2004 report to Congress on competitive bidding, which has led to subsequent competitive bidding roll out.
Final Report to Congress:
Evaluation of Medicare’s Competitive Bidding Demonstration
For Durable Medical Equipment, Prosthetics, Orthotics, and Supplies
Tommy G. Thompson
Secretary of Health and Human Services
2004
Purpose
Section 1847 of the Social Security Act, as added by section 4319 of Public Law 105-33, the Balanced Budget Act of 1997 (BBA), directs the Secretary of Health and Human Services to submit a series of annual reports and a Final Report on the impact of competitive bidding projects authorized in the BBA. The reports are to “evaluate the impact of the demonstration projects on Medicare program payments, access, diversity of product selection, and quality.” The Secretary is hereby submitting the Final Report.
Background
Section 1847 of the Social Security Act authorized the Secretary to conduct
Demonstration Projects for Competitive Acquisition of Items and Services. In these
projects, Medicare Part B items and services (other than physician services) could be
furnished under competitively awarded contracts, with competitions conducted in
competitive acquisition areas (defined under the act as a Metropolitan Statistical Area
[MSA] or smaller area within an MSA). Under this authority the Centers for Medicare &
Medicaid Services (CMS) implemented competitive bidding for durable medical
equipment, prosthetics, orthotics, and supplies (DMEPOS) in two demonstration sites
from 1999 to 2002. The CMS’ approach was to test competitive bidding in the context of
the current regulatory environment, without otherwise making major changes.
In the first site, Polk County, Florida, (pop. 491,851) the CMS conducted the first of two
rounds of bidding in 1999. Five categories of DMEPOS were put up for bidding: oxygen
equipment and supplies (required by statute), hospital beds and accessories, enteral
nutrition formulas and equipment, urological supplies, and surgical dressings. A total of
16 winning suppliers began providing demonstration products and services in Polk
County on October 1, 1999, and continued for 2 years. The second and final round of
bidding in Polk County was conducted in 2001 for the same product categories minus
enteral nutrition. (Enteral nutrition was dropped to retain only product categories that are
overwhelmingly used in private homes.) The second set of competitively bid fees took
effect in October 2001. As in round one, 16 suppliers were selected, of whom half
participated as winners previously. The new fee schedules developed from the bids in
each round replaced the statewide Medicare DMEPOS fees. The second round of the
demonstration in Polk County ended in September 2002.
Texas was the second site of the demonstration. In the San Antonio MSA’s Bexar,
Comal, and Guadalupe counties (pop. 1,593,389) the CMS conducted bidding in 2000 for
five kinds of DMEPOS: oxygen equipment and supplies, hospital beds and accessories,
wheelchairs and accessories, general orthotics, and nebulizer drugs. Fifty-one suppliers
were selected and began serving Medicare beneficiaries under the new fees in February
2001. The San Antonio site ended operations in December 2002, the statutorily required
termination date in the BBA.
1
Evaluation Study
The CMS contracted with the University of Wisconsin-Madison in 1998 to conduct the
evaluation. The University and the Research Triangle Institute led the evaluation team.
For the First Annual Report, evaluation activities included a beneficiary survey; five site
visits by the team to Polk County, Florida, and to the Medicare DME regional carrier
managing the project in 1999 and 2000 (Palmetto Government Benefits Administrators
[PGBA]); focus groups in Polk County with suppliers and members of other affected
groups; analysis of suppliers’ bids and comparison of fee schedules; and review of
operational and documentary materials such as ombudsman records and the
demonstration Request for Bid Proposals from suppliers.
For the Second Annual Report, the team conducted a followup beneficiary survey in Polk
County, enabling assessment of numerous effects of competitive bidding. The team also
analyzed the Medicare savings under the second competitively bid fee schedule in Polk
and collected information from nine Florida suppliers in a written format. The team
traveled to San Antonio for three site visits to interview demonstration and
nondemonstration suppliers, referral agents, beneficiary representatives, and the San
Antonio demonstration ombudsman. They also analyzed Medicare savings under the
competitively bid fee schedule in San Antonio. They held discussions about the San
Antonio operations with PGBA in Columbia, South Carolina. As in Polk County, a
baseline survey was administered to a sample of Texas beneficiaries.
Additional evaluation activities followed the Second Annual Report. The team conducted
a followup survey among San Antonio beneficiaries to enable comparisons with the
baseline survey. They conducted several analyses of Medicare claims data from 1997 to
2002 in order to refine earlier estimates of Medicare savings under the demonstration,
examine access to portable oxygen, assess competition in the demonstration areas, and
test for possible volume changes due to the demonstration. A supplier survey in San
Antonio and a comparison area was fielded to study impacts on product selection and on
the financial status of suppliers. The team also conducted one additional site visit to Polk
County and one to San Antonio, during which they gathered information from informants
and stakeholders such as referral agents, suppliers, and beneficiary groups.
From these many sources, the evaluation team developed a wide array of information
useful to both policymakers and Medicare program planners. Their Final Evaluation
Report, attached, details numerous facts and analyses upon which the findings rest.
Results in Brief
The evaluation focuses on five major areas of impact:
1. Medicare expenditures;
2. beneficiary access;
3. quality and product selection;
4. market competitiveness; and
5. administrative feasibility of the reimbursement system.
2
In each area evaluation data indicate mostly favorable results for the Medicare program.
• The project saved significant expenditures, nearly 20 percent overall in each
site.
• Statistical and qualitative data indicate that beneficiary access and quality of
services were essentially unchanged. With the help of traditional intermediaries
such as hospital discharge planners and physicians, beneficiaries negotiated the
new system satisfactorily. A few areas of concern surfaced in either statistical
data or site visits—most notably, possible pressure on access to portable oxygen
in Polk County, where evidence from both the beneficiary survey and personlevel
claims analysis indicated a reduction in the proportion of oxygen patients
using portable oxygen. An area of concern arising during San Antonio site visits
was reportedly poor service quality from some wheelchair suppliers. Some of the
specific concerns involved improper fitting and instances of separate billing for
accessories formerly provided gratis as part of the overall wheelchair order.
These findings suggest areas where administrative and policy tools can be
developed to forestall deterioration of access and quality in future bidding
initiatives.
• Although it was difficult to generate information on changes in product selection,
the San Antonio supplier survey and site visit data suggest that beneficiaries
experienced little or no change in the array of products available to them.
• The market competitiveness analysis indicates that adequate numbers of
bidders participated, particularly in the larger-volume product categories. A
second competition held in Polk County after 2 years resulted in 50 percent
turnover in winning suppliers; both small and large suppliers were selected.
Market concentration usually changed little, despite the fact that 53 percent to
65 percent of bidders were chosen as suppliers.
• The CMS, with PGBA serving as project manager, successfully administered
the new payment system, from early site preparation through the bid solicitation
and evaluation period to the implementation and monitoring phases. Suppliers
reported little difficulty preparing bids, and claims processing proceeded
smoothly.
Medicare’s policy objectives in terms of savings, access, quality, competition, and
administrative feasibility were largely realized under the competitive bidding
demonstration. The program design, calling for multiple winners to maintain qualitybased
competition, appears to be a critical element. For example, during site visits,
referral agents repeatedly took credit for judging service quality and steering
beneficiaries to suppliers whom they judged to be the better performers. For the dual
purposes of maintaining quality and access, and sustaining competitive markets, the
multiple winner design appeared to serve Medicare’s needs.
3
Results by Evaluation Area
Below we summarize the key evaluation findings in each impact area, comment on the
strength and consistency of the evidence, and discuss some reasons for the findings. We
also summarize the evaluation team’s recommendations regarding the suitability of the
individual demonstration product categories for competitive bidding. The Final
Evaluation Report contains further details on methods, results, and implications for future
competitive bidding initiatives.
Medicare expenditures
Fee schedules resulting from the three bidding competitions held in the two sites suggest
that substantial savings can be realized from competitive bidding. Our final estimates
suggest savings of 16-17 percent annually in Polk County’s Round 1, 20 percent in Polk
County’s Round 2, and 20 percent in San Antonio’s single round. Overall, the
demonstration in both sites saved 19 percent over what would have been paid under the
existing statutory fees. The demonstration reduced Medicare payments by $7.5 million
and beneficiary payments by $1.9 million.
For each demonstration product or service, the prices bid by winning suppliers were
combined to determine the competitively bid Medicare fees. A bid consisted of an
offered price for each item within a demonstration product category. For each bidder, the
prices for all items in a category were combined into an average price using a standard set
of weights. These “composite prices” were arrayed from lowest to highest. To determine
a cutoff separating firms in the competitive range from remaining firms, the CMS
selected bids in order of composite price until design requirements were met. These
requirements included, for example, a certain minimum number of suppliers for each
category. The offered prices of firms in the competitive range were adjusted1 and then
averaged to determine the demonstration fee.
Polk County fees and savings
Fees resulting from the Round 2 competition in Polk County were lower than the fees on
the Year 2001 Medicare statewide fee schedule for 7 of 7 oxygen items, 17 of 17 hospital
beds and accessories items, 18 of 24 urological items, and 21 of 28 surgical dressings
items. Among the six urological items and seven surgical dressings items with fees
higher than the Medicare statewide fee schedule, one-half or more are no greater than 20
percent higher. An increase in some fees under competitive bidding may be an indication
that cost growth for certain items outpaced general cost increases allowed under
statewide fees.
Overall, the new fees were favorable to Medicare. In Round 2, the average price
reduction for oxygen was 19 percent; for hospital beds and accessories, 34 percent; for
urological supplies, 7 percent; and for surgical dressings, 4 percent. The percentage
reduction in total allowed charges for these categories was 20 percent, 31 percent, 9
percent, and -1 percent, respectively (Table 1). The adjustment factor is the ratio of the cutoff composite price to the bidder’s composite price. The
minimum adjustment factor equals 1. The adjustment ensures that total revenues under the demonstration
fees equal the revenues that would result from using the cutoff composite price, assuming a standard set of
volume weights.
4
Compared to the fees in Round 1, fees in Round 2 for oxygen and hospital beds exhibited
little change, falling generally within about 5 percent of round one fees. Savings for
hospital beds and accessories grew from about 18 percent to approximately 25 percent
during Round 1, even though fees did not change for the 2-year duration. The increase
was probably due to the expiration of grandfathering clauses for capped rentals that
predated the demonstration. New rentals starting after these expired were all subject to
the demonstration fees.
Table 1 Medicare DMEPOS Competitive Bidding Demonstration:
Percentage savings in total allowed charges, by bidding round and year
Polk County, Florida San Antonio, Texas
Round 1, Year 1 Round 1, Year 2 Round 2 Year 1 (11 mos.) Year 2
Oxygen equipment
and supplies 16.6 16.7 19.8 20.7 18.0
Hospital beds and
Accessories 18.5 27.5 30.7 13.8 24.5
Urological supplies 17.8 17.2 9.4 n/a n/a
Surgical dressings -12.0 -12.7 -1.2 n/a n/a
Enteral nutrition 16.3 17.0 n/a n/a n/a
Wheelchairs and
Accessories n/a n/a n/a 14.9 23.0
General orthotics n/a n/a n/a 25.4 21.5
Nebulizer drugs n/a n/a n/a 26.4 26.0
ALL CATEGORIES 16.3 17.2 20.3 19.6 21.3
Note: Savings are based on actual utilization derived from Medicare claims data.
Fees in Round 2 were almost always lower for surgical dressings, and all fees were
higher for urological supplies. Surgical dressings fees in Round 1 were high due to
unintended consequences of the technical procedure for summarizing an individual firm’s
bid prices. An improved procedure used in Round 2 probably helped to lower surgical
dressings fees so that most are now below the statutory fee schedule. Round 2 fee
increases for urological items were generally between 10 and 20 percent. Once the
demonstration got underway, some urological suppliers discovered they had bid too low
to cover costs. The shift in fees for these two categories is reflected in the changes in
savings between Round 1 and Round 2.
5
San Antonio fees and savings
Demonstration fees resulting from the San Antonio competition conducted in 2000 were
lower than the Medicare statewide fee schedule for 10 of 10 oxygen items, 18 of 18
hospital beds and accessories items, 61 of 61 wheelchair and accessories items, 46 of 46
orthotics items, and 16 of 27 nebulizer drugs put up for bidding. The average price
reduction for oxygen was 22 percent; for hospital beds, 26 percent; for wheelchairs, 20
percent; for general orthotics, 10 percent; and for nebulizer drugs, 21 percent . Annual
savings in the second year of San Antonio’s single round were 18, 24, 23, 21, and 26
percent, respectively (Table 1). As in Polk County, savings grew from Year 1 to Year 2
for capped rental equipment (hospital beds and wheelchairs), apparently due to the
expiration of grandfathered arrangements.
Volume impacts
If price changes affect utilization—for example, by making equipment more affordable for
beneficiaries—then savings estimates should take the impacts into account. So it is
important to consider whether large volume increases significantly reduced the potential
savings from the fee reductions. The evaluation team provided alternative savings
figures predicated on estimates of volume impacts. The alternative estimates are based
on statistical models of growth in claims volume in the demonstration areas in
comparison to similar areas in the states of Florida and Texas. Each product category
was analyzed separately, and most analyses revealed no statistically significant impact on
volume.
Two notable exceptions were the analyses of oxygen concentrators and portable gaseous
oxygen systems in Polk County. The models for these categories indicated statistically
significant volume increases which would reduce the savings estimates for oxygen
equipment overall in Polk by about half.2 This result rested largely on an unexplained
drop in volume in several of the Florida comparison counties during Round 2.
Specifically, under the demonstration, raw data suggested that Polk’s oxygen growth
continued at historical rates while growth in several of the comparison counties turned
negative. The team concluded that the statistical evidence of the demonstration causing
volume growth in these two important oxygen categories in Polk is questionable.
Further, information in support of such an impact did not emerge from analysis of
volumes in San Antonio and site visits there. Accordingly, the best estimate of the
demonstration savings in Table 1 is not adjusted using results of the volume analysis.
Another category of note is wheelchair accessories, which were very low volume items
sometimes provided gratis by wheelchair suppliers in San Antonio. For five of the six
accessories tested in the analysis, estimated volume impacts were an increase of at least
100 percent, averaging about 150 percent. However, only one of these tests, for rental of
an anti-tipping device, was statistically significant. Yet there were some indications from
informants during site visits that suppliers were less likely to forgive the price of
2 Final Evaluation Report, Appendix B.
6
Wheelchair accessories under the demonstration. It is also possible that the price
reductions for the accessories, which averaged slightly more than 20 percent, led to
higher beneficiary demand. Because accessories are an exceedingly small factor in the
expenditures for the wheelchair category, and because the formal statistical findings were
very weak, the best estimate in Table 1 is not adjusted for volume impacts.
In summary, some isolated results were possibly suggestive of volume responses to the
fee changes brought by competitive bidding. These results were far from conclusive, yet
they do not completely negate the notion that a volume impact is possible in future
initiatives.
Access to DMEPOS goods and services
The evaluation’s three main sources of information on access to care were the beneficiary
surveys, site visits, and San Antonio supplier survey. Data from all three sources
suggested little or no impact of the demonstration on access to goods and services put up
for bidding. Reduced use of portable oxygen among Polk beneficiaries is the single
important result pointing to a risk of lower access under competitive bidding. We discuss
this further below.
In both sites, transition policies protected beneficiaries from disruptions in existing
relationships with oxygen companies, nebulizer drug suppliers, and suppliers of capped
rental equipment (nutrition infusion pumps, hospital beds, and wheelchairs). For oxygen
and nebulizer drugs, existing relationships could continue as long as the supplier agreed
to accept demonstration fees, and suppliers overwhelmingly agreed to this arrangement.
Capped rental fees for agreements predating the demonstration stayed in place.
The beneficiary surveys in Florida and San Antonio employed similar designs. Separate
surveys were administered to random samples of oxygen users and other equipment
users. The team analyzed total users and new users separately. New users are more likely
to be directly affected by the demonstration, because they had no previous relationship
with a supplier in their product category. New users accounted for 10 to 40 percent of the
samples (across sites and time periods). Questionnaires covered a wide range of
measures dealing with access to equipment, training, maintenance, customer service, and
delivery services. Responses collected before the demonstration were compared with
responses after the demonstration. Impact estimates took into account general trends that
might affect responses (by surveying a similar area). Estimates also controlled for
sample differences in demographic composition and health status. The Florida surveys
were conducted in the second quarter of 1999 and the first quarter of 2001; the San
Antonio surveys were conducted in the 4 months ending February 2001 and the 5 months
ending July 2002.
Access to oxygen services in Polk County and San Antonio
Beneficiary survey measures overall indicated that access to oxygen services remained
unchanged under the demonstration. At both baseline and followup, measures suggested
beneficiaries in general have good access to oxygen services.
7
Delivery and supplier accessibility indicators. Numerous delivery and accessibility
measures remained stable. About three-quarters of beneficiaries initially received their
equipment on the same day it was ordered. The frequency of portable oxygen refills
remained unchanged or, in the case of new users in San Antonio, increased. Also stable
were the rate of home delivery by the supplier, distance to supplier, number of portable
refills ordered each time, and the rate beneficiaries ran out of supplies. Access to
customer service also did not shift markedly, including the number of phone or in-person
contacts in the past 6 months, service call response time, and ability to contact the
supplier by telephone. Home health agencies became more involved in ordering and
delivery in Polk, a change with no clear explanation, although it might reflect greater
initiative taken by paid caregivers to help beneficiaries adjust to the demonstration
environment.
Access to training. Estimates of changed performance in particulars of training upon
initial delivery (e.g., receipt of written instructions, trained in replacing parts) were mixed
and usually statistically imprecise. Impacts tended to be quite small. Training indicators
were more consistently unfavorable in San Antonio than in Polk County, and a couple
were marginally nonsignificant.
Access to maintenance, important to assure continued purity of the oxygen flow, did not
change. For example, about 70 percent in Polk and 50 percent in San Antonio reported a
maintenance visit from the supplier in the past month.3 Majorities reported receiving
maintenance visits routinely every 1 to 3 months. Access to these and other services
appeared stable. Respiratory therapist visits remained unchanged, as did the time since
the last physician respiratory checkup, a measure included as an indirect effect of changes
in supplier access.
Portable oxygen
Although most survey access indicators indicated stability, an important exception was
portable oxygen use in Polk County. Portable oxygen is necessary for certain patients to
move freely about their home and to travel outside the home. It can facilitate adherence
to the doctor’s prescribed oxygen regimen. Access to portable oxygen can be important
for promoting quality-of-life and for assuring the effectiveness of oxygen therapy.
The Polk County survey results suggested a marginally nonsignificant decline of about 6
percentage points in the overall rate of portable use (p=.057) and a significant reduction
of 24 percentage points among new users (p =.025). The survey impact estimates pertain
to Round 1 of the demonstration. This result was further investigated with claims
analysis, which suggested a small, statistically significant reduction of 3 percentage
points among new users in Round 1 (p < .01) and a larger decline of 12 percentage points
among new users in Round 2 (p < .01). Unadjusted claims trend data showed a general
decline in portable use, beginning early in Round 1 and affecting Polk and the five
counties used for comparison in Florida claims, except that the Polk downtrend among
new users was steeper. New users also reported a 44-percentage-point increase in
3 See Final Evaluation Report, Appendix C.
8
Oxygen-conserving devices on portable systems. These devices can reduce the frequency
of portable refills, although they are not well suited to every patient.4
San Antonio’s results contrasted with Polk County’s. In the survey, portable oxygen use
appeared stable. There was trivial change in oxygen-conserving devices. A relatively
large increase in use of portable liquid cylinders, from about 12 percent to 18 percent,
may be a positive access indicator, because these are often lightweight systems; however,
statistical confidence in this finding was limited (p=.098). Claims analysis, using the
same comparison area used in the survey, revealed very small reductions in the portable
use rate. Claims trend data indicated no general decline.
The two statistical sources of evidence on portable oxygen in Polk County were
consistent in pointing to a possible access problem, but the impact sizes differed. Size
inconsistency may be due to differences in comparison areas between the survey and
claims analyses, sample-related variation in the survey, and factors affecting survey
responses that remain unclear. Both sources used large numbers of observations,
especially the claims data. Notwithstanding the size difference, the survey and claims
evidence seem sufficiently persuasive to conclude that lower access to portable oxygen is
a real risk posed by competitive bidding, at least in some areas and in the general
program environment in effect during the demonstration.5 Suppliers have an incentive to
reduce access, because the add-on payment for portable oxygen is fixed per month,
regardless of the number of refills needed. Previous research on the impact of price
reductions did not find lower access to portable oxygen following 25 percent national fee
reductions in 1998 mandated by the 1997 Balanced Budget Act (CMRI, 2000). But fee
reductions under the demonstration came in addition to the pre-existing reductions. Fees
may have dropped low enough to induce a market response.
How could the demonstration have brought about changes in access to portable oxygen?
Two previous studies (CMRI, 2000; GAO, 1997) mentioned indications of high
variability among suppliers in the proportion of patients offered portable or lightweight
portable systems, with some suppliers providing portable oxygen to only a minority of
patients. However, we found little indication that such suppliers may have been more
likely to win the bidding. An oxygen policy change coinciding with the initiation of the
Polk demonstration raised another possibility, because this change tightened Medicare
eligibility for portable oxygen. (Florida trend data cited above illustrate the impact.)
Perhaps the Florida winning suppliers adhered with unusual diligence to the new
coverage requirement in the spotlight of the payment system experiment. We did not
have systematic data necessary to examine this possible explanation, and no informants
mentioned this type of behavioral change during the team’s site visits. The coverage
change could have combined with the demonstration’s cost-reduction incentives, with
both resulting in less portable oxygen use. For reasons cited above, even in the absence
of the coverage change, Polk suppliers might have been highly motivated under
competitive bidding to save costs by strictly following the coverage and documentation
rules. Another portable oxygen indicator in Polk County was consistent with supplier
4 United States General Accounting Office, Medicare: Access to Home Oxygen Largely Unchanged; Closer
HCFA Monitoring Needed, GAO/HEHS-99-56, April 1999
5 Except for lowered fees and fewer suppliers, basic program requirements were otherwise unchanged.
9
Cost-consciousness—the large increase in oxygen-conserving devices on portable
systems.
It is not certain that reduced portable use rates harmed any beneficiaries. Other sources
of data did not suggest serious problems with portable oxygen access. Site visit sources
did not cite access to portable oxygen as a problem in either Florida or Texas. In general,
satisfaction measures in the survey remained high and unchanged under the
demonstration. The survey did not find large numbers of beneficiaries switching their
patronage to new suppliers. Moreover, it is not clear that all beneficiaries need or use
portable oxygen when provided. Previous research has found that a significant fraction
do not use their units.6 For these reasons, we cannot rule out the possibility that the
demonstration reduced some inappropriate provision of portable oxygen. In future
initiatives, policy and administrative tools, such as stakeholder education, can be
enhanced towards providing oxygen modalities that fit the medical necessity in each case.
Access to medical equipment in Polk County and San Antonio
Beneficiary survey estimates for medical equipment users in the demonstration suggested
larger access impacts compared to oxygen, but these were in no consistent direction.
Most findings were statistically nonsignificant, due in part to smaller sample sizes for the
medical equipment surveys. The overall impression emerging from the medical
equipment surveys was, again, few signs of problems, stability in access, and generally
strong access levels.
Delivery indicators. Findings suggested some small changes took place for medical
equipment users but these probably did not represent important shifts in access. In Polk
County the demonstration may have caused less reliance on supplier home delivery in
favor of mail and home health agency delivery, although none of these changes was
statistically significant, and supplier delivery remained dominant by far. For new users, a
decline in home delivery was more apparent (-24 percentage points, p=.017). By
contrast, in San Antonio, delivery overall appeared to shift towards home delivery, as
well as to mail delivery, and marginally away from home health agency delivery. The
shift to supplier delivery was due to new equipment users generally (+19 percentage
points, p=.034), and nebulizer drug users (+21 percentage points, p=.047). These results
may mean that suppliers sought to reduce costs by cutting back on home delivery. Also,
the survey substantiated a small lengthening of initial delivery time in San Antonio,
which increased on average from 2.3 to 2.9 days overall (p=.080) and from 1.6 to 2.7
days (p=.042) for hospital beds. In Polk, delivery time remained unchanged at two days,
on average.
Supplier accessibility. Measures here were stable or appeared improved. Supplier
service call response time changed trivially, except for surgical dressings users in Polk,
where it fell from about three days to about one day (p=.033). Medical equipment users
in San Antonio gained in their ability to contact the supplier by telephone for all users
(+10 percentage points, p=.026) and for new users (+20 percentage points, p=.049) .
6 Department of Health and Human Services, Office of Inspector General, Usage and Documentation of
Home Oxygen Therapy, OEI-03-96-00090, August 1999.
10
In both sites, it is likely that beneficiaries became less directly involved in arranging for
their own equipment. In Polk County, the rate of self-ordering fell by 14 percentage
points (p=.043) overall. This reduction was particularly associated with the subgroup of
hospital bed users (-26 percentage points, p=.001). Similarly, in San Antonio, selfordering
declined for all medical equipment users (-13 percentage points, p=.009),
wheelchair users (-12 percentage points, p=.040), and nebulizer drug users (-31
percentage points, p=.006). Survey results in both sites suggested a “caregiver” became
more involved in ordering, but with the exception of new medical equipment users in San
Antonio (+21 percentage points, p=.033), statistical precision on this impact was low.
Access to training. Specific indicators of training at initial delivery were generally
strongly improved in both Polk and San Antonio, but often had low statistical reliability.
Two exceptions were for Polk subgroups; these were large, negative and statistically
significant—a 22 percentage point rise (p=.047) in urological supplies users reporting
receiving no training, and a 40 percentage point decrease (p=.032) in surgical dressings
users reporting instruction in maintenance. Both results may be tied to more mail delivery
in these categories, a trend observed in unadjusted data that may be partly traceable to the
out-of-town location of several winning suppliers in these categories. Relatedly, surgical
dressings users reported a halving, on average, of the number of contacts with suppliers
“in the past six months” (from 4.4 to 2.5, p=.046). It is not known whether local
diffusion of new surgical dressing technology contributed to findings of fewer direct
contacts. In the First Annual Report to Congress, the team noted that a winning supplier
new to the area may have helped spread use of better dressings. To the extent surgical
dressings patients receive wound treatment from medical personnel, a reduction in
contacts and maintenance training from the supplier may not be clinically important.
Access to maintenance. In both sites indicators suggested modest, statistically
nonsignificant reductions in the likelihood of routine maintenance visits. However, in
one Polk subgroup, new users, the probability of a maintenance visit in the past 30 days
declined 34 percentage points (p=.042). Unadjusted data suggest the groups most
affected were new surgical dressings and new hospital bed users.
Other access-related information: supplier coverage area, site visits, supplier survey
Supplier coverage area. Winning suppliers helped maintain good access by offering
extensive service areas. In Polk’s Round 1, three-quarters of the winning suppliers agreed
to serve the entire demonstration area. CMS required all winners in Round 2 to serve the
entire demonstration area, potentially easing access further. In San Antonio, 80 percent of
winning suppliers agreed to serve the entire three-county demonstration site. These
coverage rates were achieved with many small suppliers represented among the winners.
11
Site visits. The evaluation team conducted interviews and focus groups among
beneficiary groups, suppliers, and referral agents such as hospital discharge planners.
The site visits produced a variety of interesting facts and perspectives. Referral agents,
beneficiary groups, nondemonstration suppliers, and others early on voiced fears about
threats to access from having fewer suppliers to patronize, from selecting out-of-county
suppliers, and from the possible loss of suppliers who might add in certain items gratis for indigent patients. In addition, they worried about the loss of “one-stop shopping,”
whereby a beneficiary might choose one supplier for all of their needs based on the price
for needed services not covered by Medicare. Charges for noncovered services can vary,
whereas covered services’ prices are uniform under the statewide Medicare fee schedule.
If a winning supplier charges more than a nondemonstration supplier for a noncovered
item, a beneficiary would either have to incur higher charges than necessary or split his or
her patronage among multiple suppliers.
Data-gathering during multiple site visits indicated that these fears generally were not
realized, in part because referral agents worked harder than before in assisting
beneficiaries to negotiate the new marketplace. In some cases they said they helped
beneficiaries find multiple suppliers to save money. During the early transition period,
when referral agents were pressed for time to familiarize themselves with new firms and
make judgments about their performance, some felt frustration and disappointment about
unexpected waiting times and low responsiveness, until they identified a winning supplier
who met their standards. By the end of the demonstration, site visit evidence suggested
referral agents and other stakeholders tended to see access as essentially unaffected by
the demonstration.
The beneficiary surveys included questions bearing on several of these concerns.
Responses did not indicate any systematic, large-scale changes that would indicate
serious problems. Average initial delivery times remained unchanged, except for hospital
bed users in San Antonio (see above). All four surveys suggested a possible trend to
using more than one supplier but had low statistical reliability for this measure; the
estimated prevalence of using multiple suppliers remained under 5 percent for oxygen
users and close to 30 percent for medical equipment users. A question about the “time
and energy it takes to get the medical equipment you need” produced a slight reduction in
the probability of the most favorable response category in three of the four surveys, but
no estimate of change was statistically significant. Three of the four surveys had answers
suggesting that beneficiaries were more likely to have changed suppliers in the last 6
months but, again, these estimates were not statistically significant, and changing was
unusual (5 to 10 percent switched under the demonstration). Beneficiaries sometimes
noted reasons for changing that had nothing to do with supplier performance.
Supplier survey. The San Antonio supplier survey was administered once, in 2002.
Asked about the timeliness of Medicare equipment delivery back in February 2001, 66
percent of nondemonstration suppliers in San Antonio said they believed Medicare
equipment was delivered on time. When asked the same question about a period during
the demonstration, only 26 percent of these suppliers thought equipment was delivered on
time. No such dramatic change in responses occurred among either demonstration
suppliers and suppliers in the comparison area; large, stable majorities assessed delivery
as timely for both periods. Even though sample sizes were small for purposes of
statistical confidence, the results suggested that supplier perceptions are influenced by
market position. Taking that into consideration, these results seem broadly consistent
with the beneficiary survey results pointing to little impact on timeliness.
12
Discussion
The survey data suggested little or no change in access overall. Where statistical
evidence suggested real change, results were mixed, with probable mild repercussions.
An apparent real reduction in the use rate of portable oxygen occurred in Polk County,
but there was no evidence that access to portable oxygen was perceived as a problem.
Polk surgical dressings users experienced faster response to service calls but less routine
maintenance contacts. San Antonio medical equipment users achieved better phone
accessibility to their suppliers. There was a tendency for more mail and home health
agency delivery in Polk, and a small lengthening of initial delivery time for San Antonio
hospital bed users. There was a weak suggestion in the data that routine maintenance
visits declined somewhat for medical equipment users, but not oxygen users, where
regular maintenance is clearly critical. Beneficiaries in both sites became less frequently
involved in ordering their equipment.
Hypothetically, a change in delivery source might be associated with a decline in certain
other indicators, particularly training received from the supplier upon delivery. However,
for the variety of training indicators on the survey, generally this was not the case. Part
of the explanation for a shift away from supplier delivery may be that mail delivery for
supplies such as surgical dressings and urological items found increasing use as a costsaving
measure. The shift in ordering party may indicate that beneficiaries sought help or
were offered help more often than before in adapting to the DMEPOS market, which
changed under the demonstration. Medicare’s pre-demonstration publicity efforts could
have contributed by preparing referral agents, particularly home health agencies, and
motivating them to assist both oxygen users in Polk County.
Considering the stakeholder perspectives gathered during site visits, it seems likely that
competitive bidding complicated the activities of referral agents, such as discharge
planners and physician offices, on behalf of beneficiaries. Competitive bidding may carry
some unavoidable costs, such as a heavier workload for referral agents, more comparison
shopping for beneficiaries, higher out of pocket payments on noncovered items, more
paperwork if multiple suppliers are used, and some discomfort and occasional disruption
in adapting to a modified list of approved suppliers after each bidding round.
As noted in previous Reports to Congress, the mostly favorable access findings appeared
related to several demonstration features. First, the design provided for multiple winners
in each product category. Second, winner selection procedures explicitly considered
bidders’ capacity and service capabilities. Third, transition policies allowed beneficiarysupplier
relationships to continue for appropriately selected categories of equipment.
Quality and product selection
The evaluation’s three main sources of information on quality and product selection were
the beneficiary surveys, site visits, and the San Antonio supplier survey. Global
measures of beneficiaries’ satisfaction with their supplier remained high under the
demonstration, and detailed quality measures were similarly favorable and stable.
13
Based on the supplier survey, products provided to beneficiaries changed little during the
demonstration. The site visits revealed issues surrounding urological supplies and
wheelchair fitting and delivery, which are discussed further below.
Quality of oxygen services in Polk County and San Antonio
As with the measurement of access, numerous survey measures captured dimensions of
quality, such as overall satisfaction, equipment reliability, and quality of training and
service. In the oxygen surveys only four indicators exhibited statistically significant
change. These four were balanced between favorable and unfavorable outcomes, and
they did not come from a common dimension. Such a pattern suggests that quality
remained essentially the same under the demonstration.
Consumer satisfaction. Both before and during the demonstration, a healthy majority of
oxygen users in both sites (60-70 percent) gave their supplier the highest satisfaction
rating on a scale of 0 to 10. More than 95 percent in both sites were willing to
recommend their oxygen supplier to a friend, another measure that remained unchanged.
Quality of equipment. More than 90 percent of oxygen users in both sites rated their
equipment “very reliable.” About four in five respondents said they experienced no
major problems with their equipment in the past 6 months. Similarly, about 20 percent in
both sites reported equipment replaced due to malfunction in the past 6 months. None of
these measures changed more than trivially in the overall sample, suggesting continued
acceptable equipment quality. In Polk County’s subsample of new users, the number of
major equipment problems in the past 6 months probably declined substantially; the
average number cited by respondents fell 75 percent (p=.048).
Quality of training. The survey measured training quality via a direct question asking
for an evaluation of training, as well as other questions about the beneficiary’s “comfort
level” when using and maintaining their equipment. In Polk County, nearly 60 percent
rated training quality as “excellent,” while in San Antonio, nearly 50 percent assigned the
same rating, with little or no change due to the demonstration in either site. The
proportions saying they were “very comfortable” (the highest rating) controlling oxygen
flow, using a humidifier, attaching regulators, and cleaning the system’s filter ranged
between 70 and 85 percent in Polk County, and between 35 and 80 percent in San
Antonio. These measures hardly changed in either Polk County or San Antonio due to
the demonstration. Patients using portable oxygen with an oxygen conserving device
were asked about comfort level using that device. In both sites, regardless of the
demonstration, approximately 70 percent reported being “very comfortable.”
Quality of customer service. Asked about the tone of the beneficiary’s contacts with the
supplier in the past 6 months, upwards of 80 percent reported always being treated with
“courtesy and respect” in both sites, regardless of the demonstration. In Polk County, 75
to 80 percent said the supplier “always” explained things understandably and “always”
gave all the information or help needed during the past 6 months. The prevalence for this
rating was lower, about 55 percent, in San Antonio, but in both sites little or no change
was attributable to the demonstration. One subgroup, new oxygen users in Polk, reported
a large improvement in having things explained understandably (+44 percentage points,
p=.002).
14
The surveys asked about problem resolution, namely, whether the beneficiary contacted
the supplier with a problem or complaint in the past 6 months, and whether this was
satisfactorily resolved. This complaint rate was stable at about one-quarter in Polk, while
there was indication it might have risen somewhat in San Antonio (from 19 to 25 percent,
p=.088). In both sites approximately 95 percent reported satisfactory resolution, which
was unaffected by the demonstration.
The surveys also asked about any after-hours calls to the supplier in the past 6 months,
and whether these contacts produced the help the caller sought. After-hours calling was
reported by less than 20 percent in both sites. Of these respondents, at least 7 in 10 said
the supplier was always thorough, with somewhat fewer making this assessment under
the demonstration (-6 percentage points in Polk County and –9 percentage points in San
Antonio). Neither change was statistically significant.
Five questions about receiving help from the supplier with insurance when getting started
on oxygen service suggested mixed favorable and unfavorable outcomes under
competitive bidding, but with few exceptions (e.g., in San Antonio, a nearly doubling of
reports of “no help” [+6 percentage points, p=.016], and a halving of reports of being told
how to get insurance information [-8 percentage points, p=.019]), these changes in
customer service indicators were not large or statistically significant.
Quality of medical equipment services in Polk County and San Antonio
In the medical equipment surveys a handful of indicators changed significantly, with
most of these suggesting a change for the better. Most such changes involved subgroups
such as new users and hospital bed users. For medical equipment users in general,
quality appears not to have changed as a result of the demonstration.
Consumer satisfaction. About 40 to 50 percent of demonstration medical equipment
users in Polk County and San Antonio gave their medical equipment supplier the highest
rating on a scale of 0 to 10, a level not significantly different from ratings in the absence
of the demonstration. At least 9 in 10 said they would be willing to recommend the
supplier to a friend. However, San Antonio’s nebulizer drug users registered a 6 percent
drop in willingness to recommend (p=.005) from a baseline estimate of near-unanimous
willingness.
Quality of equipment. Overall, three-quarters of medical equipment users in both sites
rated their equipment “very reliable,” with one subgroup, surgical dressings users in Polk
County, experiencing a significant increase in the reliability rating (+41 percentage
points, p=.035). It was unusual in either site for respondents to indicate that they had to
replace malfunctioning equipment, and impact estimates were statistically insignificant.
The average number of major equipment problems in the past 6 months fell slightly, from
.41 to .33, in Polk County, and increased from .33 to .52 in San Antonio, but neither
change measure was statistically reliable.
15
Quality of training. Initial training was rated “excellent” by 30 percent of medical
equipment users in both sites. As for “comfort level” in using equipment and in taking
care of it, about 70 percent of respondents in Polk County and in San Antonio said they
were “very comfortable” in performing these activities, with ratings essentially
unchanged under the demonstration.
Quality of customer service. In Polk County about three-quarters of respondents
reported that in their recent contacts with suppliers they were “always” treated with
courtesy and respect, while this rating was assigned by about half of San Antonio medical
equipment users. In both cases, change attributable to the demonstration was trivial,
except among the San Antonio subgroup of new nebulizer drug users, where the “always”
response tripled to an estimated 85 percent (p=.009). About half or more in Polk County
reported that the supplier “always” explained things understandably and “always” gave
all the help needed during contacts in the past 6 months compared to about 30 percent in
San Antonio reporting this frequency of good performance, but, again, little or no change
was observed due to the demonstration.
The complaint rate “in the past six months” was about one-quarter in both Polk County
and San Antonio, but in Texas this level represented a significant demonstration-related
change—up nearly +9 percentage points (p=.033). One subgroup, San Antonio hospital
bed users, experienced a doubling in the complaint rate under the demonstration, up from
about 14 percent (p=.032). In Polk, impact estimates suggested a sizable increase in
numbers reporting satisfactory resolution of complaints, but this was not statistically
reliable (p=.133). In San Antonio, satisfactory resolution declined by 11 percentage
points from 90 percent, but this change estimate was not statistically reliable, either.
After-hours calls to the supplier were unusual in both sites. Sizable positive change in
reports of supplier thoroughness during such after-hours contacts was measured in Polk,
while negative change was observed in San Antonio, but neither impact estimate was
statistically significant, given the small base of respondents. New medical equipment
users in San Antonio, however, registered a large and statistically significant
improvement in reports of after-hours thoroughness.
Indicators of receipt of insurance help upon initial delivery were generally notably
improved in both Polk and San Antonio but had low statistical reliability. An exception
was hospital bed users; in Polk, they experienced a 10-percentage-point increase in the
rate at which they received insurance information (p=.036) and in San Antonio, a 36-
point increase in the rate at which the supplier offered to bill insurance (p=.035).
Nebulizer drug users in both the oxygen survey and medical equipment surveys in San
Antonio were asked whether they experienced a delay in receiving their drugs because
the supplier was out of stock. According to both surveys, there was a small
improvement, but it was not statistically significant. In contrast, a question about
medication error, which occurred rarely, suggested somewhat worsening performance,
but the estimate also was not statistically significant.
16
Other quality-related information: site visits, supplier survey
During site visits to Polk County early in the demonstration, complaints surfaced about
the quality of urological supplies amid admissions by some suppliers that they had bid
too low. If prices did not cover costs, suppliers had an incentive to offer inferior products
or pursue other strategies that could limit product selection. When the team isolated
urological users’ survey results, they found no reliable indication of more quality
problems. Beneficiaries may have responded to quality deterioration by switching to
another supplier (perhaps at their own expense) or by obtaining a prescription for a
specific product. Prices for urological supplies rose in Round 2, and a well-respected
longtime supplier who carried a broad product line was among the winning bidders. A
site visit source believed the situation improved in Round 2.
Findings from site visits in San Antonio suggested some wheelchair suppliers attempted
to cut costs by providing fewer accessories and/or charging for accessories previously
provided gratis, and perhaps by using less-qualified staff for fitting. Referral agents said
they needed to be more detailed than before in specifying orders, and more vigilant in
assuring specifications were met. Anecdotal evidence also revealed widely differing
supplier approaches to obtaining a proper fitting. Although the survey revealed no
change in quality measures for wheelchairs, this category in particular may need special
monitoring and more explicit supplier standards in the future.
During site visits to Polk County, views on quality differed between demonstration
suppliers and nondemonstration suppliers. Nondemonstration suppliers tended to be
doubtful that quality could be maintained. In San Antonio, some demonstration suppliers
expressed concern that their competitors would lower quality under price pressure,
whereas others believed that in the bidding process Medicare had dropped lower-quality
suppliers from the area.
Yet most referral agents once the demonstration was well under way did not believe
systematic quality problems resulted from the demonstration. The role played by referral
agents in screening suppliers on behalf of beneficiaries probably helped maintain quality
during the demonstration. Referral agents described how they assessed their experience
with each supplier to identify suppliers to recommend to beneficiaries routinely. In some
cases they also adapted their documentation process to enable them to work more
smoothly with certain suppliers. These reports bespeak an important position for referral
agents in the competitive marketplace, especially considering that fewer approved
suppliers result from competitive bidding.
The San Antonio supplier survey was the evaluation’s main statistical source of
information on possible changes in product selection. During early site visits to suppliers
in Polk County, it was difficult to obtain from them detailed information needed to assess
product selection. The survey strategy was to ask about the most typical brand offered to
Medicare beneficiaries in a month before the demonstration began and in a corresponding
month during the demonstration. Separately, the survey also asked the respondent to
provide reasons for changing brands. Results have limited generalizability because
relatively few suppliers responded, and a comparison-group analysis was not possible.
The results suggested that most suppliers did not change the brands they offered, with the
possible exception of hospital bed suppliers. Five of 18 respondents began offering used
mattresses instead of new mattresses. The array of brands offered before and after the
demonstration was fairly broad.7 Respondents cited various reasons aside from “lower
cost” for changing product selection (regardless of whether their data from a separate
question indicated that their most common brand/model offering changed).
17
Discussion
The evaluation data on issues of quality and product selection suggested quality did not
change appreciably. The beneficiary surveys provided the most reliable and systematic
evidence on quality, and they reveal few quality issues. Key survey indicators of
equipment reliability, complaints, and consumer satisfaction were stable or, in a few
instances (surgical dressings users, new medical equipment users) they turned more
favorable under the demonstration.
Anecdotal reports of possible quality problems with urological items in Polk County and
with wheelchair service in San Antonio came to the team’s attention during site visits.
Evidence from the beneficiary survey did not suggest that problems were widespread or
systematic. However, these cases provided lessons and guides for possible future action.
The situation with urological supplies was eventually self-correcting, in part because
Round 2 prices rose8 and new firms became demonstration suppliers. This experience
illustrated that upward price adjustment may take place under competitive bidding, and
that some product categories, especially smaller ones such as urological supplies, may be
vulnerable to a lessening of product selection under reduced prices. Some urological
supplies are associated with strong patient preferences, and therefore may be particularly
vulnerable to supplier market strategies that narrow patient choice. Educational
intervention among beneficiaries on their options, such as obtaining a specific
prescription, is one approach that may help to limit the risk of disruption.
Wheelchair fitting and service appeared subject to widely differing supplier approaches,
based on several referral agents interviewed during site visits. Along with reported
instances of poor follow-through and other questionable practices, the wheelchair
experience raises the question whether administrative attention to supplier standards on
fitting and adjustment is needed. There were also anecdotal indications that certain
quality problems became less severe when ordering documentation became more
detailed. This could mean that documentation practices can be improved. Clearer, more
detailed written specifications may impose more of a paperwork burden on referral
agents, but there should be benefits in terms of accurately filling the order the first time
and promoting accountability among suppliers.
Referral agents indicated that they learned to refer selectively as they gained experience
with demonstration suppliers of varying quality and responsiveness. It is quite possible
that this mode of operating existed before the demonstration but it likely was not
compressed into a short time frame such as that forced by the demonstration transition.
7 Section 4.7 in the Final Evaluation Report mentions brand names reported in the Supplier Survey.
8 Previous literature on competitive bidding mentions the possibility that inexperienced bidders may bid too
aggressively low at the outset. This may have happened in Polk County with urological supplies.
18
Nonetheless, the role of referral agents as market intermediaries is an interesting and
important one. Referral agents may present an arena worthy of more administrative
attention as Medicare seeks greater efficiency under the DMEPOS benefit.
Market competitiveness
Evidence for evaluating impacts on market competitiveness comes from four sources:
bidder participation and selection data, particularly from the two rounds of bidding
conducted in Polk County, claims analysis of changes in Medicare market shares, site
visit informants, and the San Antonio supplier survey.
For the Final Evaluation Report, the evaluation team undertook extensive claims analysis
to track market share changes between demonstration and nondemonstration suppliers, to
measure market concentration, and to trace individual market shares of participating
firms. The analyses suggested that during the 3-year period of the project, the DMEPOS
markets for the demonstration product lines tended to stay close to their former
concentration levels, even while demonstration suppliers as a group gained market share.
Several other pieces of evidence pointed to good signs for competitiveness. For example,
most product categories attracted numerous bidders, firms that submitted bids had a good
chance of being selected, and although suppliers have a strong tendency to dislike
competitive bidding, there was still some opinion among them that the demonstrationrelated
markets remained competitive.
Polk County bidding results
In Round 1, the 30 bidders included firms with both small and large market shares. Both
rounds of bidding produced relatively little change in the mix of small- and large-share
firms serving the demonstration area. In the second round of bidding in Polk County,
bidders numbered 26—not many fewer than in the first round, despite the reduction in
product categories from 5 to 4. There were 22 bidders for oxygen, 19 for hospital beds, 7
for urological supplies, and 4 for surgical dressings. Entry into and exit from the market
were demonstrated in the second round: half of the round two demonstration suppliers
had demonstration status in round one, but half did not. The new winners did not
represent a disproportionate number of nonlocal suppliers, relative to the group of
winners from the first round of bidding. Two of the new winners had lost the
competition in the first round, a possible indication that they learned how to be successful
from their earlier experience. Two product categories experienced declines in bidders—
urological supplies and surgical dressings.
San Antonio bidding results
In San Antonio, the bidding competition attracted a large number of bidders. In all, 79
suppliers submitted a total of 169 bids across the 5 product categories in the bidding
competition held in 2000. Oxygen, hospital beds, and wheelchairs each generated more
than 40 bids, and nebulizer drugs drew 33 bids. There were only 14 bids for general
orthotics, the category with the lowest total allowed charges. A total of 51 firms won
supplier status. There were 32, 24, 23, 8, and 11 winners in the oxygen, hospital bed,
wheelchair, orthotics, and nebulizer drug categories, respectively. (A firm could win in
more than one category.)
19
Market competition analysis in Polk County and San Antonio
The evaluation team analyzed changes in market share in each product category for
demonstration suppliers as a group vs. the remaining nondemonstration suppliers. They
also analyzed an index of market concentration for each product category.
The market share analysis showed gains in market share for winners tended to occur over
time, especially for categories under transition policies that grandfathered beneficiary
supplier relationships—oxygen, hospital beds, wheelchairs, and nebulizer drugs. Their
market share increased as new users entered the market, while the market share of
nondemonstration firms eventually fell, but not to zero, because of continuing long-term
patronage by grandfathered beneficiaries. For urological supplies, shares of
nondemonstration suppliers fell, but not to zero, due to the grandfathering of relationships
with nursing home residents (provided that the supplier accepted the competitively bid
fees), and due to a hold-harmless policy that allowed a two-month grace period if the
beneficiary patronized a nondemonstration supplier in error. Program policy for enteral
nutrition allowed nursing homes to honor contracts with nondemonstration suppliers.
Most enteral nutrition is supplied to nursing home residents, resulting in a small reduction
of market share for nondemonstration suppliers. For surgical dressings market share for
nondemonstration suppliers actually increased due to a combination of factors related to
the project’s transition policies for nursing home residents and a reduction in total
volume owing to changes in home health payment policy external to the competitive
bidding demonstration.9 Nondemonstration suppliers of orthotics saw their market share
drop modestly, in part because of grandfathered nursing home relationships.
The evaluation team used a technical index of market concentration in its assessment of
the demonstration’s impact on market competitiveness. This index is constructed from
the market shares of the firms serving a market. The measure is normally applied to
well-defined markets characterized from detailed economic information. In this analysis,
for convenience the Polk County and San Antonio supplier shares were assumed to
define the entire market for each product category, without conducting a market
definition study.
The results suggested that the demonstration had relatively little effect on concentration,
except for surgical dressings. Although observers might have expected concentration to
grow if the number of suppliers declines, this wasn’t necessarily the case. For example,
if a supplier with a large market share does not win demonstration status, the
concentration index can even decline, as smaller suppliers come to characterize the
market. Furthermore, the analysis took into account trends in comparison counties, and
in some product categories underlying concentration was trending upward. Surgical
dressings had an unusually small number of suppliers, and this category was highly
concentrated in Polk County before the demonstration. A change in supplier
9 Fee-for-service reimbursements to suppliers of surgical dressings dropped, probably because the home
health prospective payment system bundled these supplies into the agency’s lump-sum payment.
20
Configuration created a large increase in the concentration measure in Round 1 and a
large decrease in Round 2.
Analysis of individual firm shares over time showed, as expected, that winning suppliers
generally gained market share, while losing suppliers did not. Some winning suppliers
were particularly successful in adding to their market share, including some small
suppliers. However, a higher market share did not materialize for every winning bidder.
Other competition-related information: site visits, supplier surveys
Site visit results. Two demonstration firms in Polk County that filed for bankruptcy
during Round 1 won demonstration status again in Round 2. These bankruptcies, and
another affecting a nondemonstration firm, were not related to the demonstration.
During site visit discussions in San Antonio, suppliers revealed mixed opinions about
how continued competitive bidding might eventually affect market competitiveness.
With varying reliance on Medicare revenues, not all suppliers felt their survival was
threatened by Medicare bidding.
Supplier survey responses. Business and financial data collected informally from a
small sample of Polk County suppliers suggested that some of the demonstration
suppliers experienced higher volume. Of these, some also reported increased revenues,
notwithstanding the price reductions brought by competitive bidding. The sample,
however, was not necessarily representative. These same winning suppliers tended to
perceive the Polk County market as being more competitive as a result of the
demonstration.
Measures of financial health from the San Antonio supplier survey suggested that, as
expected, overall revenues declined due to the reduced prices. Also as expected,
demonstration suppliers were less likely to have reduced revenues and lower net income
under the demonstration than nondemonstration suppliers. Suppliers’ perceptions of
changes in market competitiveness in the period coinciding with the first year of the
demonstration depended on their demonstration status. Demonstration suppliers were
nearly twice as likely as nondemonstration suppliers to rate the market as equally or more
competitive. However, they were noticeably less likely than the Austin comparison-area
suppliers to assign this rating.
Discussion
The demonstration tested competitive bidding over a relatively short time period, 3 years.
Thus, analysis of long-term effects on market competition is beyond the scope of the
evaluation. Yet several observations, site visits, and analytic results from the evaluation
team provide signs that declining competitiveness is not a necessary consequence of
competitive bidding.
21
Theory suggests that competition is easier to maintain in markets with low barriers to
entry into the business. Low barriers to entry characterize DMEPOS supplier markets, in
part because these businesses are not capital-intensive, and without heavy regulatory
requirements. The bidding design featuring multiple winners potentially adds further
protection against shrinkage of market competitiveness. In fact, the analysis of market
concentration showed little impact of the demonstration on the concentration index,
probably due to the multiple winner design. Also, a multiple-winner design can motivate
more competition at the bidding stage, by increasing the probability of achieving winner
status. Selecting multiple winners could motivate continued competition in later bidding
rounds simply by sustaining more firms in the market.10 As noted earlier, the availability
of multiple winners allowed referral agents to choose among suppliers on the basis of
quality and service, fostering competition during the marketing stage.
Analysis of market shares for demonstration suppliers and nondemonstration suppliers
suggested that under the demonstration design permitting grandfathering of certain
beneficiary-supplier relationships, demonstration suppliers as a group enlarged their
market share, but not to 100 percent. Most nondemonstration firms continued to serve
existing patients, and in Polk County some losing bidders returned to the second round of
competition and gained demonstration status. Considering that firms can rely on
Medicare revenues to varying degrees, these results and experiences do not necessarily
imply that firms should feel threatened by competitive bidding for Medicare DMEPOS.
Firm-level market share analysis found increased market share was not guaranteed for
every winning supplier. This may be explained by site visit data indicating that certain
suppliers did not make marketing efforts, that out-of-area suppliers were avoided by
many beneficiaries, and that referral agents were selective in guiding beneficiaries to
suppliers they perceived as better performers.
In Polk County, the reduction in Round 2 bidders for urological supplies raised the
possibility that small profit margins deterred bidding by more urologicals suppliers in
Round 2; low profit margins were reported by some suppliers to be a problem, according
to early site visit information. But it is worth noting that this category had relatively low
total allowed charges at stake in the bidding competitions, with comparatively few
bidders and similarly small numbers of suppliers before the demonstration started.
Unless designs to bolster participation can be developed, such small-volume DMEPOS
categories may represent lower-priority areas for conducting competitive bidding, not
only in terms of the limited savings potential on a small dollar base but also in terms of a
category’s competitive potential.
Administrative feasibility of the reimbursement system
The evaluation of administrative feasibility addressed the ease of implementing the
process of competitive bidding and of administering the post-bidding phases, including
the transition to approved suppliers, new reimbursement procedures, and site monitoring.
The evaluation team also considered the net savings from the competitive bidding project
after accounting for estimated administrative costs. They further estimated costs under a
national program using the same administrative structure used in the demonstration.
10 Jeffrey S. McCombs and Jon B. Christianson, “Applying Competitive Bidding to Health Care,” Journal
of Health Politics, Policy and Law 12:4, Winter 1987, 703-722.
22
Estimates suggested favorable returns from competitive bidding—especially favorable
under an extension to additional competitive bidding areas. Extrapolating administrative
costs to a national program and assuming conservative savings, the team illustrated that
investing in a national program might bring savings twice as large as outlays.
Implementation and operations
The CMS essentially replicated the same competitive bidding model in both Polk County
and San Antonio. Substantial early efforts to educate beneficiaries, referral agents, and
suppliers about the demonstration helped to ease the transition to the competitively bid
fees and approved suppliers list. As noted earlier, transition policies also helped
stakeholders adjust.
By the second competition (San Antonio), CMS gained experience in bid evaluation
sufficient to allow some streamlining of bid processing. Experience also informed the
weighting formula for summarizing bid prices into a summary bid, which likely produced
some anomalous fees before it was changed.
The ombudsman was well accepted. The ombudsman conducted in-person information
sessions about the demonstration, responded to inquiries about the demonstration from all
stakeholders, coordinated bid evaluation site visits to suppliers, and generally served as
Medicare’s “eyes and ears” on-site. The ombudsman also monitored supplier
performance by investigating complaints and conducting routine inspections.
PGBA encountered few problems in automated processing of claims for the
demonstration areas. Because several ZIP Codes in Texas crossed into nondemonstration
counties, claims from these areas had to be pulled from the claims stream and manually
processed, but this affected relatively few claims.
A delay in issuing the San Antonio directory of approved suppliers appeared to cause
avoidable difficulties in making DMEPOS arrangements for some beneficiaries early in
the transition. In both sites, informants recommended earlier release of the directory.
This underscores the importance of allowing sufficient time for site stakeholders to
prepare for each changeover to new approved suppliers.
Demonstration savings net of costs
Costs of administering the demonstration were estimated to be $4.8 million (in Year 2000
dollars). These costs covered research and development activities begun in 1995,
subsequent public and supplier education, bidding and bid evaluation, modifications to
claims processing, and ongoing site monitoring until project termination in December
2002. Total estimated savings in the two demonstration sites since October 1999, when
the first competitive bid fees became effective in Polk County, through termination in
each site, were $9.4 million, of which $7.5 million are Medicare savings and $1.9 million
are beneficiary savings. This implies net savings to the Medicare program of $2.7
million.
23
Spreading the large fixed-cost component of the project over additional sites would likely
increase the return substantially. For example, the cost of adding the San Antonio site
was $310,000 in the first full year, during which bidding was conducted. When bidding
was not conducted, the annual costs were about $100,000. Over 3 years (2000 to 2002)
the San Antonio site cost $510,000 to run, versus estimated savings of approximately
$4.6 million. The actual net savings from adding more sites would depend on factors
such as the size and competitiveness of the market in the additional sites, and the
particulars of bidding design and administration.
Estimated costs of a national program
The evaluation team extrapolated the costs of the demonstration program to estimate
potential costs of a national competitive bidding program. They assumed that
competitions would be conducted in all 261 MSAs and that CMS would follow the basic
design of the demonstration (including a 2-year bidding cycle). These estimates were not
precise, but were developed as illustration. About 669 full-time equivalent personnel
would be needed (10 at CMS and 659 at durable medical equipment regional carriers),
and total annual costs on this basis were estimated at $68.9 million. This result is not a
forecast of expenses under the competitive bidding program legislated in section 302 of
the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (Public
Law 108-173). Under that program, Medicare competitive bidding is to be phased in
among 10 large MSAs by 2007, 80 MSAs by 2009, and additional areas thereafter. The
Secretary has authority to select items to be phased into the program, based on volume,
price, and savings potential.
Discussion
Medicare was able to test operational feasibility of DMEPOS competitive bidding in two
sites under the demonstration. No major difficulties surfaced during various phases, from
site preparation to claims processing and monitoring.
The directory of winning suppliers was a critical document from the point of view of
discharge planners and others who needed to quickly arrange DMEPOS for patients
during the transition to the fee schedule and list of approved suppliers. Timeliness of the
directory is therefore an important issue. In any future initiatives, Internet applications
should be helpful in issuing and updating the directory virtually instantaneously.
The evaluation team’s analysis of San Antonio costs and savings suggested that the
returns to an extension of competitive bidding to further sites can grow substantially, due
to economies of scale. The national program illustrative cost estimate of $68.9 million
compares favorably to a conservative assumption on savings, say, if only oxygen were
put up for bidding and savings were 10 percent instead of nearly 20 percent. Oxygen
allowed charges in the year 2000 were $1.77 billion. If Medicare savings were $141.6
billion under competitive bidding (80 percent of $177 million in saved allowed charges),
the savings would be twice the estimated program costs.
24
Recommendations on products’ suitability for bidding
Considering the evidence from several criteria developed by the evaluation, the
evaluation team considered the implications for product selection under a national
program of competitive bidding. The criteria included allowed charges and potential
savings, number of suppliers, problems reported during the demonstration, and the
possible impact on prices of exclusions such as enteral nutrition in nursing homes.
The evaluators judged that several products were “well suited” (oxygen equipment and
supplies, hospital beds and accessories, and nebulizer drugs), another product category
was “potentially well suited” (wheelchairs), and several were “not as well suited” for
competitive bidding (surgical dressings, general orthotics, urological supplies, and enteral
nutrition). The categories viewed as “not as well suited” tended to have relatively low
allowed charges and low numbers of suppliers. Both characteristics may not indicate
sufficient potential for total savings and competition that would lead to significant price
reductions. In contrast, the “well suited” categories had high allowed charges and
relatively many suppliers. Wheelchairs had high allowed charges but there was evidence
of quality problems in this category.
The Final Evaluation Report provides more detailed discussion of these issues. The
discussion therein sets forth sound principles for evaluating future competitive bidding
markets. However, the team’s judgments on particular product categories should not be
taken as definitive for guiding future bidding initiatives. Market analysis of individual
competitive bidding areas could reveal that total savings and potential for competition are
likely for some of the categories judged “not as well suited” for bidding based on the
demonstration experience.
Conclusion
The broad variety of data used to evaluate the DMEPOS competitive bidding
demonstration suggested that the tests in Polk County and San Antonio largely met
Medicare’s objectives in terms of program savings; maintaining access, quality, and
product selection; preserving competition in DMEPOS markets; and administrative
feasibility. Savings estimates were about one-fifth relative to payments under the
statutory fee schedule, and they compared favorably to costs of running the program. In
general, access and quality changed little, and market competitiveness appeared stable.
The CMS demonstrated a workable competitive bidding design and feasible operating
procedures and policies.
This does not assume that some suppliers’ behavior remained completely the same. It is
logical to think that cost-saving measures will be pursued when prices fall. Policymakers
are interested in whether any behavioral changes to reduce costs were counterproductive
for beneficiaries and the Medicare program, or whether they represented efficiency
improvements. Further, they are interested in whether any new value was added to the
services (some of which may be cost-increasing). The evaluation revealed examples of
value added to beneficiaries’ services: improved product reliability; easier telephone
25
access to suppliers; more attention to insurance procedures at the start of the
beneficiary/supplier relationship; and higher frequency of portable oxygen refills.
Given the controversy surrounding Medicare competitive bidding, cost-saving behaviors
attract more attention. The evaluation study provided possible examples of these, too.
Examples affecting subgroups of beneficiaries included more provision of used vs. new
mattresses to hospital bed users; more use of mail delivery and less use of home delivery;
possibly, separate billing for wheelchair accessories previously informally bundled into
the wheelchair fee; fewer excess supplies; fewer maintenance visits; more use of oxygen
conserving devices on portable oxygen equipment; and less provision of portable oxygen.
Some examples may be seen as a benefit by beneficiaries (e.g., oxygen-conserving
devices simplifying logistics of travel outside the home). The shifts do not appear
harmful or pervasive enough to be a concern, and some observers may consider specific
changes justifiable from an efficiency standpoint.
A risk of lower access to portable oxygen is probably the foremost concern raised by the
evaluation results. Equally, the evidence on portable oxygen highlights the problem of
how the Medicare program can achieve an appropriate and efficient allocation of portable
oxygen to beneficiaries who need it and will use it, under either competitive bidding or
current payment methods. Policy tools such as stakeholder education, improvements in
data, and revisions to payment procedures may all have a role to play in meeting this
challenge.