DMS Classification Oddities

[jeffeastman]

When running this measure against 100 generated patient histories (v16), I see the odd result of more patients in the numerators and denominators than are in the initial-population. Seems to me the CQL ought to require initial-population membership as a criteria for numerators and denominators.

===== Server Summary =====
	denominator 1 = 94
	denominator 2 = 94
	denominator 3 = 94
	initial-population = 62
	numerator 1 = 74
	numerator 2 = 74
	numerator 3 = 74

Looking at the CQL, it seems the Numerators check for Denominator membership but the Denominators don't check for Initial Population:

define "Denominator 1":
	{First(
		"Depression Encounters" E
			where PeriodToIntervalOfDT(E.period) during "Assessment Period One"
			sort by start of PeriodToIntervalOfDT(period)
	)}

define "Denominator 2":
	{First(
		"Depression Encounters" E
			where PeriodToIntervalOfDT(E.period) during "Assessment Period Two"
			sort by start of PeriodToIntervalOfDT(period)
	)}

define "Denominator 3":
	{First(
		"Depression Encounters" E
			where PeriodToIntervalOfDT(E.period) during "Assessment Period Three"
			sort by start of PeriodToIntervalOfDT(period)
	)}