Skip Navigation Archive: U.S. Department of Health and Human Services U.S. Department of Health and Human Services
Archive: Agency for Healthcare Research Quality
Archival print banner

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to for current information.

CareScience Risk Assessment Model - Hospital Performance Measurement

Presentations from a November 2008 meeting to discuss issues related to mortality measures.

Appendix C — Select Practice Formulas

  1. Determine case-weighted arithmetic means by disease (3-digit ICD-9 code, k) for
    • mortality (Tk)
    • morbidity (Bk)
    • complications (Ck)
      Note: Records without a risk score are excluded.
  2. Compute risk-adjusted rate for each facility (h) for each adverse outcome:
    • Tkh = Tk + ΔTkh , where Δ indicates mean deviation (actual - risk)
    • Bkh = Bk + ΔBkh
    • Ckh = Ck + ΔCkh
      Note: Tkh, Bkh, and Ckh can never be negative.
  3. Compute quality indicator by disease k for each facility (h) based on CHRP weights:
    • Qkh = 0. 46(Tkh)0.96 + 0.29(Bkh)0.91 + 0. 25(Ckh)0.94
      Note: Smaller Qkh denotes higher quality.
  4. Compute efficiency ratio for length of stay by disease k for each hospital h:
    • RLkh = exp{(logLOS)kh - (logLOSrisk)kh} for hxk level means
      Note: Smaller RLkh denotes higher efficiency.
  5. Cross tabulate Qkh and RLkh by quintile across facilities for each disease k.
    Note: This is a 5 x 5 cross-tabulation matrix.
    Small facilities with less than 100 cases for disease K were excluded from the tabulation.
    No matrix is constructed with fewer than 200 qualifying facilities.
  6. Identify the select practice facilities for each disease k (set H*k) by choosing facilities falling into the top 2x2 cells in the 5x5 cross-tabulation matrix.
    Note: Select practice represents roughly 16% (4/25) of cases for each disease k. The reason for the imprecision is that select practice is a facility characteristic, requiring all cases from a given facility be sorted into one quintile only.
    For the 50 highest-volume diseases, the correlations between quality (Qkh) and efficiency (RLkh) are weak.
  7. Compute select practice risk for each disease k as
    • T*k = case-weighted mean Tkh for all h Î H*k
    • B*k = case-weighted mean Bkh for all h Î H*k
    • C*k = case-weighted mean Ckh for all h Î H*k
    • RL*k = case-weighted geometric mean RLkh for all h Î H*k
    • R$*k = case-weighted geometric mean R$kh for all h Î H*k
      Note: Compute ratio for total costs by disease k for each hospital h:
      R$kh = exp{(logCost)kh - (logCostRisk)kh} for hxk level means
  8. Compute the scaling factor for each outcome as the ratio of select practice to mean:
    • Rtk = T*k /Tk for all k
    • Rbk = B*k / Bk for all k
    • Rck = C*k / Ck for all k
    • Rlk = RL*k for all k, where l indicates LOS
    • R$K = R$*k for all k, where $ indicates Total Costs
  9. ICD-9 codes that fail to pass Step 5 are rolled up into one of 18 major diagnosis groupings (Dx_Groups) of the ICD-9-CM classification system. For example, ICD-9 codes 460 to 519 (excluding the six codes that could stand alone) are rolled up into Dx_Group 8, which represents the diseases of the respiratory system. The scaling factor for each Dx_Group is computed as outlined by Steps 5 to 8. ICD-9 codes rolled up into a broader Dx_Group share the same scaling factor.

C.1 Applying the Scaling Factors

The purpose of the scaling factors is to generate risk scores based on Select Practice hospitals. Deviation (Actual - Risk) and significance flag can then be computed on the new risk scores. The deviation and significance flags indicate the facility's performance level compared to the nation's Select Practice facilities for the particular disease group.

1. Compute risk score and stderr of Select Practice at patient level:

Ln_los_risk_select = Ln_Rlk + Ln_length_of_stay_risk
Ln_los_risk_stderr_select = Ln_Rlk + Ln_length_of_stay_risk_stderr
Ln_cost_risk_select = Ln_R$k + Ln_cost_risk
Ln_cost_risk_stderr_select = Ln_R$k + Ln_cost_risk_stderr
comp_risk_select = comp_risk * Rck
comp_risk_stderr_select = comp_risk_stderr * Rck
mort_risk_select = mort_risk * Rtk
mort_risk_stderr_ select = mort_risk_stderr * Rtk
comp_morb_risk_select = comp_morb_risk * Rbk
comp_morb_risk_stderr_select = comp_morb_risk_stderr * Rbk

(R: Scaling Ratio; k: 3-digit ICD-9 code; L: LOS; $: cost; c: complication; t: mortality; b: morbidity)

2. Report the Select Practice deviation and significance flag at the aggregate level:

Use the same method as the standard-practice report.

Return to Article Contents
Proceed to Appendix D


Page last reviewed March 2009
Internet Citation: CareScience Risk Assessment Model - Hospital Performance Measurement. March 2009. Agency for Healthcare Research and Quality, Rockville, MD.


The information on this page is archived and provided for reference purposes only.


AHRQ Advancing Excellence in Health Care