Examples of Different Models for Scoring Componentsin HPSA and MUA Designation:Discussion Paper for the Negotiated Rule Making CommitteePrepared by:National Center for Health Workforce Analysis,Bureauof Health Professions, HRSA14April 2011
1
Grouping various indicators into categories or components.Methodfor scoring indicators within each component to calculate a component score;andDevelopinga method for combiningcomponentscores for determining designation as HPSA and/or MUA.Focus on II and IIINote: Examples outlined may be applicable to models for MUP &PopulationHPSA designations although not discussed.
2
II. Scoring indicators ina component forcomponent score
Indicators combinedwith equal weights or unequalweights?
How combined indicators with different scales?
One method to “standardize” scales: Convert values intopercentiles (ranking of areas on indicator values)
Group into percentile groups (e.g.,centiles, deciles). If deciles used, 10 = “worstcase” and 1=“best case.”
Percentile groups as % of areas (e.g., % all US counties) or % of US population covered?If deciles, 10% of counties or 10% of population?
3
Component score:Indicators givenequalweight:anarea’s score forcomponent is sumof scores on eachindicatorIndicators given different weights:Weights total 1.00Eachindicator scoremultiplied by weightWeightedscores summed acrossindicators
Assumption: Distributions of indicators reasonably spread out across universe of areas?
How handle indicator notreasonably dispersed across the universe ofareas—rare event/inrelatively fewareas? (Bottom page 2 in paper)
4
5
Combiningindividual component scores for determining designation asHPSAand/orMUA:Examplesare possible modelscenarios:
Model One: A generic “index” model: all components combinedintosinglescore,withequal weight as a starting premise.Model Two: A ‘sequential’ modelrequiring threshold onone componentmetfirst beforescoredfor other components.Firstcomponentaninitial filter. Other componentstreatedwith equal weight.Model Three: A ‘sequential’ model requiring thresholdbased upon two componentsmetfirst beforescoredfor other components. Other componentstreatedwith equal weight.
6
Model One: A generic “index” model : all components combined into single score, with equal weight as a starting premise.1(Pages 4 &5)
1Components listed for purposes of illustration only and may not be the components the Committee decides to use in its final model or models.The Committee has also discussed including ability to pay as its own component. Ability to pay could also be included under Access Barriers or Social Determinants of Health. No decision has been made on this.
Component 1: (e.g., Social Determinant Index (SDI))Component 2: (e.g., Observed Health Status)Component 3: (e.g., Observed Access Indicators (e.g.,ACS hosp.,etc.))Component 4: (e.g., Access Barriers (e.g., race, language,etc)Component 5: (e.g., Providerto population ratio(P2P))
7
Step 1: For each component, calculate the distribution of the statistic (i.e., the score on the component).
Step 2: For each component, categorize the distribution of the statistic by a percentile grouping.
Step 3: For eachgeographicalarea,score eachcomponent based upon the percentile group. If decilesused,the scores would range from 1 to 10where 10 could be“worst case” and 1“best case.”
Step 4: Calculate the total score for each area by summing the scores in each component.
Step 5: Set the threshold for designation based upon the distribution of the total score, e.g., the highest XX% of the total scores are designated as an area in need.
Thiscutoff value would be the threshold value in the rule for designation of an area in need.
ModelOne--continued
8
Model Two: A ‘sequential’ model requiring threshold on one component met first before scored for other components.In this example, we focus on designation of a HPSA for illustration purposes.
Step 1. Threshold for filter component:Component5:Provider/PopulationRatio(P2P)
Thisthreshold would be the first threshold in therule
Step 2: Continue evaluation of HPSA statuson Components 1 through 4 foronlyareassatisfying Step1.
Component 1: Social Determinant Index (SDI)Component 2: Observed Health StatusComponent 3: Observed Access IndicatorsComponent4: Access Barriers
Scoring of the components 1, 2, 3, and 4 would be similar to the scoring of these components in Model One.
9
Step 3:Sum scores on Components1 through 4for total score
Step 4: Set the threshold for final HPSA designation based upon the distribution of the total score of Components 1 through4:XX% of the areas considered after Step 1 with the highesttotal score on Components 1 through 4.
Thiscutoff value would be the threshold value in the rule for HPSA designation among areas meeting the threshold of Step 1.
Model Two--continued:
10
Model Three:A ‘sequential’ modelrequiring thresholdbased upon two componentsbemetfirst.For purposes of this illustration, we focus on an MUA designation.
Step 1. Threshold for filter based upon combined score oftwo components (Components 1 and 2):
SocialDeterminant Index (SDI)
ObservedHealth Status
Score on SDI (e.g., rankingby deciles)
Score on Observed Health Status(e.g., ranking by deciles)
First Total Score(sum of SDI and Observed Health Status scores)
SetfirstthresholdinMUA designationprocess
Thiswouldbe the first threshold in the ruleto meet for further evaluation.
11
Step 2: Continue evaluation of MUA status on Components 3 through 5 for only areas satisfying Step 1.
ObservedAccessAccessBarriersProvider/Pop. (P2P)Indicators
Score on Observed Access(e.g., ranking on deciles)
Score on Access Barriers(e.g., ranking on deciles)
Score on P2P(e.g., ranking on deciles)
Second Total Score(sum of Observed Access, Access Barriers and P2P)
Step3:
Step4: Set thresholdfor final MUA designation based uponthedistribution ofthesecondtotalscore
12
Questions?Discussion/Implications?
Examples of Different Models for Scoring Componentsin HPSA and MUA Designation:Discussion Paper for the Negotiated Rule Making CommitteePrepared by:National Center for Health Workforce Analysis,Bureauof Health Professions, HRSA14April 2011
13
0
Embed
Upload