Measured by the FI, frailty predicted multiple adverse outcomes, including in-hospital death, longer hospital stay duration and worsening basic ADL.16–19 However, Soh et al reported that the FI and CFS poorly predicted mortality in geriatric rehabilitation in-patients,20 inconsistent with previous studies. Except for short-term prognosis, frail elderly patients were more likely to require hospital readmission with acute illness.21 22 This study aimed to describe the health status of hospitalised elderly patients via the FI and examine the predictive value of the FI in rehospitalisation, mortality and length of stay.


Study design and population

This study was a retrospective cohort study that involved patients aged ≥65 years hospitalised at the Geriatrics Department of Peking University First Hospital from 1 January 2018 to 31 December 2018. The Geriatrics Department mainly provides medical care to elderly patients hospitalised due to internal diseases. Participants without missing values for the FI component items were included in the analysis.

Baseline assessment

The study clinicians evaluated and collected the participants’ medical comorbidities, vital signs, body mass index (BMI), laboratory test results (including serum albumin and neutrophil–lymphocyte ratio), medication history and self-reported functional status retrospectively. Self-reported functional status was assessed using a questionnaire that included seven ADLs (feeding, bathing, bowels, bladder, transferring, ambulating and walking up and down a flight of stairs). The study clinicians reviewed the participants’ medical records during hospitalisation to extract other physical dysfunction information (defecation habits, sleep disorder, mental status, weight loss and cognitive impairment). Cognitive impairment was defined as a dementia diagnosis.

The FI was constructed following a standard procedure.11 Health-related deficits were included if they met the following criteria: the deficit involved multiple body systems and a range of physiological areas, the prevalence of the deficit generally increased with age, the deficit was not nearly universal in middle age and the deficit had a baseline prevalence of ≥0.5% in the population.23 Thirty items (table 1) were included: medical history (16 comorbidities), function disorders (constipation or diarrhoea for >3 months, sleep disorder, anxiety and depression, and cognitive impairment), ADL assessment (assessed by nurses at admission), physical measurement (weight loss >2.5 kg in the past year, BMI, serum albumin level, neutrophil–lymphocyte ratio) and polypharmacy (≥5 prescription drugs). Each deficit was mapped into the 0–1 interval, with 0 indicating the absence of a deficit (the healthiest state) and 1 indicating the maximal expression of the deficit (the unhealthiest state). The patients were classified as robust (≤0.1), prefrail (0.1–0.25) or frail (≥0.25) based on the FI (range: 0–1).23–25

Table 1

List of 30 variables included in the frailty index

Follow-up and outcomes

The duration of hospital stay was calculated as the number of days between the dates of admission and discharge. All patients were followed up after discharge to determine the time of the first readmission. The participants’ readmission was tracked using selected records and telephone follow-ups accessed via the health system’s electronic medical records. Readmission time was defined as the interval between discharge and the next readmission date or the end of follow-up (30 March 2022). The causes of readmission were based on the admitting diagnosis and included cardiocerebrovascular diseases, respiratory diseases, digestive diseases and other causes.

Statistical methods

The patients’ baseline characteristics were presented in three frailty status categories (robust, prefrail and frail). The χ2 test or Fisher’s exact test was used for categorical variables. The baseline prevalence of the prefrail and frail statuses was calculated using multinomial logistic regression, adjusting for age, sex and cause of hospitalisation at baseline. Differences between groups were analysed for continuous variables using the Kruskal-Wallis test. The associations between the FI and all-cause readmission and disease-specific readmission were determined using a Cox proportional hazards model. The cumulative risk curves of re-admission for all-cause and cardiocerebrovascular disease were described. Multivariable models were adjusted for age, sex and cause of hospitalisation at baseline. Variables were determined in the multivariable model using forward stepwise regression. A two-sided p<0.05 was considered statistically significant. The statistical analysis was performed using SPSS V.26.

Patient and public involvement

There was no patient or public involvement in the design, conduct, reporting or dissemination plans of our study.


Data from 494 hospitalised patients aged >65 years from 1 January 2018 to 31 December 2018 were collected. Twenty-four patients were excluded due to missing data. The baseline data of 470 patients were obtained. Three patients died during hospitalisation, and all of them were in the frail group. Eventually, 467 patients were followed up after discharge (figure 1). The median follow-up time was 43.99 months. Table 2 presents the participants’ baseline characteristics according to frailty status. The mean participant age was 78.2±7.9 years. Among them, 72 participants (15.32%) were categorised as robust, 272 (57.87%) as prefrail group and 126 (26.81%) as frail (table 2). The maximum FI score was 0.633 in males and 0.550 in females. The mean FI for all patients was 0.197 (SD 0.092), while the mean FI values for the male and female patients were 0.197 (SD 0.090) and 0.198 (SD 0.099), respectively.

Table 2

Baseline characteristics of the patients by frailty status

Frailty prevalence increased with age and ADL difficulty. Frailty and prefrailty prevalences among the participants were not significantly different based on tobacco smoking status and BMI (figure 2).

Figure 2

The prevalence of frailty according to baseline characteristics. Frailty prevalence was adjusted for age, sex and cause of hospitalisation, except for prevalence by age group, which was adjusted for sex and cause of hospitalisation at baseline. BMI, body mass index.

The proportions of FI deficits in the frail and prefrail groups were described in five dimensions (medical history, function disorders, physical measurement, ADL and polypharmacy). The proportion of polypharmacy deficits was highest between the two groups (92.06% and 58.46%, respectively). In the frail group, the proportions of physical measurement, medical history, function disorders and ADL deficits were 41.37, 34.28, 19.44 and 13.25%, respectively. The proportion of deficits in the above four dimensions decreased in the prefrail group and were 28.77, 18.59, 8.55 and 3.93%, respectively.

The median (IQR) hospital stay duration was 10 (8–14), 13 (9–17) and 14 (11–22) days for the robust, prefrail and frail groups, respectively. The frail group had a longer hospital stay than the robust and prefrail groups (Padj<0.001, Padj=0.007, respectively). Similarly, the prefrail group had a longer hospital stay than the robust group (Padj=0.007).

During the follow-up, 266 patients were readmitted to the hospital: 135 (50.8%) for cardiocerebrovascular diseases, 36 (13.5%) for respiratory diseases, 39 (14.7%) for digestive diseases and 56 (21.1%) for all other causes. The Kaplan-Meier analysis determined that the median readmission interval was 29.37 months (95% CI 23.09 o 35.65). In the prefrail and frail groups, the median readmission interval was 31.37 months (95% CI 25.71 to 37.02) and 11.83 months (95% CI 8.99 to 14.68), respectively. Twenty-five patients (34.7%) in the robust group were readmitted to the hospital.

The multivariable model was adjusted for age, sex, hospitalisation cause at baseline and frailty status to analyse the readmission risk factors. The multivariate Cox regression analysis demonstrated that age and frailty status were the risk factors for all-cause readmission (p=0.024 and p=0.001, respectively). The prefrail and frail participants had a higher risk of all-cause readmission and cardiocerebrovascular disease readmission than the robust participants (table 3). The overall adjusted HR for readmission per 0.1 increment in the FI was 1.40 (95% CI 1.23 to 1.60, p<0.001). Figure 3A presents the calculated risk curves. A graded increase in the risk of readmission for cardiocerebrovascular disease was observed in the prefrail and frail groups (figure 3B). The corresponding HR for readmission risk was 1.62 (95% CI 1.34 to 1.96, p<0.001) for every 0.1 increment in the FI. No significant difference was observed between the association of the FI and the risk of readmission for respiratory diseases, digestive diseases and other causes.

Table 3

The association of frailty status and readmission

Figure 3

Cumulative risk of readmission according to frailty status. (A) Cumulative risk of all-cause readmission, including cardiocerebrovascular diseases, respiratory diseases, digestive diseases, tumours and other causes. (B) Cumulative risk of readmission for cardiocerebrovascular diseases.

During the follow-up period, 16 patients died after readmission: 1 from the robust group, 5 from the prefrail group and 10 from the frail group.


This study constructed the FI for hospitalised elderly patients and confirmed that frailty is a predictor of prolonged hospitalisation days and readmission in elderly patients. This study was the first to apply the FI to evaluate the health status and readmission risk of hospitalised elderly patients in China. The frail patients had a higher risk of all-cause readmission and cause-specific readmission for cardiocerebrovascular disease. These findings emphasised the importance of assessing frailty to accurately inform patients and their families about the possibility of readmission. Taking measurements to improve frailty status might reduce readmissions and healthcare costs.

Frailty prevalence data in hospitalised elderly patients vary widely and are 27%–80%.26 In the present study, the prevalence of FI-categorised frailty and prefrailty was 26.81% and 57.87%, respectively. Frailty prevalence increased with age, whereas prefrailty decreased with age, which indicated that the frail condition progressed with increasing age. The population selection and frailty measurement can influence the variation in frailty prevalence.27 Most of the patients recruited in this study were admitted to hospital due to acute and subacute internal disease. Furthermore, the mean patient age was lower than that of previous studies,15 28 29 which might explain the lower frailty prevalence as compared with the previous studies. In this study, frailty prevalence increased with ADL difficulty. Compared with the prefrail group, frail patients demonstrated an overall decline in multiple dimensions, including physical measurement, ADL, function disorder and medical history. Interventions such as physical exercise and maintaining good nutritional status might be effective in preventing the transition from prefrail to frail in the elderly.30

In this study, the obvious association of the FI with the hospital stay duration was consistent with that of previous studies.18 19 22 However, a higher frailty score was associated with shorter hospital stay in rehabilitation settings.29 The contradictory findings might be due to the study sites. Estimating the effect of frailty on hospital readmission is challenging. Stuck et al reported that the frailty phenotype and the FI, but not the CFS, were good predictors of readmission to acute care from in-patient rehabilitation.21 Furthermore, the FI was associated with rehospitalisation and all-cause mortality after discharge.31 32 Consistent with these observations, the present study confirmed that frailty, as measured by the FI, increased hospitalisation risk during the long-term follow-up period. However, a prospective cohort study reported that the FI was not significantly related to readmission in elderly in-patients.18 Hospital readmission is not only associated with the patient’s frailty status but also with social factors such as housing instability and social support.33 Moreover, the continuous nature of the FI potentially changes with an individual’s treatment. Previous studies evaluated the frailty status at admission rather than before discharge. Therefore, the relationship between FI and readmission risk should be evaluated dynamically in further research.

Frailty and prefrailty are highly prevalent conditions and are related to adverse health outcomes among the elderly with cardiovascular diseases (CVDs).34 Aida et al used the simplified frailty scale and reported that frailty was associated with an increased risk of mortality or readmission in elderly patients hospitalised for CVD.35 Nghiem et al conducted a large cohort study that used HFRS among 75-year-old patients with CVD and confirmed that frailty increased the risk of mortality, longer duration of stay and healthcare costs. However, a significant influence of frailty on readmission within 30 days of discharge was not observed.36 The present study supported the idea that diagnoses of frailty and prefrailty, according to the FI, were independent predictors of readmission for CVD among elderly patients. Significantly, >50% of the patients enrolled in the present study had cardiocerebrovascular disease. Thus, the research conclusion might be subject to bias. After adjusting for the cause of hospitalisation at baseline, frailty remained a risk factor for cardiocerebrovascular disease readmission. The predictive abilities between the FI and readmission for other causes were not demonstrated and might be limited to the population selection and small samples.

The present study had some limitations. First, this study was a single-centre retrospective study, and internal or external validation was not performed. Second, the number of deaths during the follow-up period was relatively low, and the relationship between FI and mortality was not analysed further. Third, all patients were admitted to tertiary hospitals for internal diseases, which might have limited the representativeness of this study. Finally, the COVID-19 pandemic might have affected readmission negatively.


Frailty assessment, according to the FI, predicted a longer duration of hospital stay and higher all-cause and cardiocerebrovascular disease readmission risk in hospitalised elderly patients. The FI was useful for the risk stratification of elderly patients in clinical practice.

Data availability statement

Data are available on reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Clinical Research Ethics Committee of the Peking University First Hospital, Beijing, China (No. 2022SCI 345) and individual consent for this retrospective analysis was waived.



  • LW and XZ contributed equally.

  • Contributors Conception and design: LW and XZ; (Administrative support: XL; Provision of study materials or patients: LW and XZ; Collection and assembly of data: LW; Data analysis and interpretation: LW and XZ; Manuscript writing: all authors; Final approval of manuscript: all authors. As the guarantor, XL accepts full responsibility for the work and the conduct of the study, had access to the data, and controlled the decision to publish.

  • Funding This work was supported by National Key R&D Program of China (2020YFC2008804), Capital’s Funds for Health Improvement and Research (2022-3-70212) and National High Level Hospital Clinical Research Funding (Interdepartmental Research Project of Peking University First Hospital) 2023IR37.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.