Table 3.

MSSP Hospital Clusters

	Cluster 1	Cluster 2	Cluster 3	Cluster 4	Cluster 5
Cluster Label	MSSP - low ambulatory services but high health IT	MSSP - high health IT	MSSP - low health IT	MSSP - high tight MD alignment & health IT	MSSP -high loose MD alignment & health IT	MSSP- overall
Count of physicians in loose hospital arrangements per 1000 patient days***	0.23	0.61	1.22	1.47	9.12	1.61
Count of physicians in tight hospital arrangements per 1000 days***	0.15	0.46	0.72	5.84	0.57	0.89
Count of 5 ambulatory services***	0.82	3.09	2.36	2.85	2.71	2.47
Count of 5 basic Health IT applications***	4.92	4.79	2.32	4.92	4.85	4.44
Count of 2 advanced Health IT applications***	1.97	1.97	0.56	1.88	1.91	1.74
Number of hospitals in cluster	62	146	50	26	34	318
Percent of hospitals in cluster	19%	46%	16%	8%	11%	100%

NOTES:

Summary statistics from combined sample cluster analysis: a) number of clusters at which Cubic Clustering Criterion (CCC) statistic first exceeded value of 2: 5 clusters; b) number of clusters at which CCC value levels off: 5 clusters (CCC at 4 clusters: 1.95; at 5 clusters: 4.01; at 6 clusters: 4.84; c) number of clusters at which Pseudo F statistic levels off: 5 clusters (value equals 105 at 4 clusters; 106 at 5 clusters; and 101 at 6 clusters); d) number of clusters after Pseudo t-squared statistic reaches first peak and then declines: 5 clusters (value equals 158 at 4 clusters; 18.3 at 5 clusters).

^***characteristic significantly different a p<.01 level across clusters.

BOLD values indicate significant different values at p<.05 for particular cluster relative to other clusters.

ANOVA analysis used to test for overall significant differences across entire set of clusters; Duncan multiple range tests used to examine significant differences in means across individual clusters.