Docstoc

Geographic Variation in Health Spending Is Medicare a Good Proxy

Document Sample
Geographic Variation in Health Spending  Is Medicare a Good Proxy Powered By Docstoc
					Geographic Variation in Health Spending:
         What is it measuring?



              Louise Sheiner
              March 10, 2010
                           Conventional Wisdom from
                           the Dartmouth Atlas Project


• Medicare spending per beneficiary varies widely by geographic area
    – Average per beneficiary cost in 2004:
       $8,659 in Louisiana vs. $5,640 in South Dakota

• The variation in spending is only partly explained by differences in
  demographic factors, prices, and income.

• The remainder of the variation is attributed to differences in “practice
  styles”

• High-spending areas have same or lower quality as low-spending areas.

• Policy implications:
    – There is a lot of inefficiency in Medicare
    – We can significantly reduce spending and increase quality of health care if
      higher-cost states adopt practices of lower-cost states.
  Policy Implications : Can expand coverage AND
                    save money


• Sutherland, Fisher, and Skinner, New England Journal of Medicine, 2009.
  “Evidence regarding regional variations in spending and growth points to a
  more hopeful alternative: we should be able to reorganize and improve care
  to eliminate wasteful and unnecessary service.”

• Fisher, Bynum, and Skinner, New England Journal of Medicine, 2009:
  “But a careful look at variations in spending growth and spending patterns
  among U.S. regions reveals a more optimistic picture. By learning from
  regions that have attained sustainable growth rates and building on
  successful models of delivery-system and payment system reform, we
  might, with adequate physician leadership manage to “bend the cost
  curve.” ….. .. Reducing annual growth in per capita spending from 3.5%
  (the national average) to 2.4% (the rate in San Francisco) would leave
  Medicare with a healthy estimated balance of $758 billion, a cumulative
  savings of $1.42 trillion.”
                    Too good to be true?
              Questions addressed in this paper
• Does the Dartmouth research correctly characterize the variation in
  Medicare spending?
    – They compare Medicare spending across geographic areas, and do some
      adjustments for sex, age, race, price, and health.
    – Anything not explained by these factors – the residual – is deemed to be a
      region-specific factor attributable to practice styles.
    – Because they are focusing on the residual, it is important to control as
      accurately as possible for factors that might influence spending.

• Does the variation in non-Medicare spending mirror that of Medicare
  spending?
   – If not, is the variation in Medicare spending likely attributable to
     practice styles?

• What other factors explain the variation in non-Medicare spending?

• How does this change our view of health reform and the potential for cost
  savings?
                      My Approach

• I use CMS National Health Account Data on health
  spending by state
• I examine the variation in average Medicare spending
  by state (no individual-specific data) and control for
  attributes of the state.
• I focus only on physician, other professional, and
  hospital spending.
   – I want to compare to private nonelderly spending and
     privately-insured spending.
   – Discussion of practice styles is about hospitals and
     physicians.
States differ substantially on health measures
  25

                                                                                          WV
                                                                                               MS

                                                                     KY

                                                                          TX AL
  20




                                                 AR             OK
                                                                                TN
                                  NM                                           LA
                                         NV                               NC
                                                        CA
                                                         NY                          SC
                                                                IN
                                                           FL
                                               NJ IL PA
                                               ME
                                                MO
                                  OR                                      GA
  15




                                       RI AZ                   OH
                                                               MI
                                                          DE
                MT              KSWA
                                                  VA
                              UT
                             IA SD ID
                         WY
                     VT ND
                         MA
                         WI
                                 NE
                                                   MD
       CO
                        CT
                            NH
  10




                MN

            5                 6                    7              8                   9         10
                                                  rate of diabetes
      Medicare Spending as a Share of Per Capita
              Income varies with illness
                                                                        LA
                                                                                                MS
.25




                                                                                          WV


                                             AR                                      SC
                                                                  TX
                                                       OK    KY
                                                     MI
                                                    FL                 AL

                                                                  GA
  .2




                                       AZ   MO          OH                  TN
                             KS                                   NC
                                                        IN
                                                PA
                                               MD DE
                                                   NY
                                             IL  CA
              MT
                     ND    UT NE
                            NM ID
                                  NV NJ
                            OR
                               RI    ME
                      WI IA SD
                      MA
                VT
         CO
               MN                 WA
.15




                                             VA
                      WY
                     CT NH

         5            6                7              8                          9             10
                                   Rate of Diabetes, 2004
           Medicare Spending Regressions
Most of the variation in spending can be accounted for
  Cross-state Variation in Acute Medicare Spending, 2004
                                                             Add
                       No               Add     Add     demographics,
                     Controls Add age income   health     uninsured



Average              $5,950 $5,950 $5,950 $5,950           $5,950

Standard Deviation    $711    $633    $577     $408         $257

10th Percentile      $5,083 $5,146 $5,114 $5,482           $5,638

90th Percentile      $6,939 $6,999 $6,810 $6,445           $6,271

Lowest               $4,729 $4,702 $5,016 $5,235           $5,182

Highest              $7,521 $7,279 $7,269 $6,966           $6,498

Range                $2,792 $2,577 $2,253 $1,731           $1,316
       Does Unadjusted Spending Identify the “High-
              Spending” States Correctly?
6500




                                                                       MI

                                                     KS
                                               NE                           MA
                                       MN AZ                                TX

                                       CO               MS             DE         LA
                                               KY WV                                    NJ
                                                                                          MD
6000




                       VT    ND                  SCMO             CA
              UT      MT                           NVOH
                          WY                                                      NY
                   ID   ME                                                   FL
                           NH                                PA
                                            ARNC AL OK
                             WI               IN                   CT
                        SD                      TN
                        IA    WA                             IL
                             OR
                                                    GA
5500




              NM

                                               RI


                                  VA
5000




       4500    5000          5500      6000      6500      7000                        7500    8000
                              Unadjusted Medicare Spending
    Why can I explain so much more of the variation than the
                       Dartmouth folks?
•    Much of the Dartmouth literature did not control for health attributes. But, a
     2009 study did control for a number of health measures, as well as measures
     of income and poverty, at the individual level.

•    The conclusion: “More than 70% of the differences in spending cannot be
     explained away by the claim that “my patients are poorer or sicker”.”

•    They suggest that “social capital” may be the important factor.

     “States with high levels of social capital tend to legislate more comprehensive
     support programs for their residents and pay health care providers more,
     while the physicians who live in these high social-capital states are more
     likely to adopt new and effective innovations rather than simply performing
     more tests and procedures with questionable medical efficacy.” (Skinner,
     Chandra, Goodman, and Fisher, December 2008)
                          Why are my results different?
•   I control for health attributes of the population at the state level. Does it
    matter whether the variables are controlled at the micro vs state level?
     –   Without omitted variable bias, no.
     –   With omitted variable bias, probably yes.


•   Once diabetes is controlled for at the individual level, any additional
    explanatory power of the state’s mean diabetes rate can’t be picking up the
    effects of diabetes.

•   But, if places with a higher incidence of diabetes also have a higher
    incidence of hypertension, for example, then putting in the mean diabetes
    rate of a region will pick up some of the effects of hypertension as well.
     –   If included health variables are correlated with excluded health variables—and if that the
         correlation is greater at the state level—then state level health variables will pick up the
         effects of health more.


•   In addition, to the extent that practice styles reflect the needs of the median
    patient, then states with sicker people will adopt more interventionist
    practice styles.
“Social Capital” is extremely correlated with other measures that can
                 affect health and health behaviors
  10


                     MS

                          WV

                          SC
    9




                         TN
                        LA
                      AL
                     GA     NC   TX
                            KY
    8




                                        OK
                                       OHIN
                                           MI
                                  FL
                                    NY     DE
                                       CA
                                       PA
                                      MD
                                 AR VAIL
    7




                                              MO
                                   NJ                  ME
                NV                           AZ
                                          RI ID          WA
                                                                NE
                                    NM              KS OR                   SD
                                                      UT
                                                          NH IA
    6




                                                 MA    WIWY
                                                  CT                        ND
                                                                       VT
                                                                     MN
                                                                     MT
    5




                                                      CO

        -2           -1                    0                1                    2
                                      socialcapital
           Measure of social capital and exercise



                           KY
65



                                  TN

                            LA
60




     MS          NC
                       WV             OK
            AL
                                    GA
                                   FL
55




                                             NJ                      IA
                      AR    TX
                             SC                   IL
                                                  MO
                                            NY                      KS
                                                IN
                                             OH               MI
                                              MDPA                                     ND
50




                                       VA       CA                             MN
                                                               RI
                                                      AZ CT
                                                     MA

                                                     WA
45




                                                     CO            OR
                                                               WI UT         NH
                                                                           WY
                                                                          VT
                                                                                  MT
40




      .2          .3               .4            .5                       .6            .7
                            most people can be trusted
                 Comparing the effect of “social capital”
                    with other regional attributes


                            Log Acute Medicare Spending, 2004


Social Capital                 -0.08**        -0.09**           -0.02     0.02
Log per cap income                             0.46**           0.62**   0.53**
Diabetes                                                        0.07**
Sedentary                                                                0.005*
Poor Health                                                               .01*
Black                                                                    .56**
Share 6574                                                                1.54
Share 7584                                                               5.58**



Rsq                              0.24           0.52             0.66     0.77
    Does non-Medicare spending exhibit the same
     geographic variation as Medicare spending?
•   Determinants of spending similar:
     – Demographics
     – Prices
     – Practice Styles?

•   But could differ because:
     –   Differences in Medicare and private incentives
     –   Concentration of health insurance and hospital industries
     –   Cost shifting
     –   Paid privately instead of through government (uninsured, underinsured, demand)

Two Measures of non-Medicare spending

     – Total acute health spending by state less estimated spending of Medicare
       beneficiaries divided by non-Medicare population

          • This measure has been criticized because much of it is estimated by CMS, and also it
            includes Medicaid spending, which differs by state

     – Employer-sponsored health insurance premiums by state
Health spending of the nonelderly also varies by state, and
 average health spending is correlated with private health
                   insurance premiums
4500




                                             RI



                                                                          MA     ME
                                             NH                             VT
4000




                        MI                                           WI
                     NV       NJ
                              VA           CT    NY
                     FL                                    MN   DE
                        TX         SC     IL    OH              WY
                                    OR CO MDKS        NE
                                        PA       MT    WV
                        OK               TN WA
                          MS            IN
                                           KY IA
                                            MO
3500




                        CA NC
                      LA
              AZ     ID                               SD
                      AL
                       NM
                      GA                                   ND
                     AR
3000




               UT

       2000          2500                    3000                         3500
                    Nonelderly per capita health spending
       Medicare spending is not correlated with the health
                  spending of the nonelderly

                           ME
3500




                          VT
                                                                           MA

                               WI
                            WY                                        DE
                                ND MN
                            SD                  NE WV
3000




                          MT                                                    NY
                            IA NH                    OH
                                                 RI
                                 WA                                   CT
                                                    KS
                                                KY MO
                                                 TN              IL                   MD
                                                IN               PA
                                      CO
                                OR                   SC
                                 VA                                                  NJ
                                                NC
2500




                                                            MS
                                                       OK
                                                      GA          CA MITX
                  NM ID                    AR        ALNV                FL     LA



                  UT
                                           AZ
2000




       4000        5000               6000             7000                                8000
                    Per beneficiary acute Medicare spending
       Medicare spending is somewhat correlated with private
                   health insurance premiums
4500




                                                 RI



                               ME                                  MA
                              VT NH
4000




                                     WI                         MI
                                      VA               NV      CT       NY NJ
                                       MN                       DE FL
                                 WY                SC OH   IL     TX
                                   OR CO        NE KS                        MD
                              MT                    WV     PA
                                    WA           TN OK
                                                IN      MS
                                IA              NC MO
3500




                                                KY            CA
                                                                        LA
                         ID     SD         AZ
                   NM                                 AL
                                     ND               GA
                                            AR
3000




                    UT

          4000      5000               6000             7000                      8000
                     Per beneficiary acute Medicare spending
        Explaining variation in non-Medicare spending

• Premiums and non-Medicare spending from national
  accounts are well correlated.
• Private health insurance premiums are explainable through
  demographics and per capita income. Health variables don’t
  matter.
• Health spending of the nonelderly (which includes private
  health insurance, out of pocket spending, Medicaid, and
  uncompensated care) depends on the rate of uninsurance,
  demographics, and income. Health variables also don’t
  matter.
• There is very little correlation between Medicare spending
  and non-Medicare spending, with or without adjustments.
  But, there is correlation between Medicare and nonMedicare
  hospital spending.
               Nonelderly health spending affected by
               income, demographics, and uninsured

                             Log Nonelderly Health Spending, 2004


Log Private HI Premium          1.01**        0.73**        0.62**
Log per cap income                                          0.44**   0.51**
Rate of Uninsured                             -0.02**       -0.01*   -0.01**
Black                                                       -1.3**   -1.2**
Urban                                                      -0.63**   -0.59**
Black*Urban                                                 1.6**      1.4
Poor Health                                                 0.0003   0.0007
Share female 19-29                                          8.3**    5.9**
Share <19                                                   -1.7*    -2.7**


Rsq                              0.29          0.53          0.75      0.7
              Private health insurance premiums vary by
                      income and demographics



                                 Log Private HI Premium, 2004



Log Nonelderly Spending           0.31**       .22**        .13*

Log per cap income                             .21**       .15**   0.14**

Share of Workers 19-29                                     -3.2*   -4.4**

Female 19-29 Share of Workers                               4.1     6.1*

Black                                                      -0.08   -0.14

Poor Health                                                0.002   -1E-05



Rsq                                0.29         0.46        0.59    0.57
Medicare/NonMedicare Comparisons
              Adjusted Medicare vs. Adjusted
8000       Private Insurance Premium, 2002-2004

                                                                                         NJ


                                                LA
7000




                                   MD

                                                                 NY            MA        MI
               CA                                               TX        FL
                                                      PA
                                                                                              DE
6000




                                    KS
                                                     MS                             NV
                                             SC                 CT
                    MO                     OK        OH    NE                            IL
          AL         AZ       KY                                        WV
                                           TN                         MN
                                     GA                          CO
                                          NC          IN
5000




                                                                     NH
                                   MT                                           WY                  WI    ME
                         IA                           WA
                               OR
                    UT           VA

                                   NM
4000




       3000              3200             3400          3600         3800                          4000
                                   Adjusted Private Insurance Premium

       Adj Medicare = 3538+.61*adj private health insurance premium
       Rsq = .01
                     Adjusted Medicare vs. Adjusted
                       Non-Medicare, 2002-2004
8000




                                            NJ


                                                    LA
7000




                                                          MD

                             MI                                        NY                       MA
                                             CA            TX
                                            FL
                                                   PA
                                                                                               DE
6000




                                                          KS
                                   NV                       MS
                                                         CT
                                                          SC
                                          OK                   MO     IL
                                                                      OH     NE
                       AZ               AL                 KY               WV
                                                        TN                  MN
                                       AR GA
                                                           CO
                                                            IN
                                            NC
5000




                                             NH                 RI            ND
                                                                              VT
                                                                       MT       ME   WY   WI
                                                               IAWA           SD
                                                   OR
                                       ID
                            UT    VA

                                              NM
4000




           1500           2000            2500           3000                        3500
                         Adjusted Non-Medicare Spending per Non-Elderly


       Adj Medicare = 5816+-.10*adj non-Medicare
       Rsq = 0
4500
                                            Hospitals


                                                                                  MA



                                      MI
4000




                                 NJ
                                                   LA
                            KS                TX                     NE
               UT                           MD
                                                 PA SC         MS
                                                              WV
                                           MN                           WY
                                                              VT  MT
                    AZ            KY
                                   NH
                              AR ID                                 DE       ND
                                                       NM        OH  MO
                     NV    CA  CO OK                IN
                                                    IL
                                                               WI
                                                              ME     SDNY
                                     NC
3500




                    AL                                   IA
                          TN
                           OR         CT
                                      WA
                            FL
                            GA                                               RI
                            VA
3000




         800                           1300                  1800                 2300
                                 Adjusted Non-Medicare Hospital Spending


       Adj Medicare Hospital = 3534+.14*adj non-medicare hospital

       Rsq = .03
       Physician and Other Professional Services
2200
                                                 FL
                                                      CA




                                                                NV
2000




                                      AZ
                                                                AL
                                                                CO               KS
                                                VT         ME
                                                                            MN
                                     NJ
                     MI
                      ND                                                     TN
                           NY
                           MO
                                           IA      OK
                                                   SC
                                                  NE  LA
                                NH                    GA
                                                SD                                    DE
1800




                                                MS MT
                                               AR
                                                       OH        KY
                                                                       WA
                                              CT                  TX
                                          PA                                               WI
             RI                              ID
                                NC
                                                                       OR
                                                IN
                                                 UT         WY
                                                   MD
                                                     WV
1600




                     VA
                                                 MA
                           NM                                   IL

         1000            1100         1200          1300          1400         1500
                  Adjusted non-Medicare Physicians and other Professional Spending

   Adj Medicare Physician = 1623+.17*adj Non-Medicare Physician

   Rsq = 0
              Changes in Spending



•   How do growth rates of Medicare
    spending vary across states?

•   Are the growth rates of Medicare and
    non-Medicare spending more closely
    correlated than the levels?
      Ranking of state Medicare spending hasn’t
             changed much since 1991
                                                                          MD
                                                              NJ
8.9




                                                                   NYLA

                                                         FL
                                              TX                    MA
                                                    MI              DE
8.8




                                                     CT
                                                                    CA
                                                   IL                          PA
                                        MS
                                        OK     OH
                                          KS MONV
8.7




                         SC             WV    AL GA
                                        TN
                                         KY RI
                       NE       NC IN
                                             AR         AZ
                                MN   CO
8.6




                                        VA
                          ND            WA
                            OR
                             WI
                             NH
                         SD WYIA
                       VT ME
                           MT
8.5




             ID
                  UT       NM

       7.6         7.8                 8               8.2                          8.4
              Log per beneficiary Acute Medicare Spending, 1991
.07         But there has been convergence

                        NESC


                                            TX
                                NC     MS
               ID         ND
.06




                                                            NJ
                                  IN   OK
                       VTSD OR MN      WV
                                        KS
                    UT       WI
                          ME NH       TN            MI FL
                           MT WY        KY MO     IL CT          NY
                              IA    CO         OH
                                              NV
                                             AL                        MD
                                          AR RI                   LA
.05




                                     WAVA
                            NM                    GA              MA
                                                                  DE
                                                                  CA
                                                     AZ
.04




                                                                            PA
.03




      7.6            7.8                 8               8.2                     8.4
                Log ber beneficiary Acute Medicare Spending, 1991
            Also convergence from 1997
.07                   to 2004

                                              DE
                           NE

             OR                                                       MD
.06




                      MN
               WI

            MT SD
             VT
                 WY               NC SC
                           ND                              CT
                   ME                        KS
.05




                    IA                       WV OH                                 NJ
                             WA    AZ           OK
              ID            NH          AR IN                        MI
                                  CO         KY
                                                     MO
                                                     NV         IL
                                                                          TX FL
             NM
              UT                   VA              TNAL
.04




                                                          GA                              NY
                                                                                   MA         LA

                                                          RI          MS PA
                                                                                  CA
.03




      8.1          8.2           8.3            8.4          8.5                        8.6
                     Log per beneficiary Medicare spending 1997
High spending states have lower growth rates than low-
                   spending states
             Why do the growth rate results differ from Dartmouth?

•   Fisher, Bynum, and Skinner compare $ change in spending across areas,
    rather than percent. Also, compare $ change in spending with spending at
    end of the period.
•   This comparison allows them to argue that places that spend more also
    have higher spending growth.

                                                                         dallas
      5000




                                                                                                          manhattan



                                                                                      e long island
      4000




                                                                     houston

                                                columbus
                                    charlotte                 camden                  los angeles
                                                                 chicago
                                                                    fort lauderdale
                                                st louis
                                                                 philadelphia
                                                               boston
                                               orlando
      3000




                                           nashville             baltimore
                minneapolis   milwaukee
                                     birmingham
                        albany     phoenix
                            indianapolis   cleveland
                                  kansas city
                        seattle             washington
                                                 pittsburgh
      2000




                          atlanta

                7000                8000         9000        10000        11000                       12000
                                       Per capita Medicare spending, 2006
         I compare growth rates with beginning period spending.

This is appropriate if most of the differences in spending are attributable to
demographics, income, and health.
          6




                                                dallas
          5




                        charlotte

                                     columbus
               minneapolis
          4




                                                                                       e long island
                                                                                                              manhattan
                                               st louis                  houston
                                        milwaukee                       camden
                                    albany
                                                             orlando          chicago
                                                           nashville
                                                      birmingham                        fort lauderdale
                                                     phoenix
                                              indianapolis
          3




                                                                                           philadelphia
                                                                                        boston                los angeles
                                             seattle kansas city
                                                                       cleveland
                                                                                                  baltimore
                                                                        washington
                                                           atlanta                   pittsburgh
          2




                 4000                       5000                 6000                                  7000
                                       Per capita Medicare spending, 1992
Changes in Medicare correlated with changes in Non-Medicare,
            but only because of joint convergence
 Changes in Private Insurance premiums are correlated
with changes in average nonelderly acute health spending
                                   Conclusions
•   Much of the variation in Medicare spending across states can be explained. Most of
    the explained variation is likely attributable to factors that affect health and health
    behaviors, rather than practice styles.

•   To the extent that the remaining unexplained portion reflects greater efficiency,
    there is still scope for reducing Medicare spending by increasing efficiency.
     – Whose practices should be emulated?
           • Ranking of state spending changes after adjustment.
           • Variation in Medicare growth is negatively correlated with Medicare
              spending.

•   Most of the variation in nonelderly health spending can be explained as well, and
    the two different sources of data are correlated both across states and over time.
    However, measures of health have no explanatory power.

•   Medicare is not a good proxy for non-Medicare spending. Both the levels and the
    growth rates of spending are not correlated or only correlated a little.

•   Can’t simply assume that Medicare mirrors non-Medicare spending. This also
    casts doubt on the idea that Medicare variation reflects practice styles.

•   Furthermore, it also casts doubt (but doesn’t disprove) the assumption that changes
    in incentives in Medicare will spillover to changes in private health spending.
     Some very preliminary results on quality and safety


Quality: Risk-adjusted mortality from surgical and medical procedures
  performed in hospitals

Safety: Composite measure including things like prevalence of
  bedsores, sepsis, death from low-mortality DRGS, foreign bodies left
  in during procedure, infections, complications of anesthesia.

Source: Health Grades
                        Safety of Health System
                             Index of Safety, 2002-2004



    Medicare Spending                        1.5

    Non-Medicare Spending                                   -0.3

    Per cap Income                        -.95**          -.82**   -.82**

    Poverty                              -1039*           -1008*   -1008*

    Social Capital                        3288*            2630    2561*



    Rsq                                     0.33            0.33     0.33




• Higher real Medicare spending not associated with lower safety.

• Higher poverty associated with lower safety.

• Higher social capital does seem to improve safety.
Health Care Outcomes: Mortality and Urban share
10



                    MS
 9




                         AR
                                     SC
                                                WY
     VT                                                           NM                    NY
                              AL                OK
 8




                                                         NE
                                    NH
                                     NC        TN
                          MT                      ID
               WV                                                           OR
          ME                                              KS
                                                             VA
                                                            GA                                 MA
                                         IA                  LA                                     NJ
                                                                                               RI
 7




                                                        MOIN                     WA      CT
                                                       WI              PA         TX
                               KY                                                               NVCA
                                                                            DE         MDUT
                         SD
                                                                                         IL
                                                          MN
 6




                                                                  MI
                               ND                                      OH                 FL
                                                                                   CO    AZ
 5




     .4                             .6                                  .8                               1
                                              percent urban
               Health Care Outcomes: Mortality and Poverty Rate
               20




                                                                                        MS
                                          DC
                                                                     NM
               15




                                                       LA
poverty rate




                                              KY             WV                AR
                                                TX
                                                                    AL
                                                                            SC
                                                       GA          NC OK
                                                                  TN
                        AZ                                     ID      NY
                                               CA RI          MT
                                                             OR
               10




                         OH                   NV
                                                  WA        ME
                          FLMI     IL           PAMO
                        CO              SD
                         ND              UT
                                                 IN                NE     WY
                                         DE     WI
                                                        VA
                                                         MA
                                                      IA   KS
                                                       NJ
                                                 CT
                                         MD
                                  MN                              AK VT
                                                                  NH
                5




                    5         6                7               8                    9        10
                                         risk-adjusted mortality rates
                              Explaining Health Care Outcomes

                                   Risk-adjusted Mortality



      Medicare Spending                -0.00002                        0

      Non-Medicare Spending                                                      0

      Per cap Income                                                 .0001**   .0001**

      Black                                                          12.2**    13.4**

      Urb                                                            -2.5**    -2.7**

      Black*Urb                                                       -17*      -20**

      Poverty                                                         .13*      .12*

      Social Capital                                         -0.26    -0.2      -0.14



      Rsq                                  0                 0.03     0.41       0.4




•No relationship between health spending and risk-
adjusted mortality.

•Urban areas and richer areas (?) have higher mortality.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:7
posted:6/22/2012
language:
pages:43