Docstoc

Case Mix Adjustment and Hospice Costs

Document Sample
Case Mix Adjustment and Hospice Costs Powered By Docstoc
					The Medicare Hospice Payment System:
     A Preliminary Consideration
       of Potential Refinements


             Nancy Nicosia
            Academy Health
             June 25, 2006
               Acknowledgements

• Co-Authors
   –   Melinda Beeuwkes Buntin
   –   Elaine Reardon
   –   Karl Lorenz
   –   Joanne Lynn

• Project sponsor
   – Medicare Payment Advisory Commission
                    Motivation

• Current payment system
  – Based on HCFA study conducted in 1980-2
  – Per diem payments based on four categories of
    care
  – Largely unchanged since implemented in 1983

• Recent changes in hospice industry
  –   Rapid growth in providers, patients and costs
  –   Cancer patients no longer a majority
  –   Changes in cancer treatments
  –   Declining length of stay
                  Questions

1. How well does the per diem system reflect
    current hospice costs?

2. Should case mix adjusters be considered (e.g.
    diagnoses)?

3. Are the beginnings and ends of hospice stays
    more intensive?
Advantages of Data Provided by Large,
For-Profit Hospice Chain over Medicare

• Contains data on the number, frequency, timing,
 and duration of visits, and the type of staff
 providing them

• Additional patient-level data on marital status,
 nursing home residence, and discharge status
       Limitations of Hospice Chain Data
• Covers only 6 percent of hospice population;
 approximately two dozen sites
• Different patient mix
   – Less lung cancer, debility patients
   – More cardiovascular, cerebrovascular,
     neurodegenerative patients
   – More patients 85 and older
• Different practice patterns: greater use of inpatient care,
 no respite care
• Must still impute direct patient care costs (i.e. wages for
 visits); labor costs of inpatient care not fully captured
Q1: The Per Diem System Reflected in
 Current Visit, Visit Costs Patterns
     Percent of Variance Explained   100


                                      75
                 (R-sq)




                                      50


                                      25


                                       0
                                           Number of   Visit Labor
                                             Visits      Costs
Q2: Potential Case Mix Adjusters Explain Little
              Additional Variance
                             100
       Percent of Variance
        Explained (R-sq)
                              75


                              50


                              25


                               0
                                   Number of Visits Visit Labor Costs
                                      Days of Care (by Type)
                                      Demographics/Diagnoses
                                      Days, Demographics, and Diagnoses
                                                Ca
                                                   r di
                                                        ov                           Total Visit Costs
                                               Ce          asc
                                                 r eb          ula
                                                      r ov         r
                                                           as
                                                              cu
                                                                 lar

                                        Ill-                              HIV
                                             de
                                                f in
                                                     e      dD
                         Ne                                   eb
                           uro                                            ilit
                                                     de                        y
                                                          ge
                                                            ne
                                                                  rat
                                                                      ive
                         Lu
                                              ng
                                                 ,   lar
                        Gy                              y   nx
                           n                  ec                  ca
                                                   olo                 nc
                                                          gic               er
                                                                al
                                                                     can
                                                                           ce
                        He
                                                 Br
                                                   ea                            r
                                                            st
                          ma
                                               tol             c     an
                                                  o   gic                 cer
          Kid                                                al
                            ne                                    can
                              y,                                          ce
                                                 bla                            r
                                                       dd
                          Co           er
                                          ca
                             l or
                                 ec          nc
                                    tal         er
                        Ot                ca
                           he                nc
                             rG                 er
                                  ast
                                      ro.
                                          ca
                              Pro            nc
                                   sta          er
                                                                                                                Disease Category Small




                                       te
                                          ca
                                             nc
                               Ot               er
                                                     he
                                                       rc
                                                         an
                                                                       ce
                                                                          r
                                                                                                         Q2 (con’t): Differences in Predictions by




                         Days of Care Model
Days, Demo., Dx Model
 Q3: More Intensive Care is Delivered at
 Beginning and End of Hospice Stays

                     2.5
                                                         Average
                      2

                     1.5                                 First 3 Days*


                      1
                                                         Middle Days
                     0.5

                      0                                  Last 3 Days **
 Mean       Median         Mean       Median
Visit Labor Costs          Number of Visits
                                       *including pre-admission visits
                                       **excludes those discharged alive
                   Conclusions
• The current per diem reflects resource utilization in
 this hospice chain well; perhaps because chain has
 adapted practices to payment system parameters

• Potential case mix adjusters add little explanatory
 power conditional on days of care

• Greater compensation for the first and last days of
 hospice care before death could be warranted

• Results should be validated with more representative
 data set, with complete patient-level costs
        Sample Statistics: Demographics

Table A-1: Demographics
Age Under 65               2820     4.10    60251     5.10
Age 65 to 74               12221   17.78   249263    21.10
Age 75 to 84               25405   36.97   447308    37.87
Age 85 & over              28279   41.15   424434    35.93
Divorced/separated/widow   40183   58.47      n.a.    n.a.
Married, living together   23000   33.47      n.a.    n.a.
Single                     5542     8.06      n.a.    n.a.
Asian                       691     1.00     6688     0.57
Black                      7960    11.58    90425     7.68
Hispanic                   7807    11.36    15541     1.32
Other                       422     0.61     11417    0.97
White                      51846   75.44   1053159   89.46
Female                     41077   59.77   680877    57.64
Discharge Status (Died)    62355   90.73   978371    82.82
Nursing Home               19746   28.73      n.a.    n.a.
           Sample Statistics: Diagnoses
Table A-2: Diagnoses
Cancer – Breast                 1672     2.43   30248     2.61
Cancer – Colorectal             2720     3.96   34191     2.95
Cancer – Gynecological           1105    1.61   17988     1.55
Cancer – Hematological          1721     2.50   19284     1.66
Cancer – Kidney, Bladder        1254     1.82   23012     1.99
Cancer – Lung, larynx, pleura   6652     9.68   135228   11.67
Cancer – Other Gastroint.       3866     5.63   67417     5.82
Cancer – Other                  2749     4.00   25601     2.21
Cancer – Prostate               1648     2.40   34194     2.95
Cancer (Medicare only)           n.a.    n.a.   105523    9.11
Cardiovascular                  9768    14.21   150538   12.99
Cerebrovascular                 5880     8.56   68981     5.95
HIV                              415     0.60    5656     0.49
Ill-Defined Debility            6197     9.02   124469   10.74
Neurodegenerative               13602   19.79   163547   14.12
Other Diagnosis                 4369     6.36   63392     5.47
Respiratory                     5106     7.43   89163     7.70
Regression Results: Number of Visits
Table A-3: Number of Visits
                      Model 1      Model 2    Model 3
CC Days                3.96***          ---    3.82***
GIC Days               1.13***          ---    0.99***
RHC Days               0.57***          ---    0.57***
Demographics                  No      Yes         Yes
Diagnosis                     No      Yes         Yes
Year FE                       No      Yes         Yes
Program FE                    No      Yes         Yes
Adj R-sq                0.914        0.111      0.891
Sample                 68,725       68,725     68,725
Regression Results: Visit Labor Costs
Table A-4: Visit Labor Costs
                      Model 1    Model 2     Model 3
CC Days              300.21***        ---   295.27***
GIC Days              34.52***        ---    27.32***
RHC Days              10.91***        ---    10.79***
Demographics               No       Yes          Yes
Diagnosis                  No       Yes          Yes
Year FE                    No       Yes          Yes
Program FE                 No       Yes          Yes
Adj R-sq                 0.906     0.075       0.879
Sample                 68,725     68,725      68,725

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:118
posted:4/12/2008
language:English
pages:15