Table 1.
Baseline characteristics of three cohorts
| Characteristics | Alberta | Denmark | Scotland | ||
|---|---|---|---|---|---|
| No of participants | 67 942 | 17 528 | 7740 | ||
| Age (years), median (IQR) | 77.6 (69.3-84.4) | 77.4 (70.8-83.1) | 79.9 (73.1-85.4) | ||
| Age <65 years, n (%) | 11 106 (16.3) | 2218 (12.7) | 758 (9.8) | ||
| Age 65-74 years, n (%) | 16 838 (24.8) | 4763 (27.2) | 1606 (20.7) | ||
| Age 75-84 years, n (%) | 24 399 (35.9) | 7449 (42.5) | 3326 (43) | ||
| Age ≥85 years, n (%) | 15 599 (23) | 3098 (17.7) | 2050 (26.5) | ||
| Male sex, n (%) | 31 368 (46.2) | 9073 (51.8) | 3523 (45.5) | ||
| CKD-EPI 2009 formula | |||||
| Index eGFR, median (IQR) | 38.8 (34.0-42.1) | 39.5 (35.0-42.4) | 39.2 (34.7-42.2) | ||
| Index eGFR 15-29, n (%) | 9297 (13.7) | 1897 (10.8) | 904 (11.7) | ||
| CKD-EPI 2021 formula | |||||
| Baseline eGFR, median (IQR) | 41.6 (36.4-45.1) | 42.7 (37.8-45.9) | 42.0 (37.3-45.3) | ||
| Baseline eGFR 15-29, n (%) | 6729 (9.9) | 1244 (7.1) | 626 (8.1) | ||
| Qualifying period (days), median (IQR) | 168 (112-296) | 133 (104-194) | 168 (113-281) | ||
| Qualifying eGFR tests (n), median (IQR) | 2 (2-3) | 3 (2-4) | 3 (2-4) | ||
| Albuminuria (mg/g), median (IQR)* | 12.4 (11.9-66.6) | 22.7 (8.6-87.5) | 22.1 (8.8-98.1) | ||
| A1 (<30 mg/g), n (%) | 42 960 (63.2) | 9909 (56.5) | 4418 (57.1) | ||
| A2 (30-300 mg/g), n (%) | 14 550 (21.4) | 5403 (30.8) | 2242 (29.0) | ||
| A3 (>300 mg/g), n (%) | 10 432 (15.4) | 2216 (12.6) | 1080 (14.0) | ||
| Albumin-to-creatinine ratio type | |||||
| Measured, n (%) | 34 122 (50.2) | 17 528 (100) | 6898 (89.1) | ||
| Protein-to-creatinine ratio calculated, n (%) | 3268 (4.8) | 0 | 842 (10.9) | ||
| Dipstick calculated, n (%) | 30 552 (45.0) | 0 | 0 | ||
| Diabetes, n (%) | 30 641 (45.1) | 8051 (45.9) | 1945 (25.1) | ||
| Cardiovascular disease, n (%) | 31 821 (46.8) | 4193 (23.9) | 2193 (28.3) | ||
| Myocardial infarction, n (%) | 6719 (9.9) | 969 (5.5) | 1251 (16.2) | ||
| Heart failure, n (%) | 19 058 (28.1) | 2253 (12.9) | 969 (12.5) | ||
| Stroke or TIA, n (%) | 15 021 (22.1) | 1432 (8.2) | 684 (8.8) | ||
| Peripheral vascular disease, n (%) | 4287 (6.3) | 446 (2.5) | 621 (8.0) | ||
| Chronic pulmonary disease, n (%) | 21 485 (31.6) | 3652 (20.8) | 1034 (13.4) | ||
| Cancer, n (%) | 14 842 (21.8) | 2185 (12.5) | 662 (8.6) |
Alberta, Canada, data accrual was from 1 April 2008 to 31 March 2019. We tested the transportability of the super learner in Denmark (Central Region accrual from 1 January 2007 to 31 December 2018, North Region accrual from 1 January 2012 to 31 December 2021), and Scotland (accrual from 1 January 2011 to 31 December 2019). CKD-EPI=chronic kidney disease epidemiology collaboration; eGFR=estimated glomerular filtration rate, in mL/min/1.73 m2; IQR=interquartile range; cardiovascular disease=binary summary (one or more); TIA=transient ischaemic attack; cancer=any non-epithelial skin cancer.
Measured albumin-to-creatinine ratio or albumin-to-creatinine ratio calculated from protein-to-creatinine ratio or urine dipstick. Conversion factor for albumin-to-creatinine ratio: 1 mg/mmol=0.113 mg/g. Of note, people who had urine dipstick measures of proteinuria were included only in the Alberta cohort, because in Alberta, unlike Denmark and Scotland, dipsticks are part of usual care and workflow for the information management system. This allows generalisation of the use of the prediction tool in settings where urine dipstick testing is available.