Acute hospitalisation often causes a certain degree of damage to the abilities of daily living of elderly persons. The total number of admitted patients who were baseline-independent was 10,963 in the study period. Among them, 869 cases were ADL-dependent at discharge because of the acute illness, accounting for 7.9% of the total number of older patients discharged from hospital. This is similar to the results of the study conducted by Volpato et al. , indicating that acute hospitalisation can lead to ADL dependence in a considerably large number of independent elderly.
From the observation of ADL status, the proportion of older patients who were ADL dependent at 6 months after discharge was 43.3%, which was higher than that identified by Chen et al. . Three-fifths of the patients included in Chen’s study were from surgical wards, while the patients in this study were all from internal medicine wards. Differences in diseases and treatments between internal and surgical departments may result in different recovery rates after discharge. In addition, Chen et al. excluded patients with cognitive impairment. But they were included in this study conversely. The adverse effects of cognitive impairment on the recovery of ADL after discharge have been reported in many studies [8, 19-21].
In this study, the rate of ADL-independent older patients at 6 months after discharge was 56.7% (i.e. proportion that returned to baseline), which was higher than the results of Boyd et al. . The reason may be that Boyd et al. included older patients over 70 years old and patients were dependent at baseline, which is different from the present his study. Previous studies have shown that patients that are ADL-dependent at baseline may tend to be stable because of a long dependency time, which makes it less likely to restore ADL function [1, 3, 5, 18, 22, 23]
Similar to the results of previous studies [5, 7, 24], the ADL of hospitalised ADL-dependent older patients gradually recovered from admission to 6 months after discharge and the recovery rate was the highest at 1 month after discharge. From Figure 2, the ADL level also increased to a certain extent from admission to discharge. This may be related to recovery from the acute illness, which is consistent with the results of Mudge et al. .
As for the factors related to functional recovery after discharge, previous studies had shown that age [7, 26, 27], cognition [8, 19-21], depression, albumin, and IADL at 2 weeks before admission affected ADL in older patients. Age and cognition were confirmed again in this study. However, some studies reported that age was not the most direct factor affecting ADL [28, 29], but these mainly involved short-term follow-ups. Hardy et al.  found that the effect of age on ADL was significant for a long time after becoming dependent, namely 6-12 months or longer after discharge. And older patients who had recovered independence at an early stage were prone to ADL dependence again over a long period of time. This is consistent with the results of the present study.
ADL score at discharge and post-discharge residence were predictors of ADL recovery in elderly patients at 6 months after discharge, which had not been reported in previous studies. The higher the ADL score of older patients at discharge, the greater the probability of ADL independence recovery at 6 months after discharge. Although previous studies had confirmed the correlation between physical function at discharge and prognosis [8, 30], ADL had not been used as an index to evaluate physical function, and Barthel Index had not been used to evaluate ADL. This result suggests that, first of all, clinicians should pay attention to the ADL recovery of older patients during hospitalisation. Despite the experience of functional loss, those whose function improved during hospitalisation were 2.3-2.9 times more likely to recover than those who continued to decline . In addition, ADL evaluation should be considered as part of discharge assessment criteria. Furthermore, the post-discharge rehabilitation and nursing plans should be drawn up early, and medical and nursing resources should be reasonably allocated according to the possibility of ADL recovery.
This study also found that the ADL score at 6 months after discharge of older patients who returned home (87.64±21.51) was significantly higher than that of those going to nursing homes or other care institutions (62.22±31.31). It might be that patients going home could experience care, companionship, and support from their families, which might play a role in the recovery of their functions; or might be the ADL and diseases of the two above-mentioned groups were distinct at discharge. Patients who went to institutions after discharge had a score of (24.35±19.35), which was significantly lower than that of those who went home after discharge (55.93±19.44), which could be the reason why patients going home had a higher ADL score at 6 months after discharge.
Depression, albumin, and IADL 2 weeks before admission had no significant effect on the recovery of ADL at 6 months after discharge. In terms of depression, patients with a depressive tendency accounted for a small proportion of the sample (8.7%), which was lower than in previous surveys (17.3%) . This may be related to the following reasons. Firstly, evaluation tools were different, and the 5-item Geriatric Depression Scale was used in this study, which is shorter than the tools in previous research. Secondly, evaluation time was different, as previous studies assessed depression during hospitalisation, while this study completed assessments 1-2 days before discharge or at discharge. Older patients who were about to be discharged might readjust their mental status, as the disease had improved and their environment would change soon. Thirdly, previous studies confirmed that cognitive impairment was associated with depression. However, in this study, patients with cognitive impairment accounted for a low proportion (6.2%). In terms of albumin, studies have shown that low albumin (<35g/l) was associated with ADL recovery. The proportion of patients with low albumin in this study was only 18.0%, which is much lower than in previous studies (76.7%) . The reason may be that previous studies included populations that were dependent at baseline and whose basic nutritional status might be poor. However, the subjects in this study were independent at baseline, and most had good basic nutritional status and may not be prone to low albumin during hospitalisation. In addition, recent studies  pointed out that exploring the predictive effect of some biological parameters (e. g. albumin) on ADL in older patients were rare and some results of those studies were mutually contradictory [20, 33, 34]. Therefore, the relationship between albumin and ADL recovery after discharge in hospitalised ADL-dependent older patients needs to be further studied.
As for IADL 2 weeks before admission, some scholars held that patients who had much more IADL dependent were less likely to regain ADL independence. However, there was no association between the two in this study, which may need to be further examined in future research.
In this study, the ROC curve analysis of age, cognition, post-discharge residence, and ADL score at discharge showed that the AUC for the combined predictors was 0.763, which is higher than in previous studies (0.640-0.784) . This indicates that the predictive value of the model is strong, which is helpful to predict the recovery of ADL of the elderly dependent population. Results of study indicate that the medical staffs should pay more attention to the assessments and individualized rehabilitation guidance among the patients who are greater than or equal to 65 years, with the AMT scores was below 8 and a lower score of ADL during the hospitalization. Additionally, enlarging the knowledge about rehabilitation and improving the skill to rehabilitate of the patients as well as their caregivers is essential to regain the ADL independence. Health care workers concentrate on the follow-up of patients and the rational care resources allocation after discharge, especially for the patients in the institution. Most importantly, continual assessments and rehabilitation guidance should be executed during the follow-up in this population.
Due to practical limitations, this study could not examine other factors that may have an impact on ADL in early stages of disease, such as the comorbidity index. In addition, this study did not consider factors that may affect ADL after discharge in older patients, which need to be further studied in the future. It have been reported that patients with neurological diseases accounted for the majority of the hospitalized in this study, so the population of this study cannot represents all of acute hospitalized elderly patients.