Table 2:

Comparison of patient characteristics before and after propensity score matching

CharacteristicBefore propensity score matchingAfter propensity score matchingp value*
UnexposedExposedUnexposedExposed
No. of patients (pooled across the teams)41 133391231093109
Age at death, yr, median (IQR)80 (69–87)75 (64–84)74 (63–83)75 (64–84)NS
Cancer diagnosis, no. (%)14 443 (35.1)3073 (78.6)2481 (79.8)2469 (79.4)0.7
Female, no. (%)20 895 (50.8)2032 (51.9)1609 (51.8)1600 (51.5)0.8
Adjusted Clinical Group comorbidity weighting, mean ± SD5.12 ± 4.096.30 ± 4.046.21 ± 3.936.20 ± 3.98NS
Home care service type, no. (%)
 End of life6208 (15.1)3041 (77.7)2409 (77.5)2409 (77.5)1.0
 Long-term supportive3408 (8.3)210 (5.4)145 (4.7)145 (4.7)1.0
 Maintenance7692 (18.7)328 (8.4)278 (8.9)278 (8.9)1.0
 Rehab/acute care6478 (15.7)157 (4.0)106 (3.4)106 (3.4)1.0
 None17 347 (42.2)176 (4.5)171 (5.5)171 (5.5)1.0
Time from first receipt of most severe home care service type to death, d, mean ± SD
 End of life15.47 ± 56.4886.18 ± 107.2979.23 ± 102.0279.32 ± 102.05NS
 Long-term supportive23.92 ± 81.1731.00 ± 87.2537.96 ± 95.9227.84 ± 83.51NS
 Maintenance63.68 ± 124.8082.02 ± 131.6981.64 ± 131.0683.55 ± 132.16NS
 Rehab/acute care69.70 ± 113.2776.05 ± 113.9879.29 ± 112.6080.12 ± 114.96NS
No. of prior emergency department visits, mean ± SD1.16 ± 2.101.43 ± 2.111.40 ± 1.951.36 ± 1.96NS
Prior length of stay in hospital, d, mean ± SD7.18 ± 22.237.53 ± 17.096.84 ± 14.176.73 ± 14.85NS
No. of prior hospital visits, mean ± SD0.56 ± 1.040.75 ± 1.140.76 ± 1.150.70 ± 1.07NS
Income quintile, no. (%)
 1 (lowest)10 288 (25.0)668 (17.1)772 (24.8)536 (17.2)< 0.001
 29053 (22.0)746 (19.1)703 (22.6)605 (19.5)0.003
 37565 (18.4)720 (18.4)604 (19.4)550 (17.7)0.08
 47460 (18.1)872 (22.3)585 (18.8)675 (21.7)0.005
 5 (highest)6555 (15.9)906 (23.2)437 (14.1)743 (23.9)< 0.001
  • Note: IQR = interquartile range, NS = not significant, SD = standard deviation.

  • * For characteristics reported as means and medians it was not possible to calculate actual p values at the time of publication because the individual data were no longer accessible, for privacy reasons.