# Table 3 Estimated ATET by number of matches and pooled across the different numbers of matches. Year = 2013 and 7 RHCs that joined an ACO in 2012

Study ATET [95 % Conf. Interval] % Weight
Part 1 (Year is 2012 and the 7 RHCs that joined an ACO in 2012, n = 544)
Test of ATET = 0: z = 6.68, p = 0.000
Matches = 1 26.740 8.022 45.458 15.61
Matches = 2 29.410 11.653 47.167 17.34
Matches = 3 26.610 8.931 44.289 17.50
Matches = 4 23.040 8.066 38.014 24.39
Matches = 5 22.430 7.691 37.169 25.17
Pooled ATET (note 1) 25.193 17.799 32.588 100.00
Percentage of mean 14.02 %
Part 2 (Year is 2013 and the 7 RHCs that joined an ACO in 2012, n = 435)
Test of ATET = 0: z = 3.17, p = 0.002
Matches = 1 14.110 −10.488 38.708 18.81
Matches = 2 19.790 −5.592 45.172 17.67
Matches = 3 18.450 −5.030 41.930 20.65
Matches = 4 16.980 −6.285 40.245 21.03
Matches = 5 17.170 −5.664 40.004 21.83
Pooled ATET (note 1) 17.282 6.612 27.951 100.00
Percentage of mean 14.02 %
Part 3 (Year is 2013 and the 14 RHCs that joined an ACO in 2013, n = 434)
Test of ATET = 0: z = 3.73, p = 0.000
Matches = 1 27.590 0.621 54.559 17.60
Matches = 2 23.840 −2.267 49.947 18.78
Matches = 3 20.180 −5.535 45.895 19.36
Matches = 4 17.740 −5.976 41.456 22.76
Matches = 5 19.850 −4.552 44.252 21.50
Pooled ATET (note 1) 21.545 10.231 32.859 100.00
Percentage of mean 17.48 %
Part 4 (Panel for 2012 and 2013 and the 7 RHCs that joined an ACO in 2012, n = 984)
Test of ATET = 0: z = 5.12, p = 0.000
Matches = 1 13.490 −3.189 30.169 15.57
Matches = 2 17.020 3.300 30.740 23.02
Matches = 3 16.320 2.581 30.059 22.95
Matches = 4 19.670 4.441 34.899 18.68
Matches = 5 18.940 4.142 33.738 19.78
Pooled ATET (note 1) 17.185 10.603 23.766 100.00
Percentage of mean 14.05 %
1. Note 1: Employing the meta-analysis fixed effect model using the method of Mantel and Haenszel, 1959 
2. Note 2: Mean cost/visit = \$121.43 (2012), \$123.29 (2013), and \$122.36 (2012 and 2013)
3. Note 3: Panel 4 shows the results of using the fixed effects model (the within transformation) for a T = 2 panel. This is not the same as the fixed effect model used in the Mantel-Haenszel method 