MEDICARE SPENDING: STUDY HIGHLIGHTS GEOGRAPHIC VARIATIONS

      There are wide geographic variations in Medicare spending
 across the nation that should be remedied before extensive cuts
 are made to the Medicare program, according to a study in the
 October issue of the American Journal of Public Health.  The
 study found that "Medicare expenditures per enrollee varied
 considerably by state in 1992, ranging from a low of $2,157 in
 Hawaii to a high of $4,430 in Maryland."  The variation was
 "substantially greater" for home health and skilled nursing
 facility expenditures than for hospital inpatient and outpatient
 care and physician services.  In some instances the disparity was
 dramatic.  For instance, "[i]n Tennessee, Medicare pays more than
 11 times more per enrollee for home health care than it does in
 South Dakota," and "[s]killed nursing facility expenditures per
 enrollee are nearly seven times higher in California than in
 Maine."  
      MORE FINDINGS 
      The study also found "considerable variation in average
 payment per discharge for inpatient hospital care, although
 Medicare's Prospective Payment System has established a national
 payment rate."  The variation was attributed to the "many
 adjustments made to the [PPS] for each diagnosis-related group,"
 including "modifications for geographic region, local hospital
 wage rate, the Medicare Case-Mix index, graduate medical
 education, patients with very long stays or especially expensive
 hospitalizations, having a high proportion of poor patients, and
 being a sole community hospital."  Higher Medicare costs per
 beneficiary were also associated with a "larger proportion of the
 elderly living in urban areas ... a larger per capita hospital
 bed supply ... a higher aged mortality rate ... and a larger
 Medicare physician assignment rate."  However, the most
 significant variable associated with lower Medicare expenditures
 per enrollee was a high HMO penetration rate.  In addition, two
 other significant findings were that a "higher proportion of
 black elderly and a larger nursing home bed supply results in
 more Medicare spending," and that a "higher proportion of urban
 elderly and a larger percentage of primary care physicians
 combine to lower Medicare spending."
      CONCLUSIONS
      The authors conclude that while "[t]here are legitimate
 reasons for differences in Medicare spending by state,"
 substantial variations remain after controlling for
 sociodemographic factors.  Reasons for this include the fact that
 "Medicare spending per enrollee by state is driven to some degree
 by total per capita health care expenditures by state."  In
 addition, the authors note that cost-shifting occurs in some
 states.  For instance, "Maryland is a very high-cost state ...
 partly because Medicare pays 108% of Maryland hospital costs." 
 In addition, the authors note that the study does not adjust for
 certain variables such as PPS adjustments for "[GME] and being
 the sole community hospital."  They conclude that "[b]ecause
 variation across states in Medicare expenditures per enrollee
 .... is driven more by volume of services per 1,000 enrollees and
 number of service units per user than by average payment per unit
 of service, Medicare's [PPS] for hospital inpatient care and
 Resource Based Relative Value Scale for physician service payment
 have limited impact on reducing variation."  The say that "[n]ew
 strategies that focus on proportions of users and amount of
 service per uses are required," and that "[b]efore sweeping cuts
 are made in Medicare, more attention needs to be focused on the
 current disproportionate distribution of expenditures across
 states" (Kane et al, 10/97 issue).