Economic Value of Prosthetic Services Among Medicare Beneficiaries: A Claims-Based Retrospective Cohort Study
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
MILITARY MEDICINE, 181, 2:18, 2016
Economic Value of Prosthetic Services Among Medicare
Beneficiaries: A Claims-Based Retrospective Cohort Study
Allen Dobson, PhD*; Audrey El-Gamil, BA*; Matt Shimer, PhD†; Joan E. DaVanzo, PhD, MSW*
ABSTRACT Context: There are few studies of the economic impact or value of lower extremity prosthetic services.
Results from this study can inform the value proposition concerning prosthetic services within military health, where
Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_2/18/4209395 by guest on 27 January 2020
over 40,000 Veterans with limb-loss receive care for their amputations through the Veterans Administration health care
system. Purpose: To determine the extent to which Medicare patients who received selected prosthetic services had
less health care utilization, lower Medicare payments, and/or fewer negative outcomes compared to matched patients
not receiving these services. Methods: This retrospective cohort analysis using Medicare claims data (2007–2010) and
propensity score matching techniques to control for observable selection bias based on etiological diagnosis,
comorbidities, patient characteristics, and historical health care utilization one year before the etiological diagnosis.
Findings: Patients who received lower extremity prostheses had comparable Medicare episode payments ($6,099 per-
member-per-month for study group, $6,015 per-member-per-month for comparison group) and better outcomes than
patients who did not receive prostheses. Study group patients were more likely to receive extensive outpatient therapy
than comparison group patients ( p < 0.05). Receiving physical therapy is associated with fewer hospitalizations and
emergency room visits, and less facility-based care ( p < 0.05), essentially offsetting the cost of the prosthetic over a
12-month time frame.
INTRODUCTION to expect the reduction in health care utilization by Medicare
Lower extremity prosthetic devices and related clinical ser- beneficiaries with lower limb prosthetic services to cover the
vices are designed to provide stability and mobility to patients cost of the device and to improve a variety of patient outcomes.
with lower limb loss. There are limited studies of the extent The use of assistive technology has become more wide-
to which Medicare beneficiaries who receive these services spread over the past three decades, given the recent growth in
experience reductions in complications and/or health care both the aged population and military service–related traumas.
costs. The “lessons learned” from this study can inform the Currently, there are approximately 2 million individuals in the
value proposition of lower limb prosthetics in military health, United States who are living with limb loss. The total number
where over 40,000 amputees receive their care through the of individuals with an amputation, and those using prostheses,
Veterans Administration.1 is expected to reach 2.4 million by the year 2020.2
This study’s primary objective was to determine the eco-
nomic value of receiving lower limb prosthetics in terms of Evidence Regarding Patient Outcomes—Functional
the beneficiary’s total health care utilization and Medicare Ability and Psychosocial Issues
expenditures. Specifically, the study aimed to determine the Although the variability in measures of quality and patient
financial benefit to Medicare when mobility is restored for a outcomes in research on prosthetic services can make com-
person with limb loss through receipt of a prosthetic device. parisons difficult, a number of studies have shown that the
Financial benefit, or economic value, was determined based provision of a prosthetic device led to measurable improve-
on the health care utilization and costs for those beneficiaries ments in the quality of patient care and functional and
who received these services, compared to clinically similar psychosocial outcomes. Beyond physical health, receipt of
beneficiaries who did not receive prosthetic devices. This prosthetics is associated with improved mental health status
value can be applied directly to the Medicare program, and and social functioning, and reduced role limitation because
indirectly, but perhaps more powerfully, to the beneficiary’s of emotional problems.3 Research also suggests that the
quality of life. receipt of prosthetics can lead to benefits beyond the reduc-
A review of the literature suggests that receipt of a pros- tion of health care utilization and provider payments, such
thetic device improves patient outcomes on a variety of func- as societal gains including returning to work and reduced
tional and psychosocial measures. These services can reduce reliance on social services.
health care spending for some patients by preventing down-
stream clinical complications and reducing other types of
Evidence Concerning Cost–Benefit and
health care utilization. Based on the literature, it is reasonable
Economic Value
Long-term savings are thought to result when patients
*Dobson DaVanzo & Associates, LLC, 450 Maple Avenue East, Suite
303, Vienna, VA 22182.
receive appropriate prosthetic care. Without prosthetic care,
†hMetrix, 150 Monument Road, Suite 107, Bala Cynwyd, PA 19004. individuals live more sedentary lifestyles, which research
doi: 10.7205/MILMED-D-15-00545 has shown to lead to secondary complications, such as
18 MILITARY MEDICINE, Vol. 181, February Supplement 2016Economic Value of Prosthetic Services Among Medicare Beneficiaries
diabetes and other related clinical issues as well as increases includes one year of claims before and at least 12 months
in health care utilization and spending.4 following the receipt of the service. Within the custom
cohort database, CMS also provided Medicare claims from
2007 to 2010 across all settings for patients who did not
Current and Future Access to Prosthetics
receive lower extremity prosthetic devices. This population
Despite research that suggests that orthotics and prosthetics
served as the matched comparison group. CMS identified
(O&P) services can prevent falls, reduce downstream clinical
the comparison group patients by matching them to the
manifestations such as the development of diabetes, and lead
patients who received prosthetics (study group) based on the
to long-term savings in health care spending, patients can
presence of an etiological diagnosis, gender, age, and state
face significant barriers to access. Creation of parity laws or
of residence. CMS provided five comparison group patients
Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_2/18/4209395 by guest on 27 January 2020
inclusion of prosthetics as “essential benefits” are likely the preliminarily matched to each study group patient.
only way to ensure appropriate access to patients with these
To estimate the proportion of Medicare services (not
disabilities. Nineteen states have passed laws ensuring that
patients) reflected in our custom cohort database, we com-
people with limb loss have fair and appropriate access to
pared the lower extremity prosthetic utilization by Healthcare
prosthetic care under private insurance, and seven states
Common Procedure Coding System Code in the custom
have included orthotics in those laws as well. These 19 states
cohort dataset in calendar year 2008 to the utilization reported
include Arkansas, California, Colorado, Indiana, Iowa, Illinois,
in the Physician/Supplier Procedure Master Summary File, an
Louisiana, Maine, Maryland, Massachusetts, Missouri, New
aggregate procedure level claims file released by CMS that
Hampshire, New Jersey, Oregon, Rhode Island, Texas, Utah, includes 100% of Part B Carrier and durable medical equip-
Vermont, and Virginia.5
ment Regional Carrier claims). We estimate that the sample
methodology presented above captured approximately 37% of
METHODS the lower extremity prostheses during this period. The low
The analytic methodology consisted of three key compo- proportion of lower extremity prostheses included in our cus-
nents, including: (1) developing clinical episodes of care for tom cohort dataset was due to the requirement that the pros-
each individual beneficiary; (2) developing patient cohorts thetic must have been provided to a patient within 12 months
of lower extremity prosthetics users and a matched compari- of his/her amputation. Therefore, our results can only be
son group using the Medicare claims; and (3) calculating extrapolated to “new prosthetic” users and not the overall
descriptive statistics and analyzing outcomes associated with Medicare prosthetic population.
lower extremity prosthetics on overall Medicare episode
payments for both cohorts. Developing Patient Episodes
All of the analyses were conducted using Medicare
We developed patient episodes that would capture health care
claims for a custom cohort requested from the Centers for
diagnoses, utilization, and expenditures before and after, receipt
Medicare & Medicaid Services (CMS) (Use Agreement No.
of the prosthetic device. Figure 1 shows the episode structure.
21558). We requested claims across all settings from 2007
All study group patient episodes contained the following
to 2010 for patients who received and did not receive lower
key features.
extremity prosthetics and services. This database served as
the analytic sample for all of our analyses.
Receipt of Prosthetic Service
The sampling methodology utilized by CMS to extract
the custom cohorts allowed our analyses to reflect those Across all episodes, the study group patient must have
Medicare beneficiaries who received specified prosthetic ser- received the prosthetic services between January 1, 2008
vices between January 1, 2008 and June 30, 2009. Patients and June 30, 2009. This allowed us to maximize sample
were required to have received the prostheses during the size, as only patients with 18 months of claims were consid-
specified time period and must also have had appropriate eti- ered for matching.
ological diagnoses and an amputation within 12 months
before the receipt of the prosthetic. Codes used to identify Etiological Diagnosis for Which Patient Receives Service
an amputation and the etiological diagnoses of interest for The etiological diagnosis was for the condition which ulti-
each group are included in a separate technical methodology mately led to the need for the lower extremity prosthetic ser-
available from the authors. vice (likely functional diagnosis), not the diagnosis linked to
Health care claims across all care settings from 2007 to the claims at the time of receipt of the service. The etiologi-
2010 were obtained for the beneficiaries who met sampling cal diagnosis is used to match the prosthetic users to
specifications. Care settings include inpatient and outpatient nonusers (study to comparison group) and must be present
hospitals, long-term care hospitals, skilled nursing facilities, during the preservice window. Furthermore, etiological diag-
inpatient rehabilitation facilities, home health agencies, hos- nosis is also used to control for mortality across groups. The
pice, physician/carrier visits, and durable medical equipment, etiological diagnosis (as defined by the Agency of Healthcare
prosthetics, orthotics, and supplies. Therefore, the database Research and Quality’s Clinical Classification Software6)
MILITARY MEDICINE, Vol. 181, February Supplement 2016 19Economic Value of Prosthetic Services Among Medicare Beneficiaries
Figure 1.
Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_2/18/4209395 by guest on 27 January 2020
Structure of the patient episode definitions for the study group with embedded illustrative example. O&P, orthotics and prosthetics.
was identified with assistance from the study’s clinical com- Developing Patient Cohorts
mittee. The distribution of possible etiological diagnoses is Based on the patient episode definitions described above, we
presented in Table I. identified two patient cohorts: those who had the etiological
diagnosis and received prosthetic services (i.e., the study
“Clean” Period Before Etiological Diagnosis group) and those who had the etiological diagnoses but did
To ensure proper matching to the comparison group, we not receive the prosthetic service (i.e., the comparison
required a three-month minimum clean period for each patient group). The comparison group was matched to the study
episode before the etiological diagnosis to prevent the study group through propensity score matching. (Before matching,
group from containing patients with a lengthy history of the average age for study group was 69.47 years, 74.32 for com-
etiological diagnosis, which may have impacted the clinical parisons [ p < 0.0001] and after matching, average age for
outcome as well as their use of the prosthetic service. study group was 72.9 years, 72.5 years for comparisons
[ p < 0.7263].)
Propensity score matching techniques are widely used in
Preservice Window Before the Receipt of the Prosthetic Service observational studies when randomized controlled trials are
The etiological diagnosis was identified within 12 months not possible or are unethical or impractical to administer.7
before the receipt of services (preservice window). This Literature suggests that applying these techniques to obser-
preservice window also allowed us to identify comorbid con- vational studies is sufficient to remove observable selection
ditions, patterns of institutional care, and other health care uti- bias among treatment and comparison groups and can result
lization used for risk adjustment during the matching process. in findings that look like randomized controlled trials.8–11
We used propensity scores to identify a one-to-one match
Postservice Window across study group and comparison group patients based on
Postservice period captured up to 12 months after receiving etiological diagnosis, comorbidities, patient characteristics
the prosthetic service used to track trends in overall health (age, gender, race), and historical health care utilization one
care utilization and expenditures. year before the etiological diagnosis. Patients were also
matched on death to further control for selection bias. That
TABLE I. Etiological Diagnoses for Lower Extremity Prostheses is, if a study group patient died within their episode, they
(2008–2010) were matched to a similar comparison group patient who
died during their postservice window. Following the
Etiological Diagnosis Percent of
matching, any matched pair that died within three months of
(Clinical Classification Software Category) Matched Pairs
the etiological diagnosis was excluded from the analysis.
Diabetes Mellitus with Complications 24.3 Since comparison group patients do not have an index date
Peripheral and Visceral Atherosclerosis 22.2
Skin and Subcutaneous Tissue Infections 15.0
(date they received the O&P service), etiological diagnosis
Other Non-traumatic Joint Disorders 12.2 was the only metric we identified that was consistently
Chronic Ulcer of Skin 11.9 defined among the study and comparison group patients.
Other Circulatory Disease 4.4
Complication of Device; Implant or Graft 3.5
Open Wounds of Extremities 2.6 Role of Mortality
Gangrene 2.3
Septicemia (Except in Labor); 1.6
Before matching study to comparison group patients, we
Rehabilitation Care; Complications of compared the mortality rates of patients who received O&P
Surgical Procedure services to those who did not. This was particularly impor-
tant among lower extremity prosthetic patients, as the clini-
Source: Dobson DaVanzo analysis of custom cohort Standard Analytic Files
(2007–2010) for Medicare beneficiaries who received orthotics and prosthet-
cal severity (and risk of imminent death) may have been a
ics services from January 1, 2008 through June 30, 2009 (and matched com- driver of whether the patient received a prosthetic or not. By
parisons), according to custom cohort database definition. matching patients on whether or not they died, we were able
20 MILITARY MEDICINE, Vol. 181, February Supplement 2016Economic Value of Prosthetic Services Among Medicare Beneficiaries
to more appropriately compare Medicare episode payments conclusions could no longer be drawn with regard to the
across groups. As a result, mortality across the groups was health care utilization and payments for the study group
excluded as a study outcome. compared to the comparison group. The temporal autocorre-
lation function is defined as,
Determining Post-Service Window for Comparison
hðX 1 μ 1 ÞðX t μ t Þi
Group Patients A ðt Þ ¼
σ1 στ
The study group patients’ postservice window was triggered
by the receipt of the prosthetic service. Since comparison where Xt is the Medicare episode payments for a beneficiary
group patients did not receive a prosthetic service, we devel- after t months, μτ is the mean Medicare episode payments
of all beneficiaries after t months, and στ is the standard
Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_2/18/4209395 by guest on 27 January 2020
oped a proxy start date for comparison group patients.
In order to ensure the same postservice window for which deviation of the Medicare episode payments after t months.
health care utilization and expenditures were tracked and The temporal autocorrelation function measured the corre-
compared across cohorts, the length of the comparison lation between the Medicare expenditures of a beneficiary’s
group’s episodes was contingent on the episode length of first month (index month) and his/her Medicare expenditures
the study group. Proper matching of the study and compari- at each month within the episode. As expected, the correlation
son group patients limited the number of episodes included decreased as the episode length increased as unrelated acute
in our study, but helped to ensure that the study and com- events and underlying chronic conditions began to dominate
parison group patients were clinically and demographically overall episode expenditures as opposed to receipt of the
similar.12 Figure 2 shows the number of study group and eti- prosthetic service.
ological patients included in each service group before and The correlation dropped rapidly until Month 12, then
after matching. The relatively small number of lower plateaued for the remainder of the episode. This plateau in the
extremity prostheses was due to the required amputation temporal autocorrelation function signified that an underlying
within 12 months of receipt of the prosthetic. As noted confounding correlation may have been dominating the
above, this excluded long-term users who received replace- effects of the treatment after 12 months. As a result, we could
ment prosthetics during the study window. not distinguish between the effect of the treatment (receipt
of the prosthetic service) and unrelated expenditures after
Calculating Descriptive Statistics and Analyzing 12 months. Accordingly, we limited episodes to 12 months
Impact of Lower Extremity Prosthetic Devices on (as opposed to 18 months) to more precisely measure treat-
Overall Patient Medicare Expenditures ment effects and outcomes.
Descriptive statistics were calculated for the study and com- Across both study and comparison cohorts, we compared
parison cohorts after the propensity score matching. As the average Medicare per-member-per-month (PMPM) pay-
noted above, the two groups were compared to each other ment, distribution of payments by care settings, and outcome
based on the distribution of patient characteristics including measures, such as falls, hospitalizations, and days of rehabil-
but not limited to age, gender, race, and comorbidities. itative/physical therapy.
We used a temporal autocorrelation function in order to
determine an episode length that would capture the effects Data Limitations
of the prosthetic service without capturing the effects of
The key limitation of our methodology was the reliance
other comorbidities or unrelated events. This was a critical
on administrative data as opposed to clinical data recorded
analytic step, as once the Medicare episode payments were
in medical records. While our dataset included all fee-
no longer correlated to the receipt of the targeted service,
for-service health care utilization and payments, clinical
indicators, such as functional status, were not available
from our administrative data. Propensity score matching
relied on all patient demographic and clinical characteris-
tics in order to control for observable selection bias among
those who received prosthetic services compared to those
who did not, thereby isolating the effect of receiving the
prosthetic service.
Another limitation of the claims data was the lack of
Medicare Advantage discharges and Medicaid long-term
care-related expenses for dually eligible patients. The rela-
tionship of the Medicare to Medicaid payment systems is
problematic for analyses that involve episodes of care, as the
Figure 2. Distribution of pairs (study group and comparison group exclusion of Medicaid claims for dually eligible patients
matches). O&P, orthotics and prosthetics. prohibit us from identifying patients who receive care in
MILITARY MEDICINE, Vol. 181, February Supplement 2016 21Economic Value of Prosthetic Services Among Medicare Beneficiaries
TABLE II. Descriptive Statistics Across Matched Pairs 72.9 years of age (comparison group patients were 72.5 years
(2008–2010) of age) and approximately one-third of patients in both groups
Percentage
were female. More than two-thirds of patients included in
Study Comparison Point the matched pairs were Caucasian.
Demographic Characteristic Group Group Difference Table III presents health care utilization and payments by
Number of Beneficiaries 428 428 0.00 care setting for those who received lower extremity prosthe-
Average Age 72.9 72.5 0.40 ses (study group) compared to those who did not (compari-
Duals 40.2% 45.3% −0.05 son group).Across the 12-month episode, the study group
Female 36.7% 37.6% −0.01 patients had average PMPM Medicare payments across all
Death 45.6% 45.6% 0.00
care settings that were only slightly (not significantly) higher
Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_2/18/4209395 by guest on 27 January 2020
Black 22.7% 29.9% −0.07
White 70.8% 64.0% 0.07 than the comparison group ($6,099 for the study group com-
Hispanic 3.3% 3.5% 0.00 pared to $6,015 for the comparison group). About one-quarter
of the total PMPM payment for the study group patients is
Source: Dobson DaVanzo analysis of custom cohort Standard Analytic Files
(2007–2010) for Medicare beneficiaries who received orthotics and prosthet-
attributed to the prosthetic ($1,554 of the total PMPM of
ics services from January 1, 2008 through June 30, 2009 (and matched com- $6,099). The prosthetic cost was nearly amortized within
parisons), according to custom cohort database definition. 12 months due to a reduction of care in other care settings.
As shown in Table III, the second largest driver of
PMPM Medicare payment among the study group (and the
largest among the comparison group) was the acute care
long-term care facilities as compared to the community. hospital admission. Study group patients had a lower rate
Additionally, with 50 different Medicaid program policies of hospitalization than the comparison group patients
reflected in the data for dual eligibles, there is variability for (1.18 admissions for the study group compared to 1.51 admis-
which we cannot explicitly account. sions for the comparison group), resulting in lower PMPM
Medicare payments for acute care hospitalizations ($1,498 for
RESULTS the study group compared to $2,338 for the comparison group)
Although lower extremity prostheses are often provided to ( p < 0.05).
younger (not Medicare eligible) beneficiaries because of Study group patients were more likely to remain in the
trauma or disease progression, we only investigated the home and receive outpatient visits and less likely to receive
impact of prostheses on Medicare beneficiaries with a recent facility-based care including Inpatient Rehabilitation Facilities,
amputation. Our propensity score matching resulted in 428 Long-Term Care Hospital, Other Inpatient, and hospice ser-
matched pairs of recent amputees who received a pros- vices than the comparison group patients ( p < 0.05). Study
thetic matched to a new amputee who did not. These patients group patients had higher outpatient visits (11.37 visits com-
were matched on demographic and clinical characteristics and pared to 9.52 visits; p < 0.05) and comparable home health
are accordingly risk-adjusted. admissions (1.29 admissions compared to 1.18 admissions),
Table II presents the descriptive statistics and the distri- and skilled nursing facility admissions (1.48 admissions
bution of matched patients by etiological diagnosis. On aver- compared to 1.81 admissions) to comparison group patients.
age, patients who received lower extremity prostheses were Study group patients also had lower Medicare payments for
TABLE III. Distribution of Claims and Per-Member-Per-Month (PMPM) Payments by Care Settings (2008–2010)Care Setting
Study Group Comparison Group Difference
Average Average Average Average Average Average
Number of Claims PMPM Number of Claims PMPM Number of Claims PMPM
Physician 53.94 $649 60.68 $990 −6.75 −$341*
Durable Medical Equipment 14.02 $1,554 9.00 $211 5.02* $1,343*
Outpatient 11.37 $781 9.52 $608 1.86* $173*
Skilled Nursing Facility 1.48 $699 1.81 $735 −0.33 −$36
Home Health 1.29 515 1.18 474 0.11 41
Acute Care Hospital 1.18 1,498 1.51 2,338 −0.34* −839*
IRF, LTCH, Other Inpatient, 0.35 402 0.55 658 −0.20* −256*
and Hospice
Total 83.63 6,099 84.26 6,015 −0.63 85
Source: Dobson DaVanzo analysis of custom cohort Standard Analytic Files (2007–2010) for Medicare beneficiaries who received orthotics and prosthetics
services from January 1, 2008 through June 30, 2009 (and matched comparisons), according to custom cohort database definition. IRF, Inpatient Rehabilita-
tion Facilities; LTCH, Long-Term Care Hospital. *Statistically significant at p < 0.05.
22 MILITARY MEDICINE, Vol. 181, February Supplement 2016Economic Value of Prosthetic Services Among Medicare Beneficiaries
TABLE IV. Lower Extremity Prostheses: Average Use of Inpatient and Outpatient Therapy and Patient Outcomes by Cohort
(2008–2010)
Therapy Use and Outcomes Study Group Comparison Group Difference
Average Number of IRF Days 1.61 1.19 0.42
Average Number of Outpatient Therapy Visits 56.1 28.9 27.18*
Average Number of Fractures and Falls 0.90 0.72 0.18
Average Number of Emergency Room Admissions 1.55 2.10 −0.55*
Total Average Medicare Episode Payments $68,040 $67,312 $728
Source: Dobson DaVanzo analysis of custom cohort Standard Analytic Files (2007–2010) for Medicare beneficiaries who received orthotics and prosthet-
Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_2/18/4209395 by guest on 27 January 2020
ics services from January 1, 2008 through June 30, 2009 (and matched comparisons), according to custom cohort database definition. IRF, Inpatient
Rehabilitation Facilities. *Statistically significant at p < 0.05.
physician visits, which may indicate improved patient clini- admitted to the emergency room less often than comparison
cal stability that is not otherwise captured directly through group patients (1.55 admissions compared to 2.10 admis-
administrative claims ($649 PMPM compared to $990 sions; p < 0.05).
PMPM; p < 0.05). Across the entire 12-month episode, study group patients
Patients need to be trained and receive extensive therapy had an average Medicare episode payment that was compa-
to properly use a prosthetic device. One driver of the differ- rable (only $728 higher or approximately 1% of $68,040) to
ence in the PMPM payment is the use of therapy. As shown the comparison group, including the cost of the prosthetic.
in Table IV, study group patients had considerably higher Figure 3 presents the cumulative episode payment for the
utilization of both outpatient therapy (56.1 visits compared study and comparison group by episode month. This chart
to 28.9 visits). Both inpatient and outpatient therapy sessions indicates that the cost of the prosthetic in Month 1 was
are critical for patients with the prosthetic to help teach them slowly amortized over time; by the end of Month 12, the
balance and mobility with their new device. Additionally, the cost of the prosthetic was fully amortized.
high use of therapy may be associated with increased ambu-
lation, which suggests that the study group patients with
prostheses were less bedbound than the comparison group. DISCUSSION
Adverse events or outcomes, defined as the number of The literature indicates that the receipt of prosthetic services
fractures and falls and emergency room admissions, were could increase a patient’s mobility, ultimately reducing their
also comparable or higher for the study group patients. health care utilization and increasing their quality of life.
Despite the increased independence of study group patients, Based on this possibility, this study investigated the eco-
the number of falls and fractures was comparable to compar- nomic impact and value of lower extremity prostheses.
ison group patients. However, study group patients were Using propensity score matching techniques to compare
clinically and demographically similar patients who received
these services to those who did not, we were able to deter-
mine the economic impact of these services on the Medicare
population. This study excludes other sources of economic
value and outcomes, such as the ability for patients with
prosthetics to return to work or become more independent
from social services.
Our analyses show that over a 12-month period, patients
who received lower extremity prosthetics had reduced
Medicare payments such that the cost of the prosthetic was
essentially “covered.” Through a reduction in acute care hos-
pitalizations, physician visits, and facility-based care, patients
experienced better quality of life at a comparable Medicare
episode payment. Despite a comparable number of fractures
and falls among lower extremity prosthetic users, the rate of
emergency room admissions was lower than for those who
did not receive the service. Part of the savings as a result of
Figure 3. Lower extremity prostheses: cumulative medicare episode pay- reduced facility-based care was offset by extensive physical
ment by cohort (18-month episodes from 2008 to 2010). Source: Dobson therapy and rehabilitation to teach patients how to properly
DaVanzo analysis of custom cohort Standard Analytic Files (2007–2010) use their prostheses.
for Medicare beneficiaries who received orthotics and prosthetics services
from January 1, 2008 through June 30, 2009 (and matched comparisons), If used properly, lower extremity prostheses have the
according to custom cohort database definition. potential to increase quality of life and reduce facility-based
MILITARY MEDICINE, Vol. 181, February Supplement 2016 23Economic Value of Prosthetic Services Among Medicare Beneficiaries
care for newly amputated Medicare beneficiaries. Across all 2010. Available at http://www.legislature.maine.gov/opla/IFS2009.pdf;
analyses presented above, our results suggest that prosthetic accessed on November 16, 2015.
5. The Orthotic and Prosthetic Alliance: Inclusion of Prosthetic and
services provide value to the Medicare program, as well as a
Orthotic Coverage in the Essential Health Benefits Package under the
value to the patient. The cost of the services are nearly, if Affordable Care Act. Available at http://www.oandp.org/assets/pdf/
not completely, amortized through reduced acute care hospi- WhitePaper-OPandEHB.pdf; accessed on November 16, 2015.
talizations and facility-based care. 6. Healthcare Cost and Utilization Project (HCUP): HCUP CCS. 2014.
Rockville, MD, Agency for Healthcare Research and Quality. Available
at www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp; accessed on Novem-
ACKNOWLEDGMENT ber 16, 2015.
This study was supported by the Amputee Coalition and the American 7. Trojano M, Pellegrini F, Paolicelli D, Fuiani A, Di Renzo V: Observa-
Downloaded from https://academic.oup.com/milmed/article-abstract/181/suppl_2/18/4209395 by guest on 27 January 2020
Orthotic and Prosthetic Association. tional studies: propensity score analysis of non-randomized data. Int
MS J 2009; 16: 90–7.
8. Austin PC: An introduction to propensity score methods for reducing
REFERENCES the effects of confounding in observational studies. Multivariate Behav
1. 2015 Federal Advanced Amputation Skills Training (FAAST) Sympo- Res 2011; 46: 399–424.
sium, May 19–21, 2005. Available at http://www.army.mil/article/ 9. Kuss O, Legler T, Borgermann J: Treatments effects from randomized
149964/Joint_DoD_VA_training_symposium__collaboration_keeps_experts_ trials and propensity score analyses were similar in similar populations
on_cutting_edge_of_limb_loss_care/; accessed November 16, 2015. in an example from cardiac surgery. J Clin Epidemiol 2011; 64(10):
2. American Academy of Orthotists and Prosthetists: O&P trends and sta- 1076–84.
tistics, 2006. Available at http://www.oandp.org/grants/awareness/06_ 10. Dehejia R, Wahba S: Propensity score-matching methods for non-
statistics.pdf; accessed November 16, 2015. experimental causal studies. Rev Econ Stat 2002; 84(1): 151–61.
3. Samuelsson KA, Toytari O, Salminen AL, Brandt A: Effects of lower 11. Rosenbaum PR, Rubin DB: The central role of the propensity score
limb prosthesis on activity, participation, and quality of life: a system- in observational studies for causal effects. Biometrika 1983; 70(1):
atic review. Prosthet Orthot Int 2012; 36(2): 145–58. 41–55.
4. Department of Professional and Financial Regulation. State of Maine: A 12. Austin PC: Optimal caliper widths for propensity-score matching when
report to the Joint Standing Committee on Insurance and Financial Ser- estimating differences in means and differences in proportions in obser-
vices of the 124th Maine Legislature: review and evaluation of LD20, vational studies. Pharm Stat 2011; 10: 150–61.
24 MILITARY MEDICINE, Vol. 181, February Supplement 2016You can also read
