Extended Data Fig. 3: Sensitivity analysis using alternative cognitive scale and dementia classification: trends in the proportion of people receiving no care for ADL and IADL functional limitations among persons with dementia and functional limitations in the HRS, ELSA, SHARE, and CHARLS, 2012–2018.
From: Absence of care among community-dwelling older adults with dementia and functional limitations
a, Care absence for ADL and/or IADL limitations (HRS: n = 4,191; ELSA: n = 1,035; SHARE: NA; CHARLS: n = 2,037). b, Absence of informal care for ADL and/or IADL limitations (HRS: n = 4,180; ELSA: n = 1,012; SHARE: NA; CHARLS: n = 2,037). c, Absence of formal care for ADL and/or IADL limitations (HRS: n = 4,168; ELSA: n = 1,011; SHARE: n = 5,120; CHARLS: n = 2,037). d, Care absence for ADL limitations (HRS: n = 3,159; ELSA: n = 865; SHARE: n = 1,843; CHARLS: n = 1,273). e, Absence of informal care for ADL limitations (HRS: n = 3,151; ELSA: n = 865; SHARE: n = 1,845; CHARLS: 338). f, Absence of formal care for ADL limitations (HRS: n = 3,127; ELSA: n = 865; SHARE: n = 3,759; CHARLS: 338). g, Care absence for IADL limitations (HRS: n = 3,253; ELSA: n = 656; SHARE: NA; CHARLS, n = 1,686). h, Absence of informal care for IADL limitations (HRS: n = 3,236; ELSA: n = 639; SHARE: NA; CHARLS: 501). i, Absence of formal care for IADL limitations (HRS: n = 3,236; ELSA: n = 638; SHARE: n = 3,876; CHARLS: 501). In panels (a-i), data are presented as adjusted means ± 95% confidence intervals (CIs). Dotted points represent the estimated proportion of people receiving no care (a, d, g), no informal care (b, e, h) and no formal care (c, f, i) at each wave, as estimated using the GEE model among samples who developed dementia. In this sensitivity analysis, participants’ dementia status was evaluated using the same 25-score scale and defined based on 1.5 SDs threshold in the four surveys. Error bars indicate the 95% CIs. Estimates for certain waves in SHARE and CHARLS are unavailable due to lack of data (details provided in Supplementary Table 5). Corrections for multiplicity were not made. Panels (a-c) include samples with at least one ADL or IADL limitation, (d-f) with at least one ADL limitation, and (g-i) with at least one IADL limitation. Adjusted Annual Percent Change (AAPC) estimates derived from the GEE model are shown on the right of each panel. Asterisks denote the significance of the AAPC (time trend): *** p < 0.001, ** p < 0.01, * p < 0.05. ‘NA’ indicates unavailable estimates. Two-sided tests were performed using GEE model, with detailed test statistics and P values provided in the Source Data.
