Skip to main content
. 2016 Oct 3;18(7):869–882. doi: 10.1007/s10198-016-0834-x

Table 4.

Sensitivity analysis: three scenarios (1: excluding dead, 2: cost truncation, 3: intention-to-treat) with difference-in-difference estimators (ATT) and their relative change compared to the baseline scenario

Excluding dead (1) (n: TM = 630, C = 6607) Cost truncation (2) (n: TM = 651, C = 7047) ITT (3) (n: TM = 815, C = 7047) Baseline
ATTa Δ % ATTa Δ % ATTa Δ % ATTa
Total costs −776.26 13.28 −553.62 38.15 −706.30 21.09 −895.11*
 Inpatient treatment −936.43* 11.33 −826.14* 21.77 −919.54* 12.93 −1056.04**
  Thereof due to COPD −624.71*** 2.74 −597.94** 6.90 −554.96** 13.60 −642.28***
 Outpatient treatment 63.98 7.99 68.82* 1.03 65.11* 6.36 69.54*
 Pharmaceuticals 101.71 −10.79 130.19 −41.82 145.13 −58.09 91.80
 Medical appliances/rehabilitation −5.51 −1266 −0.40 0.00 2.99 −642 −0.40
Indicators for healthcare utilisation
 Average length of hospital stay −1.48*** −2.91 n.a. n.a. −1.24*** 13.64 −1.44***
  Thereof due to COPD −1.72*** 2.18 n.a. n.a. −1.68*** 4.673 −1.76***
 Inpatient bed days −3.01*** 2.72 n.a. n.a. −2.82*** 9.05 −3.10***
  Thereof due to COPD −2.07*** −0.21 n.a. n.a. −1.86*** 10.03 −2.07***
 Inpatient stays −0.21** 4.12 n.a. n.a. −0.16** 26.04 −0.21***
  Thereof due to COPD −0.19*** −4.95 n.a. n.a. −0.15*** 14.03 −0.18***
  Thereof ED visits due to COPD −0.15*** −3.75 n.a. n.a. −0.13*** 10.50 −0.14***
 Proportion hospitalised (in %) −14.95*** 1.38 n.a. n.a. −11.77*** 22.36 −15.16***
  Thereof due to COPD −19.98*** 1.42 n.a. n.a. −18.86*** 6.95 −20.27***
  Thereof in ED due to COPD −16.72*** 1.64 n.a. n.a. −15.12*** 11.06 −17.00***
 Physician visits 1.09*** 14.82 n.a. n.a. 1.21*** 5.10 1.27***
  Thereof due to COPD 0.77*** 10.03 n.a. n.a. 0.76*** 11.85 0.86***
 Prescriptions 0.91 45.35 n.a. n.a. 1.48** 11.31 1.67**
Indicators for mortality
 All-cause mortality (in %) n.a. n.a. n.a. n.a. −3.73*** −24.58 −2.99***

Δ %: deviation (in %) of respective sensitivity analysis value from baseline scenario value

* < 0.05; ** < 0.01; *** < 0.0001

aAverage treatment effect for the treated represents excess resource utilisation attributable to DMP