Table AI.
Frequency tables of hospice-use reports from medical records and medicare claims, by study sites.
| GHC Medicare claims yes |
Medicare claims no | |
| Medical records yes | 23 | 2 |
| Medical records no | 38 | 22 |
| HPHC Medicare claims yes |
Medicare claims no | |
| Medical records yes | 2 | 0 |
| Medical records no | 0 | 0 |
| HFHS Medicare claims yes |
Medicare claims no | |
| Medical records yes | 7 | 3 |
| Medical records no | 2 | 11 |
| KPHI Medicare claims yes |
Medicare claims no | |
| Medical records yes | 1 | 0 |
| Medical records no | 4 | 8 |
| KPNW Medicare claims yes |
Medicare claims no | |
| Medical records yes | 41 | 7 |
| Medical records no | 20 | 23 |
| NCCC Medicare claims yes |
Medicare claims no | |
| Medical records yes | 43 | 12 |
| Medical records no | 130 | 179 |
| UAB Medicare claims yes |
Medicare claims no | |
| Medical records yes | 20 | 0 |
| Medical records no | 82 | 101 |
| UCLA Medicare claims yes |
Medicare claims no | |
| Medical records yes | 62 | 9 |
| Medical records no | 78 | 122 |
| IOWA Medicare claims yes |
Medicare claims no | |
| Medical records yes | 190 | 21 |
| Medical records no | 70 | 112 |
| UNC Medicare claims yes |
Medicare claims no | |
| Medical records yes | 6 | 0 |
| Medical records no | 21 | 39 |
Note: The subsample consists of 2669 CanCORS patients who died within 15 months of diagnosis and had at least one observed hospice-use record from two sources. An example for calculating unadjusted site-specific capturing rate for two data sources: For GHC, the estimate of Medicare Claims is 23/(23 + 2) ≈ 92%; the estimate of medical records is 23/(23 + 38) ≈ 38% (see also the footnote of Table IV).