We found 253 relevant articles from three databases, PubMed (116), Embase (128), and Cochrane Library (9). After using Endnote software to delete the duplications, 175 articles remained. At this stage, two authors checked the titles and abstracts to identify closely relevant studies, with 28 assessed for eligibility. Consequently, after checking the full text, there were 11 studies included for quantitative analysis in terms of the pre-defined inclusion criteria. Detailed information on the reasons for exclusion is presented in Figure 1.
There are 11 studies for quantifying the relation between frailty and mortality, with a of 22,105 patients with COVID-19. Overall, a majority of the studies that were included studies only focused on older adults. The study design in five studies was prospective cohort studies[8, 12, 18-20] , while the others were all retrospective cohort studies[9, 11, 13, 15-17]. There was a variety of countries, ranging from the U.S. to European countries, with 2 in UAS[11, 16], four in the UK[8, 13, 15, 20], 1 in Turkey[19],1 in France[17], 1 in Spain[9], 1 in Sweden[12], and 1 in the UK and Italy[18]. Of these, ten study settings were in hospital[8, 9, 11-13, 15, 17-20] and one was in a nursing home[16]. The prevalence of frailty ranged from 11.00% to 71.3%, producing the pooled results was 51%(95%CI:42%-60%)(SFigure1). The majority of outcomes was in-hospital mortality, with only one study reporting 30-day mortality. The largest sample size was in Turkey[19] with 18,234, and the smallest was in France[17] with 94 patients. Among all of the included studies, seven studies used clinical frailty scale as an assessment tool for frailty[8, 9, 12, 13, 15, 18, 20], one for hospital frailty risk score[19], one for frailty index[16], one for palliative performance scale[11] and one for frail non-disabled questionnaire[17]. (shown in Table 1)
Meta-analysis of the effects of frailty on mortality
There are 11 studies for meta-analysis, with the results showing pooled HR value of 2.27 (95%CI:1.79-2.89) among frail patients, compared with COVID-19 patients without frailty, indicating that frailty can be an independent predictor for mortality among patients with COVID-19 (Figure2)
Subgroup analysis
Five studies were prospective cohort, and the others were retrospective cohort studies, thus, we performed subgroup analysis based on study design. The results indicated a statistically significant association between frailty and mortality in both of these groups (cohort study versus retrospective cohort study: HR=2.23,1.81-2.75 versus HR=2.44,1.20-4.96)(SFigure2). We also performed subgroup analysis based on geographic region, due to the different prevalence levels of COVID-19 worldwide. The results found the associations in the U.SA. were higher than in European countries(SFigure3), with both suggesting that patients with frailty have an incremental greater risk of mortality than non-frail COVID-19 patients.
Subgroup analysis was based on different populations.
A majority of studies focused on hospitalized patients, with only one study reported among nursing home residents. Older nursing home residents infected with COVID-19 coexisting with frailty had a 2.95-fold risk of morality, compared to non-frail patients. (HR=2.95,95%CI:1.19-7.32). Meanwhile, hospitalized patients also had similar results. (HR=2.24,95%CI:1.74-2.89)(Figur3).
Subgroup analysis was based on different frailty assessment scales.
A majority of studies included the CFS frailty assessment scale, and the results showed that frail patients with COVID-19 presented an increased risk of mortality when using CFS tool to assess frailty syndrome (HR=2.41,95%CI:1.60-3.62). Other frailty assessment instruments included the Frailty index (HR=2.95,95%CI:1.19-7.32), hospital frailty risk score (HR=1.96,95%CI:1.72-2.15) and palliative performance scale (HR=2.89, 95%CI:1.42-5.87), which all indicated similar results, with the exception of the Frail non-disabled questionnaire (HR=1.15,95%CI:0.22-5.99),showed in Figure4.
Quality assessment
A majority of studies had more than a seven-point score, and one study had six points according to the criterion of the Newcastle–Ottawa Scale. (Table 2)
Sensitivity analysis and potential publication bias
Begg’s test was used for publication bias and the results showed no potential bias (p=0.512) (SFigure4). We also conducted a sensitivity analysis, and the results indicated our study was stable and robust (SFigure5).