Table 2.

MSE for fit of different models. MSE is calculated as ${\sum }_{i=1}^n{\mathit{MSE}}_i$ where ${\mathit{MSE}}_i=\frac{{({\widehat{Z}}_{ijt}-Z_{ijt})}^2}{T_i}$. For the models with multiple indicators, the total MSE is just the sum of the MSE’s for each type of encounter.

	Multiple Indicators		Total Encounters
	HMM	Poisson regression	HMM	Poisson regression
Number of parameters	17	21	11	9
Total MSE	6369	10,926	6069	8705
Not Alcohol MSE	3995	8398
Alc Acute MSE	1210	1304
Alc Specific MSE	913	973
Alc Chronic MSE	251	251