Recently published data from the Lower Extremity Assessment Project (LEAP)
study indicate that functional outcomes are similar following amputation or
reconstruction for the treatment of a limb-threatening lower-extremity
injury1-3.
Given this similarity in outcomes, it is important to examine health-care
costs associated with these two treatment pathways so that the trade-offs
between costs and effectiveness can be adequately addressed. The few
investigators who have examined the costs of amputation and reconstruction
generally have agreed in their conclusions that hospital and physician fee
charges for limb salvage often exceed those for early
amputation4-9.
Coupled with evidence of little difference in treatment outcomes, one might
conclude from those studies that early amputation is a reasonable strategy for
saving money. This conclusion has been challenged, however, as the studies
conducted to date have been based on small numbers of patients and have not
adequately taken into account longer-term treatment costs and the cost of
purchasing and maintaining prosthetic
devices4,8.
The goal of the present study was to compare total two-year health-care and
prosthesis-related costs incurred by patients enrolled in the LEAP study. We
hypothesized that, although total medical treatment costs for patients treated
with reconstruction would exceed those for patients undergoing amputation, the
long-term costs of prosthetic devices and related services would be high for
amputees, resulting in higher total two-year costs for amputation than for
reconstruction. On the basis of less-detailed seven-year follow-up data and
estimates of life expectancy, we also projected lifetime costs for the two
groups of patients.
Setting and Patient Population
Atotal of 601 patients were enrolled in the LEAP study over a forty-month
period (March 1994 through June 1997) and were followed prospectively in
person at three, six, twelve, and twenty-four months after the
injury1. At
approximately eighty-four months after the injury, an attempt was made to
locate and interview by telephone all LEAP participants who had completed at
least one follow-up interview within the first two years after the
injury2,3.
Patients were eligible for the LEAP study if they were sixteen to
sixty-nine years old and had been admitted to one of eight level-I trauma
centers for treatment of one or more of the following injuries to the lower
extremity:
Gustilo10
type-IIIB, IIIC, and selected type-IIIA fractures; dysvascular limbs (knee
dislocations, closed tibial fractures, or penetrating wounds with vascular
injury); major soft-tissue injuries (degloving or severe crush/avulsion
injury); and severe foot and ankle injuries (Gustilo type-IIIB ankle and all
Gustilo type-III intra-articular distal tibial [pilon] fractures and severe
hindfoot or midfoot injuries). Exclusion criteria have been previously
described; most notably, patients with a major brain injury (a Glasgow Coma
Scale score of <15 points at twenty-one days after the injury or at the
time of discharge), a spinal cord deficit, a prior amputation, or a
third-degree burn of the injured
limb1.
For the present cost analysis, thirty-two patients with bilateral injuries
were excluded as were twenty-four patients who had been enrolled in the study
at the time of discharge but could not be located subsequently. Of the
remaining 545 patients, 149 underwent amputation during the initial
hospitalization (including thirty-seven who had a traumatic amputation). After
the initial hospital discharge, twenty-seven additional patients underwent
amputation: twelve had the amputation within three months after the injury;
six, between three and six months; seven, between six and twenty-four months;
and two between twenty-four and eighty-four months. There were no significant
differences in the sociodemographic characteristics between the patients who
had undergone an amputation and those who had undergone a
reconstruction1. The
majority of the study patients were male (76.7%) and between the ages of
twenty and forty-five years (72.1%). Approximately one-third (35.2%) of the
LEAP patients lived in households below the federal poverty line at the time
of the injury. An additional 21.3% lived in households classified as near-poor
(a household income within 125% to 200% of the federal poverty line). Most
injuries resulted from motor-vehicle (29%), motorcycle (22%), or
pedestrian-vehicular (13%) collisions.
As previously
reported1, rates of
two-year postinjury follow-up exceeded 80% (92.1% were followed for three
months; 92.3%, for six months; 90.5%, for twelve months; and 84.4%, for
twenty-four months). The rate of follow-up for eighty-four months was somewhat
lower (75.8%)2.
Patients with incomplete data due to attrition were more likely to be male and
without a high-school education (p < 0.05). However, attrition rates did
not differ significantly between the two treatment groups (amputation and
reconstruction) or according to the extent and severity of the lower-extremity
injury.
The study was approved by the institutional review boards at the
coordinating center and at each study site.
Data Sources and Procedures for Estimating Two-Year Costs
Direct health-care costs for the first two years after the injury were
calculated as the sum of costs associated with (1) the initial
hospitalization, (2) all rehospitalizations for acute care related to the limb
injury, (3) inpatient rehabilitation, (4) outpatient doctor visits, (5)
outpatient physical and occupational therapy, and (6) prosthetic devices and
related services. In order to obtain dollar amounts that more accurately
reflected the opportunity costs of care, billed charges were adjusted by
Medicare cost-to-charge ratios (for hospital charges and prosthetic devices)
or by insurer or payer reimbursement rates (for billed provider fees). Costs
of inpatient rehabilitation stays and follow-up outpatient services were based
on average reimbursed amounts, rather than billed amounts. All dollar figures
were inflated to constant 2002 dollars with use of the medical service
Consumer Price Index for the year in which the admission or cost occurred.
Consumer Price Index adjustments were obtained from United States Department
of Labor, Bureau of Labor
Statistics11.
Details regarding the estimation of individual cost components follow.
Hospital Costs
Data on billed inpatient charges associated with the patient's initial
hospitalization as well as all limb-related acute rehospitalizations were
collected directly from the hospitals where patients received treatment. These
data were available for 513 (94.1%) of the 545 index hospitalizations and 187
(64.5%) of the 290 patients with one or more rehospitalizations. When actual
charge data were not available, they were imputed with use of coefficients
obtained from generalized linear regressions of the log of total charges (for
patients with complete charge data) as a function of several variables known
for all patients, including length of hospital stay, amputation status, Injury
Severity Score
(ISS)12,
Abbreviated Injury Scale (AIS) score of the contralateral limb
injury13, mechanism
of injury (motorcycle, pedestrian, fall, or other blunt and penetrating), type
of health insurance, and several variables describing the type and extent of
injury to the bone and surrounding soft tissues. Estimates of inpatient
hospital costs were obtained by multiplying billed charges by the Medicare
cost-to-charge ratio from each hospital's Medicare Cost Report for the year in
which costs were incurred.
Physician Fees Associated with Hospitalizations
Billed professional fees associated with the index hospitalization and
rehospitalizations were determined from three of the eight participating
sites, which together accounted for 196 (36.0%) of all patients in the study.
Professional fees at all other sites were imputed by multiplying individuals'
hospital charges by the mean ratio of billed fees to hospital charges
calculated from reporting sites, stratified by amputation status and ISS (a
score of <17 points or a score of =17 points) at the index admission and
by amputation status for rehospitalizations. To more accurately reflect
opportunity costs, billed fees were further adjusted to reflect each site's
average payer-specific fee recovery rate.
Inpatient Rehabilitation
The costs associated with inpatient rehabilitation were estimated by
multiplying the patient-reported length of stay for rehabilitation by the
average per diem cost for rehabilitation. The average per diem cost was the
average of the amounts reported (for patients with a lower-limb injury but no
head or spine injury) by the rehabilitation facilities that were used most
frequently by LEAP sites. Per diem costs were reported in 2000 dollars and
inflated to 2002 dollars as described above.
Outpatient Visits
Costs of outpatient visits were determined by multiplying the average per
visit cost by the number of outpatient medical, physical, or occupational
therapy visits reported by the patient in follow-up interviews. Average per
visit costs for each type of visit were calculated with use of data from the
Medstat MarketScan, a proprietary database containing health insurance claims
data for approximately 4 million insured persons. Estimates were derived for
patients in the MarketScan database who had comparable injuries and were based
on average reimbursed amounts, rather than billed amounts.
Prosthetic Charges
The information regarding the prosthetic charges was obtained directly from
the prosthetists for 134 (77.0%) of the 174 patients who had undergone an
amputation at some time during the twenty-four months after the injury. These
charges included those associated with the purchase of devices as well as the
labor and related hardware needed to maintain functional use of the devices.
For patients for whom charge data were not available, estimates were imputed
with use of coefficients obtained from generalized linear regressions of
logged charges (for patients with complete charge data) as a function of the
level of amputation, age and gender, ISS, AIS score of the contralateral limb
injury, type of health insurance, and number of prosthesis-related visits
reported by the patient as part of the follow-up interview. Average billed
prosthetic charges were estimated for each of four amputation categories
(partial foot, below-the-knee, through-the-knee, and above-the-knee). The
Medicare Part-B fee schedule for each of these five categories in 2002 dollars
was then divided by average billed charges for each category to obtain a
cost-to-charge ratio that was used to adjust billed charges to reflect costs.
These cost-to-charge ratios were 0.46, 0.50, 0.87, and 0.90 for partial foot,
below-the-knee, through-the-knee, and above-the-knee prostheses,
respectively.
Data Sources and Procedures for Projecting Lifetime Costs
To project lifetime costs beyond two years, we multiplied the number of
patient-specific expected life years based on gender and age (at two years
after the injury) by an estimate of future annual health-care costs associated
with hospitalizations and total ambulatory care visits, including those for
both general medical care and physical or occupational therapy. Average life
expectancy by age and gender was estimated on the basis of the United States
Life Tables for
200314. To these
costs, we added an estimate of future costs associated with the purchase and
maintenance of new prosthetic devices for amputees.
Projected Health-Care Costs
Projected annual costs associated with hospitalizations and total
ambulatory care visits were derived by multiplying mean utilization reported
by patients in their seventh year after the injury by per diem or per visit
costs as described above. This approach assumes that the average annual number
of hospital days and number of ambulatory care visits after two years until
death are constant and approximate utilization reported in the seventh year
after the injury.
Projected Prosthetic Costs
Projected costs for the use of prostheses were derived by multiplying the
average cost of the device worn by the patient at seven years after the injury
by the frequency with which a new prosthesis was purchased within two to seven
years after the injury. The average cost of the device worn at seven years
after the injury and the frequency with which new devices were purchased were
estimated from information obtained at the seven-year interview. To account
for costs associated with routine supplies and repairs, we added an annual
amount equal to 20% of the cost of the prosthesis. This percentage was based
on the expert opinion of a prosthetist with experience managing patients who
had undergone trauma-related amputation and projecting the lifetime costs of
prosthetic care for a variety of clients, including case managers and
lawyers15.
Total lifetime costs were calculated by adding the two-year costs to the
projected lifetime costs beginning in the third year after the injury and
occurring until death. Recognizing the importance of our estimates of
projected prosthetic costs in the comparisons between patients treated with
reconstruction and those treated with amputation, alternative estimates of
total lifetime health-care costs were derived by varying both the frequency
with which prosthetic limbs were assumed to be replaced and the cost of
supplies and repairs.
Data Summary and Analysis
Mean health-care-utilization and associated costs were summarized for two
groups of patients defined by their treatment status at three months after the
injury. The first group consisted of 384 patients who had undergone
reconstruction and had an intact lower limb at three months after the injury.
Included in this group were six patients who had undergone an amputation
between three and six months, seven who had undergone an amputation between
seven and twenty-four months, and two who had undergone an amputation between
twenty-four and eighty-four months. It should be noted that, when calculating
both the two-year and lifetime costs for the reconstruction group, we included
the costs for the subsequent amputation for these fifteen patients. The second
group consisted of 161 patients who had undergone amputation within three
months after the injury, including twelve patients who had been initially
discharged with the lower limb intact but were subsequently rehospitalized for
an amputation. We further classified patients treated with reconstruction by
the type of injury (fracture of the tibial shaft, articular fracture of the
tibia, foot fracture, and soft-tissue injury). Patients treated with an
amputation were classified according to the level of amputation.
Multivariate regression techniques were used to determine whether observed
differences in costs persisted after adjustment for the type and extent of the
lower-limb injury (including type and extent of osseous damage, extent of
soft-tissue injury, initial pulse assessment, and plantar sensation), AIS
score of the contralateral lower-limb injury, the presence of injuries to
multiple body regions (as defined by an ISS of =17 points), and selected
patient demographics (age, gender, preexisting medical conditions, and
insurance coverage at the time of the injury). To specifically examine the
impact of associated injuries and limb-related complications on treatment
costs, total costs are separately reported according to the Injury Severity
Score (ISS) and any rehospitalizations for one or more of the following
conditions: late amputation or stump revision, fracture nonunion, hardware
failure, flap loss, wound infection, or osteomyelitis.
Two-Year Costs
The LEAP patients stayed in the hospital for an average (and standard
deviation) of 17.8 ± 12.8 days (range, two to eighty-five days) for
initial treatment of the injuries (Table
I). During the index admission, the LEAP patients underwent an
average of 2.1 ± 0.83 surgical procedures (range, one to seven
procedures).
Although there was no difference in the average length of the hospital stay
or the number of surgical procedures between patients treated with amputation
and those treated with reconstruction, there was variation according to the
type of injury (among patients treated with reconstruction) and the level of
amputation (among patients treated with amputation). This variation is
reflected in the costs related to the index hospitalization (including
professional fees) (Table II).
Initial treatment costs were highest for through-the-knee amputations
($81,086), above-the-knee amputations ($66,912), and tibial shaft
reconstructions ($63,389) and lowest for partial foot amputations ($30,493)
and foot fractures ($44,550).
While rehospitalization for acute treatment related to the injuries was
more frequent for patients treated with reconstruction (60.7% were
rehospitalized) than for patients treated with amputation (35.4% were
rehospitalized), patients treated with amputation were more likely to be
admitted to a rehabilitation hospital or a rehabilitation unit within an
acute-care hospital (18.0% were admitted compared with 10.4% of patients
treated with reconstruction) (Table
I). Of particular note are the high rates of rehospitalization for
acute care among the patients treated with a tibial reconstruction (72.4% of
those with a fracture of the tibial shaft and 62.9% of those with a tibial
articular fracture) as well as the high rate of inpatient rehabilitation among
those who had undergone an above-the-knee amputation (38.7%). The mean numbers
of both outpatient doctor visits and visits for physical and occupational
therapy were slightly higher for patients treated with amputation than for
those treated with reconstruction. Like inpatient service use, outpatient
service use varied somewhat according to the type of injury and the level of
amputation.
When costs associated with rehospitalizations and post-acute care were
added to the cost of the initial hospitalization, the total costs for
reconstruction and amputation were similar ($81,091 and $78,221, respectively)
(Table II). When
prosthesis-related costs were added ($225 for reconstructions and $12,885 for
amputations), the difference in health-care costs between the two groups of
patients ($81,316 for patients treated with reconstruction and $91,106 for
patients treated with amputation) increased. These differences persisted after
we controlled for injury severity and patient demographics in a multivariate
regression analysis and were significant with and without adjustment (p <
0.05). Average total two-year healthcare costs were highest for those with a
through-the-knee amputation ($112,811) or an above-the-knee amputation
($110,039) and lowest for those with a partial foot amputation ($50,153). It
should be noted that the inclusion of the twelve patients who had an
amputation within three months after the injury because the reconstruction
failed or as a result of delayed decision-making did not significantly
increase the average health-care costs for the amputation group. The average
cost for the entire amputation group was $91,106
(Table II); the average
excluding the twelve patients was $90,460.
Rehospitalizations for limb-related complications added substantially to
the two-year health-care costs for the patients who had undergone
reconstruction. The costs for those with complications were 46% higher than
the costs for those without complications
(Table III). When the costs for
the patients who were not rehospitalized because of complications were
compared ($88,010 for the amputation group and $66,987 for the reconstruction
group), the difference between the reconstruction and amputation groups
increased. Overall, rehospitalizations accounted for 15% of the total two-year
costs for the patients treated with reconstruction compared with only 5% for
the amputees.
Health-care costs were substantially higher for patients with injuries to
multiple body systems, as reflected by an ISS of =17 points ($115,887
compared with $75,359 for those with an ISS of <17 points,
Table III). However, the
relative difference in the mean total two-year costs between the patients
treated with reconstruction and those treated with amputation was fairly
similar after stratification by trauma to other body systems. (The costs for
those with an ISS of 17 points was approximately 55% higher than that for the
patients with an ISS of <17 points.)
Projected Lifetime Costs
The total lifetime health-care cost projected for patients undergoing
amputation ($509,275) was more than three times higher than that for patients
undergoing reconstruction ($163,282). This difference can be ascribed largely
to the additional costs associated with the purchase and maintenance of new
prosthetic devices. On the average, the LEAP patients reported purchasing a
new prosthesis every 2.3 years, at an average cost (in 2002 dollars) of $7784
for a below-the-knee prosthesis, $16,028 for a through-the-knee prosthesis,
and $18,722 for an above-the-knee prosthesis. They also reported visiting
their prosthetist an average of four to five times per year.
Limb-related rehospitalizations in the seven years after the injury were
similar between the patients treated with amputation and those treated with
reconstruction. Approximately 3.9% of all LEAP patients reported one or more
days in the hospital per year, with an average length of stay of 5.8 days for
admitted patients. However, amputees reported a somewhat higher number of
total outpatient visits (4.5 compared with 3.7 for the patients treated with
reconstruction).
To examine the robustness of our estimates of lifetime costs, we derived
alternative estimates by decreasing the frequency of prosthetic replacement
(after the seventh post-injury year) from once every 2.3 years to once every
five years. This change resulted in projected costs of $345,456 and $157,553
for patients treated with amputation and reconstruction, respectively.
Although this change in assumption decreased the difference between the
groups, total lifetime costs for patients treated with amputation were still
more than two times higher than those for patients treated with
reconstruction. We further decreased the estimate of annual costs associated
with routine supplies and repairs from 20% to 10% of the cost of the
prosthesis. This adjustment resulted in projections of $331,159 and $157,072
for the patients treated with amputation and reconstruction, respectively.
The results of this analysis show comparable two-year health-care costs
associated with amputation and reconstruction following severe lower-extremity
injury. However, total lifetime health-care costs were substantially higher
for patients treated with amputation than for patients treated with
reconstruction. This was largely due to the costs associated with the repair
and replacement of prostheses for the forty to forty-five years of remaining
life years in this sample.
Our estimates of both two-year and projected lifetime health-care costs
were subject to bias given the limitations of the data and the assumptions
required to derive the estimates. First, we did not include all direct costs
incurred by either the patients treated with amputation or those treated with
reconstruction. Most notably absent were pharmacy-related costs, costs
associated with adaptive equipment, and those associated with modifications of
the patient's residence, motor vehicle, or workplace. All of these are likely
to increase estimated costs significantly. It is also important to emphasize
that our estimates did not take into account indirect costs associated with
lost work and residual disability. As previously reported, however, rates of
return to work and self-reported disability at two and seven years after the
injury were comparable for the amputation and reconstruction
groups2.
A second limitation relates to the data available for estimating two-year
costs. Although the costs associated with the initial hospitalization and
injury-related rehospitalizations were based on actual billed inpatient
charges, the costs associated with inpatient rehabilitation and outpatient
services relied on self-reported use and average per diem or per visit costs
estimated from other sources of data. Although the reliability of
self-reported data on outpatient visits has been shown to be adequate at
recall intervals of two to four weeks, it is more questionable after longer
recall periods16.
We suspect that recall bias results in underreporting of use of services,
leading to underestimates of the true costs. There is no reason to believe,
however, that the magnitude or direction of the bias would be different
between patients who had undergone amputation and those who had undergone
reconstruction.
Our estimates are also limited by missing data. Although data for
estimating hospital costs for the index hospitalization were available for 94%
of all patients, similar data for acute rehospitalizations were missing for
64.5% of the patients. Most problematic were data on professional fees.
Physician fees associated with the index hospitalization and
rehospitalizations were available for only one-third and one-quarter of the
patients, respectively. Finally, charge data were available from the
prosthetists for only 77% of the amputees. However, the multivariate models
that we used to impute values for missing data were robust. Furthermore, the
imputation methods employed were conservative, tending to shrink estimates
toward the mean of the distributions from which they were estimated.
Consequently, while these estimates should be viewed with some caution, they
should also be viewed as conservative.
Projecting total lifetime costs required many assumptions, most of which
are not testable because of the paucity of data on long-term consequences of
major limb injuries. First, we assumed that life expectancy is similar between
patients treated with an amputation and those treated with a reconstruction
and is comparable with that of the general population. In addition, average
costs per rehospitalization, per visit, and per device were trended forward on
the basis of the assumption that service use and device replacement during the
seventh year after the injury were similar to those over each patient's
lifetime. Abandonment of devices and increased use of wheelchairs are common
among the elderly, but those possibilities were not taken into account in the
derivation of our projections. We also did not account for future health-care
costs associated with the onset of long-term complications and secondary
conditions such as arthritis and chronic pain, which may affect both groups,
or the development of neuromas, loss of myodesis, and heterotopic bone, which
may lead to higher health-care utilization and revisions among the amputees.
It is interesting to note that the average number of outpatient visits was
somewhat higher for the amputees than for the patients treated with limb
salvage. These additional visits may be related to the treatment of skin
sores, infections, and cysts, which are common among amputees.
We also did not account for the purchase of multiple devices, which is
becoming more common among young, active, and high-end users. Nor did we
account for the increase in the cost of devices due to technological advances
in both materials and mechanical functioning. Finally, it is important to note
that, because the LEAP study is observational, it is possible that the
patients who underwent amputation may have been more severely injured,
requiring more complex treatment. If those patients had undergone
reconstruction, their overall health-care costs might have been greater.
Although we attempted to adjust for differences in the severity of both the
lower-limb and the associated injuries in the multivariate regression
analysis, the possibility of this bias in our results cannot be
eliminated.
Despite these limitations, we believe that our estimates add support to our
previous conclusions that efforts to improve the rate of successful
reconstructions have
merit1-3.
Not only is the reconstruction of a severely injured extremity below the
distal part of the femur a reasonable goal at an experienced level-I trauma
center, it appears to result in lower lifetime health-care costs. At the very
least, it is clear that direct costs per se should not be used to deny a
patient the opportunity for limb reconstruction. ?