Hip fractures are a burden to both society and the individual.
In 1990, the estimated total number of hip fractures in persons
who were fifty years of age or older was 300,000 in the United States1 and 1,700,000 worldwide2. Furthermore, it has been estimated
that the total number of hip fractures in persons who are fifty years
of age or older will increase to 512,000 by 2040 in the United States3 and to 6,300,000 by 2050 worldwide2. This increase can be explained primarily
by the aging of the population, and it is based on the assumption
that no major changes will occur in the health of elderly people
or in the treatment or prevention of hip fractures4,5.
Hip fractures in elderly individuals cause substantial mortality
and morbidity6-8. In fact, the
mortality rate for patients at one year after a hip fracture is
approximately 15% higher than that for nonaffected people
of similar age and gender9. Moreover,
most hip-fracture patients fail to regain their prefracture level
of activities of daily living, and up to 20% of hip-fracture
patients need to be institutionalized because of the fracture10.
Recently, more attention has been placed on the economic consequences
of hip fracture11-15. Originally,
operative procedures for hip fractures were targeted for cost control
because of their high cost and the increasing rate at which they
were being performed. Consequently, in most of the earlier studies
on the economic cost of hip fracture, cost estimations were limited
to the short-term costs16-27.
Those studies indicated that the major element contributing to the
cost of hip-fracture surgery was the length of hospital stay16,17,22,24,28. Therefore, prior attempts
to reduce the economic impact of surgery have focused on shortening
the initial hospital stay13,19,22,29-32.
These savings, however, do not necessarily reflect a cost-savings
to society30,33-36. An early transfer
to a rehabilitation center is a method of cost-shifting that can
reduce the cost for the hip fracture at the hospital at which the
operation was performed, but it does not necessarily reduce the
total cost associated with the hip fracture30,37-39.
Thus, the next step in reducing the economic impact of a hip fracture
may involve addressing the long-term costs4,30,40-44.
Unfortunately, very few studies have documented the total cost during
the one-year period following hip fracture4,43,44.
Even fewer studies have evaluated the excess cost—that
is, the additional costs resulting from the fracture4,44. We are not aware of any study
in which a control group was included to assess, with use of a fully prospective
design, the financial costs resulting from the hip fracture after
hospital discharge. In the absence of an appropriate control group,
all costs during the period following hospital discharge are likely
to be attributed to the hip fracture45.
In reality, however, these costs may partially result from pre-existing
or intercurrent comorbid conditions4,44,45.
To address this issue, we conducted a prospective study of elderly
women with a first hip fracture and matched controls to quantify:
(1) the cost of the hospitalization for the initial fracture, (2)
the total economic cost during the one-year period following the
initial hospitalization, and (3) the excess (or additional) cost
due to the fracture during the one-year period following the initial
hospitalization.
The primary objective of this prospective study was to itemize,
value, and sum the costs of a first hip fracture, with the goal
of quantifying its economic burden. The secondary objective was
to enable clinicians, administrators, and health-policy makers to understand
the economic aspects of this major health-policy issue. In a previous
study, we used statistically complex economic models to assess the influence
of selected factors on medical costs during the one-year follow-up
period after hospital discharge46.
That study primarily targeted decision makers in the public and
private sector who have an interest in the study results46. The goal of our present study was
to present a clinician-oriented, detailed cost analysis of elderly women
with a first hip fracture and matched controls, with special emphasis
on the cost utilization over time.
Only the key points are summarized in this section because a
detailed description of the study design, recruitment strategy,
participant characteristics, resource utilization, cost calculations,
and statistical analysis have been previously reported46.
Study Design and Participant Characteristics
A one-year, prospective cohort study with matched pairs was performed.
Patients with a hip fracture who were managed at four Belgian hospitals, between
November 1995 and July 1996, were matched, with respect to age and
residence, with women from the same neighborhood (control subjects).
The patients were women who were at least fifty years old and
had been admitted to the hospital because of a first hip fracture
(a femoral neck or intertrochanteric fracture). A history of a fracture other
than a hip fracture or difficulty in responding to the questionnaire
were not exclusion criteria. Of the 187 eligible women with a first
hip fracture who were identified at hospital admission, seventeen (9%)
(or their relatives) refused to participate.
The controls were chosen from the same municipality and were
matched for residence and age (within five years). A history of
hip fracture (but not a fracture of another bone) was an exclusion
criterion. When the patient lived in a nursing home, the control
subject was randomly chosen from a list of eligible women living
in the same nursing home. When the patient was living in her own
house or in the house of a relative, the control subject was selected
from a house (but not from a nursing home or hospital) located in
the immediate neighborhood. As it is difficult to use population
registers as a source for locating control subjects in Belgium,
the "nearest neighborhood" method was used47-49. Starting from the residence
of the patient, a direction was chosen at random. Next, a search
for a control subject who met the age criterion was conducted in
that direction by means of a door-to-door exploration. Although
the potential control subjects were given ample information about
the study with great care, 50% of them refused to participate.
Such low response rates, however, are common in studies involving
elderly patients50,51.
Resource Utilization
All four hospitals provided the invoices for the initial hospitalization.
These invoices included the hospital costs (emergency room; operating
room; recovery room; intensive-care unit; room and board; nursing
care; laboratory services; medications; and other supplies, such
as blood administration, electrocardiography, and renal dialysis);
the costs of the surgical implant materials; and the costs of physician
services for surgery, anesthesia, diagnostic radiology, medical
management, and physical therapy.
During the one-year period after hospital discharge, the patients
and the control subjects were interviewed once a week by trained
personnel (Health Management Creative, Brussels, Belgium), with use
of structured questionnaires and diaries. Specific questions were
asked concerning the use of the following medical services: outpatient
physician visits, outpatient physical-therapy visits, home visits
by a nurse, hospital admissions, rehabilitation-center admissions,
nursing-home admissions, and one-day clinic admissions (defined
as a stay of less than twenty-four hours in a hospital for a treatment or
an investigation). For each of these medical services, utilization
was recorded according to the number of physician visits per month,
the number of physical-therapy services per month, the number of
home visits by a nurse per month, the number of days in the hospital
per month, the number of days in a rehabilitation center per month,
the number of days in a nursing home per month, and the number of
days in a one-day clinic per month.
Sixty-eight percent of the patients were able to answer the questionnaires
directly or to complete the diary. Thirty-two percent of the patients
were unable to respond to questions properly because of illness
or mental disability. In these situations, the interview and the
completion of the diaries were requested of a "proxy-responder"—that
is, a person in close contact with the patient (generally, a family member).
When information for a patient was given by a proxy-responder, a
proxy-responder was identified for the matched control. At the
end of the one-year follow-up period, the diaries were collected.
Cost Calculations
In Belgium, the costs of these different services are almost
completely covered by the federal health-care system. Therefore,
the official 1996 Belgian Health Insurance Association reimbursement
rates were used. Calculation of all costs was assigned to a cost
center. To estimate the cost of outpatient physician visits, outpatient
physical-therapy visits, and home visits by nurses, we multiplied
the number of visits by the official 1996 Belgian Health Insurance Association
reimbursement rate per visit. To estimate the cost of the other
hospitalizations, the rehabilitation-center admissions, the nursing-home admissions,
and the one-day clinic visits that occurred in the year of follow-up,
we multiplied the number of days by the mean per diem reimbursement
rate for routine inpatient services in hospitals, rehabilitation
centers, nursing homes, and one-day clinics affiliated with the
Belgian Health Insurance Association.
In the present study, we estimated the excess cost of a hip fracture
by calculating the difference between the sum of the costs for the
hip-fracture patients and that for the control subjects.
The results are reported in United States dollars and are based
on the currency exchange rate in August 1998 (US$1 = 0.909
EURO = 36.67 Belgian francs). In Belgium, no value-added
tax is applied to health-care costs.
Statistical Analysis
Differences in the means of continuous data were analyzed with
the Student t test. The unadjusted chi-square test was used to compare
proportions. Survival curves were calculated with use of the life-table
method. For univariate comparison of survival data, the log-rank
test was used. The 95% confidence intervals were calculated
for odds ratios and hazard ratios. All statistical tests were two-sided.
Participant Characteristics
During the period from November 1995 to July 1996, 170 elderly
women with a first hip fracture were enrolled in the study. Before
the hip fracture, sixty-two women (36%) lived in their
own house; fifty-four (32%), with relatives; fifty-three
(31%), in a nursing home; and one, in a hotel. Eleven (6%) died
during the initial hospitalization. The 159 patients who were discharged
alive from the hospital were matched with 159 control subjects.
At the time of recruitment, forty-eight (30%) of the 159 control
subjects resided in a nursing home and 111 did not. The clinical
characteristics of both groups are given in Table I. The mean age
of the patients was 79.3 years, and that of the control subjects
was 78.7 years. During the one-year follow-up period after hospital
discharge, none of the patients or the controls were lost to follow-up.
Of the 159 patients, twenty-one (13%) died within one year
after the fracture. During the same time-period, five controls (3%)
died. The survival curves of both groups are shown in Figure 1.
One-Year Resource Utilization
The mean length of hospital stay for the patients with an acute
hip fracture was twenty-nine days. This stay included not only the
surgical procedure and postoperative observation but also the initial phase
of rehabilitation, which is customary in the Belgian system.
The one-year utilization of health-care services by the patients
and controls was analyzed according to the monthly data, with use
of 159 patients and 159 controls as the denominators (Appendix).
Hence, the results reflect the difference in mortality rates between
the patients and the controls. Outpatient physician visits, outpatient
physical-therapy visits, and rehabilitation-center admissions steadily decreased
after hospital discharge. In contrast, the rates of new hospital
admissions for a health problem other than the hip fracture, home
visits by a nurse, and new nursing-home admissions remained high
for the patients compared with those for the controls. The slight
decrease in the number of nursing-home stays at the end of the one-year
follow-up period was essentially due to patient deaths. We also
observed a shift in resource utilization after hospital discharge.
The total health-care utilization among the 159 patients and the
159 controls during the one-year follow-up period after the initial
hospitalization is shown in Table II.
Costs
The mean cost of the initial hospitalization for the fracture
was $9534 (range, $2703 to $37,406).
The mean costs of the medical services during the one-year follow-up
period for the patients and the controls are listed in Table III. The mean
total costs in the year after hospital discharge were $13,470
for the patients and $6170 for the controls. Thus, the
one-year excess cost after hospital discharge averaged $7300
for the hip-fracture patients who survived the acute hospital stay. These
cost calculations incorporated the higher mortality rate of the
women with a hip fracture compared with that of the controls. The
largest cost differences were attributable to nursing-home stays (31%),
rehabilitation-center stays (31%), hospitalizations (16%),
and home physical-therapy services (14%). The cost difference
for the patients who lived in the community before the hip fracture
was $8481 compared with $3827 for those who resided in
a nursing home (p < 0.001).
The monthly distribution of excess costs demonstrated that two-fifths
of the excess costs were spent during the three months following
hospital discharge (Appendix). At the seventh month, excess costs
seemed to level off. With use of the data on excess costs from the
seventh to the twelfth month, we estimated (extrapolated) the mean
excess cost during the second year after the hip fracture at $3514
for the women who survived the acute hospital stay.
To our knowledge, we are the first to report cost estimates based
on data obtained prospectively from hip-fracture patients and from
matched control subjects. Overall, the costs associated with the treatment
of hip-fracture patients were about three times greater than those
resulting from the treatment of age and residence-matched controls
without a fracture. Two-fifths of these excess costs were incurred
during the first three months following hospital discharge.
The strengths of our study are its prospective, controlled design;
complete patient follow-up; and detailed methods used for direct-cost
calculations. The study has several potential limitations, some
of which have been previously discussed in more detail46. Hip fractures occur more commonly
among the frailest women who receive more medical or social care
than the average woman52,53. In
our study, the high proportion of patients who lived in a nursing
home reflects the high prevalence of institutionalization of elderly
people who are in poor medical condition. In that respect, control
subjects who were closely matched with respect to both age and residence
had the best chance to mirror the health condition of the patients.
Unfortunately, the refusal rate among potential control subjects
in our study was high. This high refusal rate was due in part to
the necessity for controls to accept a one-year follow-up period
with regular interviews and to record all health-related events
in a diary. Hence, there was a certain degree of selection bias
toward the inclusion of control subjects who were more motivated
and thus probably healthier than the average45.
To correct the cost estimates for the difference in the degree of
frailty between patients and control subjects, we calculated a "fragility
component ratio," with use of the data summarized in Table I. The proportion
of women without a neuropsychiatric disorder in the patient group
divided by that in the control group ([100 — 12]/[100 — 3])
yielded a fragility component ratio of 0.91. The same calculation
performed with regard to the proportion of women who were not treated
for a chronic disease ([100 — 51]/[100 — 47])
yielded a fragility component ratio of 0.92. The fragility component
ratios suggest that the costs in the control group could have been
up to 9% higher if the proportion of women with a neuropsychiatric
disorder or of those treated for a chronic disease had been equivalent
in the two groups. With use of the data presented in Tables II and III, the one-year cost
after hospital discharge corrected for the fragility component can
be calculated as $13,470 — ($6170 ¥ 1.09) = $6745.
The corrected one-year total excess cost per woman with a hip fracture
then would be $16,279 instead of $16,834. Another
more sophisticated tool to adjust further for these differences
in the clinical characteristics of our study groups is economic
modeling. To estimate the simultaneous effects of important comorbidities,
we performed separate least-squares multivariate regression analyses46. Although these mathematical methods
may allow more precise control for the differences in the characteristics
of the study groups, the economic models yielded results similar
to those calculated with use of the fragility component ratio46.
We confined our report to estimates of direct costs
of health-related services rather than charges4,45,54.
These direct costs must be distinguished from less readily available
costs of formal nonmedical care and informal care provided by family
and friends. In our prospective study we did not include the costs
associated with ambulance services, outpatient medications, home
services (assistance with housekeeping, personal care, and outdoor
and indoor mobility), emotional support, help arranging for services,
and special types of transportation. Nonetheless, the diaries of
our study groups suggested that these costs were less important
than the costs of outpatient physician visits or home visits by
nurses, as they represented less than 1% of the documented
direct costs. Similarly, according to Borgquist et al., outpatient
costs in primary health care represent a relatively small part of
the total cost of hip fracture, amounting to only about 1% of the
total short-term cost within four months after the fracture40. In addition, we did not consider
the productivity losses, such as changing or giving up a job, that
are incurred by the patient, her family, and her friends4,54. The fact that only six hip-fracture
patients (4%) in our study were in the fifty to sixty-year-old
age-group suggests that these costs may be of minor importance.
Thus, we did not include indirect and intangible costs, and in this
respect the calculated costs in our report probably underestimate
the total costs. The purpose of the current study was to adopt a
societal point of view. Whenever possible, we took the public prices
used by the Belgian Health Insurance Association as the basis for
establishing the reimbursements for medical or social care services,
in order to have costs that reflected as much as possible the real
costs and not the charges.
The vast majority of studies on the costs of hip fractures have
focused on the cost of the initial hospitalization1,4,16-28,30,40-42,44,46,55-64. The
reported costs of the initial hospitalization have varied substantially
among different countries. The mean cost of care during the initial
hospitalization of hip-fracture patients is about $7000,
with the lowest cost reported for Norway ($739)27 and the highest cost reported for
Switzerland ($44,000)26.
Our estimate of the cost of the initial hospitalization ($9534)
is well within the range of those studies26,27.
In addition, our estimate of the cost of the initial hospitalization
was remarkably close to that reported in another study from Belgium
($8977)65 and in a recent
one from the United States ($11,480)4.
Nevertheless, a comparison of the costs reported in those countries
is hampered by differences in study design, failure to adjust for
exchange rates, and variation in clinical care.
The reported long-term costs associated with hip fractures also
vary widely. Comparison of long-term costs is limited by the variability
in the follow-up periods, which have ranged from four months4,19,30,40-42,55-64 to one year1,4,43,44,46,55,57,58,61,62 after the
initial hospitalization (Appendix). Studies on the long-term costs
associated with hip fractures have relied on national survey data1,55,57,58,61,62, cross-sectional designs30,41-43, and prospective designs4,40,44,46. Even the latter studies,
however, have not always clarified the true impact of a hip fracture
on the cost of care28. Older persons
who are at risk for a hip fracture often have comorbidity and functional
impairment, resulting in an increased need for medical and nonmedical
services irrespective of the fracture4,44.
Therefore, it is unclear whether the cost estimates from these studies
represent an increase in resource utilization attributable to the
hip fracture28.
The most reliable method to determine the true financial impact
of hip fractures is to assess the excess cost28.
This is done most effectively by following a cohort of hip-fracture
patients prospectively and evaluating all of the additional costs
incurred after the fracture during a period of at least one year
following hospitalization28. To
our knowledge, the excess cost of a hip fracture has been assessed
with use of a control group in only two studies, one from the United
States4 and one from Sweden44. In those studies, the excess one-year
cost incurred by one hip fracture was estimated to be $18,727
and $17,704, respectively (Appendix). In both studies, the
hip-fracture patients were used as their own controls; this involved
a comparison of the costs incurred in the year after the fracture
with those incurred before the fracture4,44.
The authors of those studies were dependent on the recall of the
patients and their families to identify the costs prior to the fracture.
Excess cost was calculated as the cost incurred one year after the fracture
(data gathered prospectively) minus the cost incurred one year before
the fracture (data gathered retrospectively)4,44.
In contrast, the current investigation was designed as a cohort
study, with use of a matched-pair analysis. Both patients and controls
were followed prospectively. Interestingly, however, the excess
costs incurred during the year after the fracture were very similar
in these three studies (Appendix), despite differences in study
design, patient selection, country of study, and cost of the items
considered46. The results of these
three studies reflect a striking variation in the utilization of
resources, suggesting that financial considerations were influenced
by the availability of care. In studies from the United States,
hip-fracture patients stayed in the hospital for a much shorter
period of time unless they had a complication15,29,66.
In a study from continental Europe, patients who stayed in the hospital
for the longest time were awaiting transfer to a nursing home67. After hospital discharge, services
provided to the patients in the American study included formal nonmedical
care and informal care provided by family and friends. These types
of services were rarely provided to the Swedish and Belgian patients.
During the study period, the United States insurance system covered
mainly the direct cost of services to inpatients. In the Belgian
and Swedish studies, the much higher volume of outpatient physical-therapy
visits, home visits by nurses, rehabilitation-center admissions,
and nursing-home admissions contributed to a striking difference
in the utilization of resources. In Belgium and Sweden, these services
have become a common component in the care of many hip-fracture
patients, and, perhaps more important, are almost completely covered
by the health-care system funded by the government. Although the
comparative analysis of these three studies indicated a striking
variability in the utilization of health-care services, their overall aggregate
economic consequences had a comparable weight. Thus, regardless
of the system of health care, there are obvious and universally
similar costs associated with hip-fracture treatment, which are probably
related to the financial resources that each industrialized Western
country is prepared to spend on health care.
The results of our study confirm that the economic burden associated
with hip fracture is substantial13-15,38,
highlighting the need for strategies to prevent this type of fracture.
Such strategies are urgently required, given the fact that, in the
United States alone, hip fracture occurs in an estimated 300,000 persons
each year, with an estimated cost of more than seven billion dollars4. These figures are expected to at
least double by the year 20404.
In the immediate future, however, efforts at cost-savings will
have to focus on rehabilitation after the fracture, reducing the
need for nursing-home care. In our study, the excess cost related
to hip fracture increased dramatically for women who were not institutionalized
before the hip fracture, essentially because of new admissions to
nursing homes. The lower cost difference among women who already resided
in a nursing home was partly due to a more limited utilization of
physical-therapy visits or rehabilitation-center admissions. As
health care moves further into an area of cost containment, health-care
payers will continute to demand evidence of cost containment from
health-care providers33,35,36,39.
While the current study cannot provide definite answers to the questions
of cost containment, it provides evidence about the need to reduce
the number of patients who require institutionalization. Obviously,
appropriate methods for controlling the cost of hip fractures will
have to be developed, but these reductions in cost must be achieved
without lowering the standards of patient care6,8,14,29,31-33,38,66.
As it is the responsibility of the surgeon to ensure that each patient
receives state-of-the-art care, not only in his or her institution
but also after hospital discharge, it remains of utmost importance
that orthopaedic surgeons take the lead in the implementation of
such modalities. We hope that cost calculations such as those presented
in the current study might serve as a resource to catalyze future efforts
to reduce the need for institutionalization after hip fracture.
Note: The Belgian Hip Fracture Study Group consists
of the following investigators: P. Autier, MD (Division of Epidemiology
and Biostatistics, European Institute of Oncology, Milan, Italy),
J.M. Baillon, MD (Department of Orthopedics, Ixelles-Etterbeek Hospital,
Brussels, Belgium), M. Barette, MD (Unit of Epidemiology and Prevention
of Cancer, Jules Bordet Institute, Brussels, Belgium), J. Bentin,
MD (Service of Rheumatology, Louis Caty Hospital, Baudour, Belgium),
S. Boonen, MD, PhD (Leuven University Center for Metabolic Bone
Diseases and Division of Geriatric Medicine, Katholieke Universiteit
Leuven, Leuven, Belgium), R. Bouillon, MD, PhD (Leuven University
Center for Metabolic Bone Diseases and Division of Endocrinology,
Katholieke Universiteit Leuven, Leuven, Belgium), P. Broos, MD,
PhD (Leuven University Center for Metabolic Bone Diseases and Division of
Traumatology and Emergency Surgery, Katholieke Universiteit Leuven,
Leuven, Belgium), M.C. Closon, PhD (Interdisciplinary Center in
Health Economics, Université Catholique de Louvain, Brussels,
Belgium), A.R. Grivegnée, PhD (Unit of Epidemiology and
Prevention of Cancer, Jules Bordet Institute, Brussels, Belgium),
P. Haentjens, MD, PhD (Department of Orthopaedics and Traumatology,
Vrije Universiteit Brussel, Brussels, Belgium), and P. Opdecam,
MD, PhD (Department of Orthopaedics and Traumatology, Vrije Universiteit
Brussel, Brussels, Belgium).