Composite Hospital Compare Measures (Hospital) (All-payer)

Queries

Acquiring data on All Composite Process of Care Measures

select a.* from entity_values as a, measure_groups_measures as b
where a.measure_id = b.measure_id
and a.entity_id = ######
and b.measure_group_id = 5

NOTE: The Commonwealth Fund has ADDITIONAL rules for ADDITIONAL visualisation/sorting/ranking: see High performers
The instructions below are Pellucid-specific, NOT Commonwealth/WNTB specific. The rules on this page are applied to ALL Pellucid properties.
Summary rule: all process measures are aggregated into composite scores. All composite scores are posted to Pellucid. Web Application Only will decide to suppress scores where d < 30. If a hospital submits ONE data point for one measure, we still produce a composite score. Web Application will suppress it as necessary.

Calculation Method

Composite score in Pellucid is a weighted average of the selected indicator rates, which is the sum of numerators divided by the sum of denominators.

Example: Given 4 fractions, 12/20, 11/30, 10/40 and 9/50, the composite score will be:
∑ numerators / ∑ denominators = (12+11+10+9) / (20+30+40+50) = 42 / 140 = 30%

Components

Composite scores were created for each of the 4 major disease conditions, including AMI, HF, PNE and SCIP. Number of measures increase over time.

Ranking Criteria

For comparability, we take a subset of the composite scores and rank them. Ranking has stricter inclusion than compositing.

*Measures must contain all available quarters worth of data, hospital must have submitted data for all reportable quarters (i.e. original CMS footnote was not equal to 3. ).

• Must report all indicators within the topics (i.e. fully reported). For example, HF composite score will be ranked only if all 4 HF indicators were reported.

• At least one indicator in the topic had denominator >= 30.

• Rank 'overal composite' when all 4 topic composite measures were available. i.e. if all four topics meet the criteria, the Overall Composite meets the criteria.

New Footnotes Created for CMS Composite Scores

11 = denominator was less than 30. (Note: this footnote was created when every indicator had less than 30 cases)

90 = eligible for ranking (i.e. "fully reported" + "at least one indicator in each topic had denominator >=30")

91 = high performer (i.e. in top 1st percentile.)

This is JMKing's old note:

There are three tiers to the data logic:

1. Calculation of all possible measures for inclusion in the master scores table

2a. Ranking of the five composite measures using specified inclusion criteria for use in the "top 10%" subset, and flagging those entities who are satisfy the inclusion criteria - this generates a subset of hospitals who satisfy Commonwealth fund inclusion criteria, and only their scores are used to rank the top ten percent. (NB, Commonwealth may chose to change this to the "top 1%" or use both)

2b. Ranking of all remaining measures.

3. Publication on a Web site of all scores that meet a second inclusion criteria, specifically 30 or more cases per measure, this will by definition include all scores that meet the above Commonwealth inclusion criteria and part of the remaining subset of hospitals.

This means that a rate and all composite scores must be generated for all hospitals for all measures. Whether or not they are to be included in various subsets for ranking and/or publication is moot, calculate all available rates and composites.

The instruction below are specific to the Web site, and may not be the same as separate data analysis performed for Commonwealth.

FOR EACH HOSPITAL reporting data in "quality scores"

1. calculate a numerator for every reported measure using supplied rate X denominator, rounded from two decimal places. Numerator must be an integer.

2. calculate a weighted average for each topic that includes at least one data point to produce a topic composite

3. calculate a weighted average for all topics to produce an all-topic composite

4. indicate all measures which were reported in each topic, eg 1,4,6,7,14,16,21

5. for any topic where all included measures were reported mark topic as 'fully reported' - use measures reported as zero (0) denominator (hospital reported but had no cases)

- do not use measures "na" or -1

6. for any topic where all included measures were reported AND at least one indicator in the topic has a denominator of >29 mark topic as 'topic eligible'

7. for any hospital where all included measures were reported AND at least one indicator in each reported topic (of four topics) has a denominator of >29 mark topic as'all-topic eligible'

8. using only those hospitals that satisfied #7, produce a report that displays in rank order the top ten percent of hospitals satisfying the inclusion criteria including tied placement (e.g. 1,2,2,4) for the all-topic group - using this report produce an average score of the top ten percent

9. using only those hospitals that satisfied number #6, produce a report that displays in rank order the top ten percent of hospitals satisfying the inclusion criteria including tied placement (e.g. 1,2,2,4) for each topic group - using this report produce an average score of the top ten percent -- Patient satisfaction is simpler, all reporting hospitals report all measures. Therefore, we rank all hospitals to derive the top ten percent. Separately, we need to know the top ten percent stratified by survey size, split into three options: <100, 100-300, >300 surveys. It is possible that the comparator set for Commonwealth will be based on all hospitals or only hospitals reporting >300 surveys, for beta we are using all hospitals. -- - ranking should occur using two decimal places for all measures - rates are published to two decimal places, e.g. 98.35% - the Web site should suppress any display of a single data point where that data point is <30 cases, based on the relevant footnote - however this may change, therefore we store the suppressed score - the Web site may display hospital scores that met the inclusion criteria differently that those that did not, probably by use of an asterix or footnote to identify hospitals that did not meet the inclusion criteria - PLEASE NOTE: THE ALL TOPIC COMPOSITE MEASURE ID is 5099, not 9000.