data weighting committee report n.
Skip this Video
Loading SlideShow in 5 Seconds..
Data Weighting Committee Report PowerPoint Presentation
Download Presentation
Data Weighting Committee Report

Loading in 2 Seconds...

play fullscreen
1 / 29

Data Weighting Committee Report - PowerPoint PPT Presentation

  • Uploaded on

Data Weighting Committee Report. For Discussion May 18-20, 2011 By Negotiated Rulemaking Committee on HPSAs and MUAs. Where we are and where we have come from. Variables have been proposed and considered

I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
Download Presentation

PowerPoint Slideshow about 'Data Weighting Committee Report' - clancy

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.

- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
data weighting committee report

Data Weighting Committee Report

For Discussion May 18-20, 2011

By Negotiated Rulemaking Committee on HPSAs and MUAs

where we are and where we have come from
Where we are and where we have come from
  • Variables have been proposed and considered
  • Probable useful variables are being assessed – their quality, utility and feasibility; their distributions and other characteristics
  • We have looked at several approaches to combining variables
      • Combining into an index (we have used “factor analysis” for this),
      • using menu,
      • considering “bonus points,”
      • considering thresholds.
  • Discussion of “factor analysis” – when is it appropriate?
  • Framework
  • Components – candidate variables
  • Weighting of components (and their elements) within Framework
variables 1
Variables –1

Access to Care Barriers --

  • Race – Nonwhite (…& non-Hispanic)
  • Ethnicity (Hispanic)
  • Linguistic isolation/Limited English Proficiency
  • “Travel” -- Distance/time and geo barriers Rural/Frontier – can use mapping resources online, &/or consider RUCA score
  • Population Density – can be modifier of thresholds or P2P
  • Disability (physical, cognitive; overall measure or specific?)
  • Other characteristics/conditions -- LGBT, other characteristics, HIV/AIDS or disease conditions with stigma and/or need for expertise with primary care
  • Other? (check barrier group list; consider ECONOMIC BARRIERS as ABILITY TO PAY measures)
variables 2
Variables – 2

Ability to Pay

  • Uninsured – demonstrated strong factor
  • Low Income (<200 FPL, or between 100-200 FPL; American Community Survey has data up to 500% FPL)

Other measures that have been proposed:

  • RWJ County Indicators: Income inequality/GINI index
  • Regional cost of living variations – not agreed to
  • Income: per capita income or median household income
variables 3
Variables - 3
  • Socio-economic indicators of health -- Index:

(combined into “index”)

    • Poverty
    • Less than high school education
    • Adults not employed
    • Single parent household
  • Direct measures of health (menu – SMR plus another from MENU)
    • SMR
    • Low birth weight / Infant mortality rate
    • Chronic disease – diabetes
    • Preventable/ACSC Hospitalizations
analytical work by jsi
Analytical Work by JSI

“Factor analysis” – considering those proposed to reflect underlying “underservice” concept

THREE “factors” (clusters of variables) identified

(examine the results – request Eric present JSI analysis)

framework for discussion hpsas
Framework for Discussion: HPSAs


Determine P2P and Health Status/Barriers/Ability to Pay Levels for population in a Rational Service Area.

Step 1. EXTREME SHORTAGE: If P2P (reflecting age/gender structure of population) shows extreme shortage (<1:2 x Y where Y is the either national norm or selected threshold for “appropriate” population per provider), designate as HPSA.

Step 2. POTENTIAL SHORTAGE: If (a) P2P is between 1:Y and 1:2Y, consider first for total population, for geo HPSA (then, if not qualifying, consider population HPSA):

b. Health status (SDI 4 variable index and direct measures – SMR and others)

c. Ability to pay (low income and/or uninsured and/or not employed 18-64

d. Access barriers (select two or more(?):

  • Race AND
  • Time/distance to care and/or
  • Limited English proficiency (LEP) or Linguistic Isolation %
  • Disability prevalence
  • Other

Step 3. NO GEOGRAPHIC SHORTAGE: If not geo HPSA, consider population HPSA

framework for discussion mua p
Framework for Discussion: MUA/P

(Option 1) Create an index (index of medical underservice) using scores for each of the 4 factors, each ranging from 0 (best) to 100 (worst), for rational service area (same definition as for HPSAs).

Sum of the scores may be weighted as follows:

(TENTATIVE – to be refined with analysis of factors considered to be representing “underservice,” review of literature, and discussion of values – policy decision to be made)

  • Health Status: 30% (Health status is made up of 50% SDI + 50% direct measures – potentially SMR plus selected one or two)
  • Access barriers (race, time/distance, LEP, other?): 20%
  • Ability to Pay (uninsured, low income): 30%
  • Provider Availability for Population of RSA (P2P): 20%
mua new index of medical underservice
MUA: new index of medical underservice

For rational service area – evaluate MUA and if area does not qualify, consider MUP options.

MUA components:

(a) Health Status(SEIH and direct measure(s) (use SMR & one or more of LBW or IMR or diabetes prevalence or Ambulatory care sensitive conditions (ACSCs) or ?; most other candidate measures are not as generally available for all geographies so some “places” would be unable to use them, and they won’t be available for testing )

(b) Access barriersincluding Race (non-white), and one or more of the following: linguistic isolation or LEP, portion of the population with geographic barriers or excessive travel time to care, and disability prevalence.

(c) Ability to pay: uninsured &/or low income percent (<200% of poverty – or use just the percent between 100 and 200% of poverty threshold)

(d) Population to provider ratio– pop will be weighted to reflect age and sex composition of the RSA (using MEPS – some of us recommended and thought it had been agreed by large committee that we’d use the age/sex use rates of people without impeded care which is about 10% higher than impeded) – the age/sex adjustment may not be possible for special pops MUPs because we don’t always know their age/sex distribution.

mua scoring of the components
MUA scoring of the components:

Could be based on:

  • our “expert opinion” – such as 30-20-30-20 (replacing old scales) -- high score == high need.
  • potential ultimate “factor analysis” if we can agree on if/how these components contribute to an underlying “underservice” concept (see JSI analysis and Hofer et al based on MEPS individual level analysis of factors predictive of more use or less use (i.e., reflecting EXPRESSED NEED or ACCESS BARRIERS)

The barriers and ability to pay components can be used in one of several ways:

(a) to discount the apparent capacity (P2P) reflecting that the apparent capacity isn’t as available as it appears or

(b) to add to the score for the health status/need dimension reflecting that there are even greater needs to be met than the health status component suggests.

  • Maintaining HPSA and MUA designations as distinct
    • HPSAs for provider availability
    • MUAs for facilities/infrastructure needs of a population
  • Using variables defined in same way for both, weighted differently
  • Both will require coming to terms with identifying an acceptable P2P ratio, which means both
      • Counting PC providers to reflect need with/without federal resources
      • Pointing to a norm or acceptable level for P2P

(Deemed “sufficient” even if health status of population is poor, and access and economic barriers are present. Could be national average recalculated annually, or a preferred ratio based on evidence, which may also take into account different models of service provision over time.)

options for combining factors
Options for combining factors
  • HPSA: geographic assessment first – consider low income or other special pop HPSA if geo area does not work.

INDEX – relies on factor analysis – variables that in combination show one DIMENSION of interest.

Score associated with index factor #1? (could be 0 – 50 points)

MENU – relies on analysis of variables that are not necessarily related to each other, but are direct or indirect measures of a factor -- such as direct measures of health status: SMR, diabetes prevalence, or low birth weight. Pick for example SMR and one OR the other – depending on which is more reflective of your area’s health status.

…or such as “barriers to care”

Examples: Race + one: Limited English Proficiency, Travel time excessive, or Ethnicity

scoring of the components
Scoring of the components:

Could be based on

  • our “expert opinion” – such as 40-30-20-10
  • potential ultimate “factor analysis” if we can agree on how these components contribute to an underlying “underservice” concept – if we are satisfied with the statistics
  • combination – useful if we have some variables with statistics we feel are robust in the application stage, but we don’t have a statistical approach for combining either variables into factors, or factors into total score
framework and graphic presentation
Framework and Graphic Presentation
  • Adjusted graphic (SH) having highest need (fewest providers per 1000 (x-axis) and lowest health status, ability to pay and access (y-axis)) in lower left by the “origin” will let us look at the relationship of provider availability to composite need measure (like economic supply/demand curves).
  • Consider that slope of line and distance from origin may be adjusted as norms change, but actual resources available would be represented by another overlay that could follow the slopes of the lines or could be closer or farther from the origin to include more or fewer designated areas as far out as the designation lines permit.
  • Being sure that the designated places are inclusive of those with greatest shortages/underservice (or would without existing public resources) is the goal.

Graphic Reflecting Continuum of “Barriers and Health Status” (Y Axis) and P2P (X Axis) (Ability to pay factor may be included with factors on either axis or be a third dimension)

p2p considerations
P2P Considerations
  • Including NP/PA/CNM may strongly affect P2P so may need to be re-scaled.
  • CONSIDERATION offered in a meeting: include NP/PA/CNMs in areas with physicians available at least as >50% of the count – on assumption that areas with more than half (or other proportion) of primary care capacity provided by non-physicians may need assistance in recruiting/retaining PC physicians.
  • Predominance of NP/PA/CNM likely pertains in RURAL/FRONTIER areas.
health status and barriers y axis
Health Status and Barriers (Y Axis)

Y axis: represents components of “need” – measured by risk factors (SEIH) of poverty, single parent household, educational attainment < high school, and not being employed, which may limit current use as well being predictive of worse health status which is indicative of need for services, as well as direct measures of health status indicative of need for services (and likely higher use than average); additional barriers to access and additional measures of potential higher than average need could be “added” into this dimension.

x axis p2p population to provider ratio
X-Axis P2P Population to Provider Ratio
  • Express in terms of PROVIDERS (weighted) per 1000 people (adjusted)? (avoids the pop/providers where provider count is zero – unless we make provision for this situation)
  • Or people (adjusted count) per provider (weighted)?
  • Final determinations of inclusion rules
  • Testing of impact of counting NP/PA/CNMs (always or only where there is physician capacity also?)
  • What is the norm or level deemed sufficient regardless of poor health status? (1:1000? Ref.?)
  • What shortage is extreme enough to warrant designation without having to demonstrate bad health status and/or other access barriers or need?
  • Those factors/variables associated with likely higher use driven by need (in spite of possible barriers) could be moved over to the P2P axis. This could be conceptualized as weighting the population to reflect the higher need – not accounted for in adjusting for age and gender. Assumes higher NEED requires MORE provider capacity or infrastructure.
  • However Committee has liked “seeing” need (worse health status – risk factors and direct measures) on the Y axis.
  • OPERATIONAL PROCESS: will be stepwise taking into account the FOUR DIMENSIONS
issues outstanding
Issues Outstanding
  • Being sure to account for areas with needs to which resources “should” be directed in conditions of relative plenty.

(that is, don’t be overly restrictive).

  • Deciding on total points for each dimension and on relative points for the factors within each component.

Data Weighting Sub-Committee Report

issues continued
Issues continued
  • Deciles – useful for viewing national distribution – but more of the information available in the measures will be useful in the actual application. Plan to use the information available rather than the strata or rank?

Counties: 10% of counties in each decile regardless of population?

PCSAs? 10% in each decile regardless of population?

Or weight by population.

Or use natural breaks in the distribution – use the top 10% of the range for the top decile etc. (extremes can be truncated), more appropriate for rarer or skewed characteristics

  • Examine the sensitivity of leaving out nominated variables – if they are already well represented they’ll not add much new information.
more to come
More to come!
  • Get input from committees on variables and their contributions
data weighting sub committee
Data Weighting Sub-Committee
  • Members/Participants

Babitz, Scanlon, Hawkins, Holloway, Owens, Diaz, Camacho, Phillips, Taylor, Jordan, Lee, Proser, Rarig (chair)

  • Meetings (April-May 2011)
  • Additional materials

Hofer, Abraham, Moscovice paper; JSI summary of variables status; graphic presented at April meeting (Holloway), Weighting Considerations.xls (Rarig, Holloway criteria), socio economic indicators of health (SEIH) and direct measures of health status and barriers.