| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Department of Pathology and Laboratory Medicine, University of Pennsylvania, 3400 Spruce St., Philadelphia, PA 19104-4283.
a Author for correspondence. Fax 215-349-5090; e-mail donaldyo{at}mail.med.upenn.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResults2DiscussionReferences |
|---|
Methods: Hospital costs were analyzed for 486 diagnosis-related groups (DRGs) relating to >1.3 million patient discharges from 60 University Hospital members of the University HealthSystems Consortium. For each DRG, length of stay, total cost, and key cost components were analyzed, including accommodation, intensive care, and surgery.
Results: In general, total costs of diseases classified as surgical exceeded those classified as medical. Diseases involving organ transplantation typically cost more than other diseases. However, within the studied population, the two DRGs accounting for most total healthcare dollars were percutaneous cardiovascular procedures and management of neonates with immaturity or respiratory failure.
Conclusions: Considering six key cost components, as well as disease complexity and length of stay, the best predictors of total costs for medical conditions were the length of stay and accommodation (housing, meals, nursing services) costs, whereas for surgical conditions, the best predictor of total costs was laboratory costs. This analysis may be used within an individual institution to identify surgical or medical diagnoses with total or component costs at variance with the group mean. A hospital may focus its cost reduction efforts to make decisions to expand, alter, or eliminate particular clinical programs based on comparison of its own total and component costs with those from other hospitals in the database.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults2DiscussionReferences |
|---|
The objective of this study was to determine the typical costs of managing different diseases and to identify the component sources of these costs. Since 1989, the University HealthSystems Consortium (UHC)1 has collected all-payor hospital discharge data from many of its academic medical center members. This resource, the UHC Clinical Data Base, contains information on length of stay (LOS), costs, and clinical outcomes such as inpatient mortality and complications rates. Cost estimates of inpatient care are available for every discharge, and this information can be aggregated on multiple levels, including diagnosis-related groups (DRGs). For each hospital, the total costs have been broken down by component sources. The database enables each hospital within the UHC to benchmark its costs for individual services for a given disease against those of its peers. Such an approach allows an institution to identify component costs for a specific service that deviate markedly from those in comparable institutions, thus enabling focused efforts to eliminate inappropriate resource utilization or to create efficiency. The database provides the background information to enable a hospital to make rational decisions as to whether it should embark on new programs or close or modify existing programs. The UHC database provides information on both the median and mean costs for any component service. In the absence of more detailed information on the variability within the service, we have compared mean costs, which are greatly influenced by unusual situations, with the more typical median costs to provide a measure of the heterogeneity of these costs.
The database highlights those diseases that have major ramifications on each of the primary cost centers in a hospital. Reports of charges on all patients discharged, irrespective of payor, were developed from the Health Care Finance Administration (HCFA) Prospective Payment System Uniform Billing Code of 1992 (UB-92) reports of charges submitted to Medicare by each hospital. The data thus provide a largely uniform pool of information, permitting the component expenses of one disease in one hospital to be compared with the same components of the same disease in other hospitals. The data permit an entry point into the examination of resource utilization in acute care hospitals. Because of the size of the database, this analysis is confined to an assessment of the costs of the major components of the different diseases. The DRGs associated with the highest costs for each of the major components or services are identified. Key cost components, LOS, and disease complexity are analyzed to determine the best predictors of total costs.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults2DiscussionReferences |
|---|
The database comprises not only the total costs for all patients
within each DRG, but also costs broken out by certain key services.
Those cost components that we have selected for analysis are listed in
Table 1
. Additionally, we have compared the 1995 Medicare reported
average LOS for patients with the same diseases to determine whether
the LOS for the UHC teaching hospitals differed materially from the
national norm for patients in both community and university hospitals.
|
Selected data from the UHC database were entered into a FileMaker Pro database, Ver. 4.0 (Claris Corporation), which was used to calculate measures such as medians and means of costs across DRGs and costs per day. Statistical correlations were performed using GraphPad Prism, Ver. 2.0 (GraphPad Software).
calculation of costs
The costs used in the calculations in this study were
derived by converting the charges billed to patients for different
services within each hospital to the hospital’s costs of providing
these services by applying the ratios of costs to charges. These
ratios are calculated from the detailed HCFA Prospective Payment System
cost reports supplied by each hospital to Medicare. The resulting
calculated costs included those pass-through costs, such as capital and
medical education, allowed by HCFA. The total costs are derived from
the sum of the individual departmental charges within a hospital.
This method has shown a correlation of >0.90 with internal accounting
costs (3). This method of calculation provides consistent
data, but because it is not designed to be a cost accounting system, it
does not measure absolute costs with complete accuracy. Nevertheless,
the consistent methodology and the large number of patients in this
study make this methodology ideal for comparing costs of disease across
institutions.
Elimination of outliers tends to generalize costs and does not identify, for example, differences in costs for room and board among different types of disease, e.g., medical, surgical, and pediatric. All of the hospitals in the database are general acute care hospitals, although many provide services such as obstetrics, oncology, and pediatrics, which are frequently provided in specialty-oriented hospitals. The costs attributable to any specialized services required to manage patients with these conditions are included in the appropriate category, such as surgical or laboratory costs.
Throughout our analysis we have elected to focus on median rather than mean costs because the mean costs were consistently higher than the median costs for all components studied, reflecting the influence of what could be a small number of much more complicated cases within a single DRG. Ratios of mean to median costs were examined to assess heterogeneity among patients in a given DRG. When costs were referenced to LOS, we elected to use both the mean and median costs and LOS within the UHC. Using the mean LOS enables comparisons with Medicare data because Medicare bases its LOS information on mean LOS (arithmetic mean).
Issues related to laboratory, blood, and drug costs are discussed in separate reports, as are those related to the complications of diseases and the influence of age on the consumption of resources.
|
Results2 |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults2DiscussionReferences |
|---|
. Three of the 20 most common medical diagnoses and 3 of the 20
most common surgical diagnoses were pregnancy-related. Two of these
surgical diagnoses were related to Cesarean sections and the third to
sterilization or dilation and curettage. Five of the listed conditions
(three medical and two surgical) were related to newborns (three
medical and two surgical). Among the DRGs with the most patients,
medical conditions typically had more patients than surgical
conditions.
|
los
The median LOS varied from a low of 1 day for many conditions to
33 days for extreme immaturity or respiratory distress in neonates (DRG
386). The diseases with a median LOS >10 days are listed in Table 3
. The mean LOS of 42.6 days for extreme immaturity or
respiratory disease in newborns exceeded the mean LOS of any other
condition by more than 10 days. The mean LOS of patients with heart
transplants exceeded the median by 10 days, and the mean for extreme
immaturity or respiratory distress in neonates exceeded the median by
9.5 days, clearly reflecting great variability among the patients with
the same conditions.
|
The median LOS exceeded the mean LOS for only 11 conditions. These were conditions where the outliers tended to have lower costs than the typical patient. The condition with greatest ratio of median to mean LOS (1.16) was malignant breast disorders with complications, closely followed by normal newborns with a ratio of 1.15. The ratio for vaginal delivery without complicating diagnosis was 1.14. This DRG encompasses all normal deliveries. The ratio above 1.00 indicates that a small number of mothers are discharged earlier than the majority.
Those DRGs whose mean LOS exceeded the median by the greatest amount, i.e., had the lowest ratio of median to mean LOS, are conditions where the outliers tend to have higher lengths of stay than the typical patient. The mean LOS for fractures of the femur (DRG 235) was almost six times the median, and the mean LOS was approximately three times that of the median for three other conditions. The ratio for operating room procedures with diagnoses of other contact with health services was 0.29, and the ratios for full-term neonates with major problems and neonates, died or transferred to another medical center, were 0.30 and 0.34, respectively.
The Medicare-reported mean lengths of stay in all hospitals exceeded
those in the 60 UHC hospitals for all except 50 DRGs. Those conditions
in which the UHC mean length exceeded the Medicare arithmetic mean LOS
by the greatest amount are illustrated in Table 4
. Seven of the listed conditions were among the 20 conditions
with the longest mean LOS in UHC hospitals. These conditions included
diseases of neonates, malignant diseases, and mental disorders. The
mean LOS in UHC hospitals for four DRGs was twice or more that of the
reported Medicare mean. The mean LOS in UHC hospitals of patients with
DRG 385 (neonates, died or transferred) was 6.6 times that reported by
Medicare. The lengths of stay in UHC hospitals of patients with DRG 456
(burns transferred to another facility) and DRG 386 (extreme immaturity
or respiratory distress syndrome in neonates) were 2.5 and 2.4 times
the Medicare reported mean lengths of stay, respectively.
|
In contrast, there were 68 DRGs in which the mean LOS in UHC hospitals
was less than one-half of the Medicare mean for the same condition.
Table 5
lists the 17 conditions for which the Medicare arithmetic mean
LOS exceeded the UHC mean by >5.0 days. All of these DRGs, except
viral meningitis (DRG 021), were among the most complicated conditions
classified by the DRG system. For DRG 454 (other injuries, poisonings
and toxic effects with complications and/or comorbidities; data not
shown), the mean LOS in UHC hospitals of 1.06 days was approximately
one-sixth of the reported Medicare mean of 6.1 days. Patients in UHC
hospitals included in DRG 466 (aftercare without history of malignancy
as secondary diagnosis; data not shown) also stayed for less than
one-third of the mean time of the patients reported by Medicare.
|
overall costs
The DRGs with the lowest and highest ratios of median to mean
costs demonstrated the greatest variability in costs to manage patients
within the same DRG. The lowest ratio of 0.29 was for patients with
burns who were transferred to another acute care facility (DRG 456).
Four of the lowest ratios (0.31, 0.36, 0.48, and 0.53) were related to
newborns: prematurity without major problems, neonates died or
transferred, full-term neonates with major problems, and neonate with
other significant problems, respectively. Each of these DRGs contained
>4000 patients, clearly reflecting the considerable variation in the
amount of care that has to be delivered to these patients classified
within the same DRG. Several of the other DRGs with lowest ratios of
median to mean costs involved malignant diseases, but they also
included conditions in children: cardiac congenital and vascular
disorders, hernia procedures, nutritional and miscellaneous metabolic
disorders, and septicemia. Only for DRG 412 (history of malignancy with
endoscopy) did median costs exceed the mean costs. This DRG included
only nine patients, and the ratio was 1.09. For eight conditions, the
mean and median costs were within 5% of each other.
Among the 20 DRGs with the lowest median-to-mean ratios, 15 were medical conditions and 5 were surgical. Of the 20 DRGs with the highest median-to-mean ratios, 15 were surgical conditions, reflecting a greater degree of homogeneity among these patients. These data suggest a greater variability among medical cases than among surgical cases.
A breakdown of the mean proportion of the total median costs for the
major cost components for both medical and surgical diseases is shown
in Fig. 1
. The costs for surgical diseases typically are much greater
than for medical diseases, as illustrated for the 20 most expensive
medical and 20 most expensive surgical diseases, and shown in Table 6
. Because of the very different lengths of stay of patients with
different conditions, the ranking of diseases by costs per day is quite
different from that by total costs. Thus, the daily costs for lung and
kidney transplants are comparable but greatly exceed those of heart,
bone marrow, and liver transplants. The ranking of diseases with the
highest total mean costs, all more than $30 000, together with their
associated daily costs is illustrated in Table 7
. The only DRG for which the mean cost did not exceed the median
cost was for patients with a history of gastrointestinal malignancy
with endoscopy (DRG 412), which included only nine patients.
|
|
|
When diseases were ranked on the basis of their costs per day, as in
Table 8
, it was apparent that some quite simple procedures, such as
arthroscopy (DRG 232) or sinus and mastoid procedures in children (DRG
054), have daily costs that substantially exceed those of much more
complicated conditions. This may be attributable to the relative
intensity of service during the short lengths of stay of the patients
with the simple procedures. For most of the conditions listed, the
median daily costs exceeded the mean daily costs. Lower mean daily
costs often were associated with longer lengths of stay for patients
within the same DRG.
|
When there were both medical and surgical DRGs within a MDC, the median
costs for surgical conditions were consistently higher than those for
medical conditions with the exception of MDCs 6 (disorders of the eye)
and 22 (burns; Table 9
). For only 2 MDCs, burns and trauma, did the mean of the median
costs of medical MDCs exceed $10 000, but this occurred for 11 of the
surgical MDCs. After burns and trauma, the most expensive medical MDCs
were mental disorders (MDC 19) and diseases of the hepatobiliary system
or pancreas (MDC 7), with mean of median costs of $9150 and $7616,
respectively.
|
For the surgical MDCs, the mean of the median costs of the trauma DRGs was almost twice that of the next most costly MDC, the single DRG associated with infectious and parasitic diseases (MDC 18). The lowest mean of the median cost for a surgical MDC, disorders of the eye, exceeded the average median costs of four medical MDCs. The costs of $3338 exceeded the average median costs of the five medical conditions in MDC 15, normal newborns and the perinatal period, by $2158.
surgical costs
The five disorders involving transplantation had higher median
surgical plus transplant costs than surgical costs alone for any other
disease. These combined costs were $56 110 for transplants of the
heart, $35 416 for liver, $31 686 for lung, $27 186 for kidney, and
$26 363 for bone marrow, but the costs directly attributable to
surgery for bone marrow and kidney transplants were only $2954 and
$4200, respectively (9.9% and 4.7% of the total costs). When
transplant-specific costs, related to the harvesting and processing of
the donated organs, were excluded, the highest median surgical costs
were for heart transplants. The median surgical costs for heart
transplants of $34 270 were more than twice those of the next most
expensive: extensive burns with an operative procedure ($14 367) and
cardiac valve procedures with cardiac catheterization ($12 500). The
identified surgical costs for liver transplants and cardiac valve
procedures without cardiac catheterization also exceeded $10 000.
The surgical costs for DRG 261 (breast procedure for nonmalignancy except biopsy and local excision) comprised 71.1% of the total costs for the condition. For 29 conditions, surgical costs comprised >60.0% of the total costs, and for 75, they comprised more than one-half the total costs. This large proportion is attributable to the typical short LOS for these conditions. Indeed, of the 10 conditions with the highest proportion, 9 had median lengths of stay of 1 day and the other had a median of 2 days. For heart transplants, surgical costs alone accounted for 39.8% of the total costs, and when combined with transplant costs accounted for 65.2%. For kidney transplants, the median surgical costs accounted for only 9.9% of total costs, but when combined with transplant expenses, they accounted for 61.8% of the total.
costs per day
When the identified costs of surgery, including transplant costs,
were subtracted from the total costs, the ranking of costs per day was
quite different from the ranking with all costs included, although
kidney, lung, and liver transplants were still contained in the 10
diseases with the highest daily costs. Although the overall cost of
transplantation of kidneys was substantially less than that of the
other major organs, the associated LOS was shorter than that of other
organ transplants so that the cost per day was high, with or without
the inclusion of all of the surgical and transplant-related costs. The
cost per day for kidney transplants, when surgical costs were excluded,
of $4266 was more than $400 greater than for any other condition.
Many conditions had median daily costs, even with the exclusion of surgical costs, that exceeded $2000 per day. These included $3855 for concussion in adults with complications and/or comorbidities. The median daily cost for managing patients with fractures of the femur, with surgical costs excluded, was $3269. The median daily cost for lymphoma or nonacute leukemia with an operative procedure and/or complications and comorbidity was $2751. Because of the short lengths of stay for the conditions, the per diem median costs for simple conditions such as nasal trauma and deformity ($2397) or percutaneous cardiovascular procedures ($2320) were among the highest.
The lowest median costs per day of the conditions included in the
database are shown in Table 10
. The lowest daily cost was $243 for normal newborns. The next
lowest costs per day of $350 and $395 also were for newborn infants but
with some medical complications, i.e., neonate with other significant
problems (DRG 390) and prematurity without major problems (DRG 388).
The other conditions with lowest costs per day often were for
ophthalmologic conditions or for psychiatric conditions. For both of
these types of disease, the costs per day were largely attributable to
accommodation costs. Whereas patients with eye disorders frequently had
short lengths of stay (overall mean, 2.4; range of means, 2–4 days),
the lengths of stay for most patients with psychiatric disorders were
typically relatively long (overall mean, 7.0; range of means, 3–11
days).
|
aggregated costs
When the aggregated costs of the conditions in the 60 University
hospitals were calculated, the condition with the highest overall cost
was percutaneous cardiovascular procedures at $229 million. This is
attributable to the large number of patients in the database, as
illustrated in Table 11
. The cost of managing the 57 661 normal births, the condition
with the most admissions, was only $131 million. The 5 organ transplant
DRGs were ranked in the 10 most expensive conditions, but the only
condition involving organ transplantation to rank in the top cumulative
cost list was renal transplantation, for which the cost of managing all
the patients in the 60 UHC hospitals was $183 million, incurred by only
4292 patients.
|
accommodation costs
Accommodation costs comprised all nursing costs and the
"hotel" type of services, including food service. The accommodation
costs of nine conditions exceeded 3 SD of the mean of all conditions.
These conditions are listed in Table 12
. The accommodation cost of DRG 472 (extensive burns with an
operative procedure) were more than $6000 higher than for any other
condition. Although all of the transplant DRGs, with the exception of
kidney transplants, were included in this outlier category, the
proportion of total costs attributable to accommodation varied from a
low of 17.8% for liver transplants to a high of 40.3% for bone marrow
transplants. When all DRGs were considered, the average proportion of
total costs of the surgical DRGs attributable to accommodation costs
was 33%, substantially less than the average 48% for medical
diseases. The highest median accommodation cost per day for any
condition was $1544 for fractures of the femur.
|
For surgical conditions in general, the high costs attributable to surgery caused accommodation costs to comprise a smaller proportion of the total costs when compared with medical conditions. For example, surgical costs for breast procedures for nonmalignancy comprised 71.0% of the total costs for the condition. For 26 surgical DRGs, surgical costs comprised >60% of the total, for 70, they comprised >50%, and for 124, and for one-half of all surgical DRGs, they comprised >40%.
Accommodation costs of 15 DRGs constituted >75.0% of the total costs. Several of these conditions were mental disorders, often associated with prolonged lengths of stay and with relatively nonintensive medical care. Two of the conditions with the highest proportional accommodation costs involved drug administration. These comprised 96.0% for alcohol/drug dependence with rehabilitation therapy and 86.2% for poisoning and toxic effects of drugs in adults with complications and/or comorbidity, although the median LOS for the latter was only 2 days. Median accommodation costs could be anticipated to be high for DRGs associated with long lengths of stay. Thus, they were 85.6% of the total costs for the median 10 days of hospitalization for childhood mental disorders and 78.9% for the median 6 days for patients with disorders of personality and impulse control. Four of the conditions involved newborns: 76.8% for the median 13 days of hospitalization for premature newborns with major problems; 76.3% for normal newborns; 82.3% for prematurity without major problems; and 75.3% for neonates with other significant problems.
In contrast, accommodation costs accounted for <20% of total costs for 27 DRGs. Whereas these conditions typically are associated with short median lengths of stay, accommodation costs were only 12.3% of the total costs for kidney transplants, which have a median LOS of 9 days. Accommodation costs as a proportion of total costs of 12.2% were lowest for cardiac pacemaker device replacement. The relatively high costs for surgical procedures for many conditions associated with short lengths of stay meant that accommodation costs as a proportion of total costs were small for these conditions.
intensive care unit costs
For only 33 conditions were there identifiable median intensive
care unit (ICU) costs. The greatest median ICU costs were $19 561 for
extensive burns with an operative procedure (DRG 472), $8937 for
craniotomy with multiple significant trauma (DRG 484), and $8532 for
heart transplants (DRG 103). However, ICU costs comprised 83.9% of all
accommodation costs for unexpired patients with cardiac arrest (DRG
129); 63.3% and 61.7%, respectively, for patients with extensive
burns without (DRG 457) and with an operative procedure (DRG 472); and
54.8% for patients with a craniotomy for multiple significant trauma
(DRG 484).
other services
At 5% of total costs for medical diseases, median radiology costs
were 2.5 times those for surgical patients (2%). Median
laboratory costs as a proportion of total costs were 9% for medical
patients, 50% higher than for surgical patients (6%). Pharmacy costs
comprised 7% of total costs for medical conditions and 8% of total
costs for surgical conditions. The laboratory and drug costs will be
discussed in separate reports. Only 214 conditions incurred median
ancillary service costs. In decreasing order, the median ancillary
service costs for the most expensive conditions were $5841 for
extensive burns with an operative procedure (DRG 472), $2898 for
rehabilitation (DRG 462), and $2789 for craniotomy with multiple
significant trauma (DRG 484). As a proportion of total costs, ancillary
service costs for rehabilitation (25.0%) greatly exceeded those for
the next highest, extensive burns with an operative procedure (7.8%).
complexity
The relative weight of a condition is derived by Medicare from the
average charge for each DRG divided by the national average
standardized charge per case. The relative weight value is intended to
reflect the relative resource consumption associated with each DRG and
thus, presumably, reflects the complexity of each DRG. The higher the
relative weight, the greater is Medicare’s payment to a hospital. The
relative weights of the DRGs ranged from a low of 0.0896 for allergic
reactions in children <18 years to 16.3066 for liver transplantation.
The relative weight for normal newborns (0.1465) was the second lowest.
The highest weight for a medical condition was 4.1529 for chemotherapy
with acute leukemia as a second diagnosis. The lowest relative weight
for a surgical condition was 0.1479 for circumcision in children. The
15 highest relative weights were all for surgical conditions. All five
organ-transplant DRGs, except for kidney transplants with a relative
weight of 4.1370, were among these 15 surgical conditions.
correlations
Table 13
identifies key correlates with total costs for the 486 DRGs in
the database, broken out by medical (Table 13
A) and surgical (Table 13
B) DRGs. For medical conditions, the accommodation costs were the
best predictor of overall costs (r2 =
0.9231) and of LOS (r2 = 0.9553). LOS,
laboratory costs, and drug costs were also good predictors of total
costs for medical conditions. For surgical conditions, the best
predictor of total costs was laboratory costs
(r2 = 0.9578), but all cost
measurements, with the exception of radiology, correlated well with
total costs. Accommodation costs were again the best predictor of LOS
(r2 = 0.9618). For the surgical DRGs,
eliminating the surgical and transplant costs had little effect on the
correlations. For the surgical DRGs, there was a much better
correlation of the relative weight with key components than there was
among medical conditions. Thus, the correlation coefficients of total,
accommodation, laboratory, and drug costs and LOS all exceeded 0.8,
whereas the highest correlation of a cost measurement for medical
conditions with the relative weight was 0.6698 for total costs. There
was a much better correlation of relative weight with LOS for surgical
conditions (r2 = 0.8071) than for
medical conditions (r2 = 0.5990).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults2DiscussionReferences |
|---|
The observation of generally greater costs for surgical DRGs compared with medical DRGs is consistent with other reports using a different approach to evaluate hospital costs (5). Barnett (5) determined that that the mean cost of surgical DRGs of $11 836 per relative weight unit was more than twice that of medical DRGs.
There is considerable difference between the mean and median total costs of diseases. For all of the most costly diseases, the mean cost exceeded the median. For patients with extensive burns without major operating room procedures, the difference was $19 217, or 25.4% of the median cost. This discrepancy presumably reflects the considerable differences in the extent of the burns or severity of the conditions of the patients included in this one DRG. For no other condition was the difference as great, either in terms of percentage or dollars.
We focused our assessments on medians to minimize the influence of a small number of outliers on the apparent costs of a disease, in particular when the number of patients in the category was small. The great majority of conditions with the highest mean-to-median ratios were medical conditions, whereas the majority of those with small differences were surgical conditions. This presumably reflects the greater variability, e.g., severity, acuity or chronicity of the disease, age of the patient, or prior health, inherent in medical conditions than in surgical conditions.
Although Medicare categorizes patients’ diseases by DRGs, this classification forces the inclusion of many different diseases into a single DRG. Sumner and Moreland (6) studied patients with DRG 243 (medical back problems) to highlight the heterogeneity of diseases included within a single DRG. The DRG classification also does not reflect the radically different amounts of care required for the treatment for the heterogeneous diseases included in certain DRGs. The DRG system does not differentiate between mild and severe forms of the same disease; therefore, the costs for university hospitals, which often receive more complicated patients than are admitted to community hospitals, may appear to be disproportionately high. Horn et al. (7) discussed the failure of unadjusted DRGs to adequately account for severity of illness. Subsequently, a Computerized Severity Index has been developed that facilitates the prediction of LOS within a DRG (8)(9). None of the data presented in this study relating to lengths of stay and costs were adjusted for severity of illness. That is, within each DRG, age, sex, and severity on admission were not specifically considered. Likewise, apart from the conditions that specifically state an age in their description, the data were not adjusted for age.
By far the most common cause of admission to the 60 University hospitals included in this study was uncomplicated vaginal delivery, and the discharge of normal newborns was the second most frequent DRG. The number of normal vaginal deliveries was more than twice the number of discharges for chemotherapy without leukemia as a secondary diagnosis, the third most common cause of admission.
University hospitals typically have had higher costs than community hospitals, although they admit patients with many of the most common diseases as well as those with more complex diseases. In 1985, Frick et al. (10) showed that the average cost per case was 60% higher in teaching hospitals than in nonteaching hospitals, but only approximately one-fourth of the difference was attributable to differences in case mix. It is not possible to determine how much of the current differences may be attributed to differences in severity because the differences in cost between teaching and nonteaching hospitals has been forced down. In teaching hospitals, such as those that are members of UHC, there typically has been greater use of resources than in community hospitals (10)(11)(12)(13). Exceptions to these generalizations, however, should be noted. For example, there is evidence that for some common disease treatments, such as cholecystectomy, charges and lengths of stay may be comparable in teaching and community hospitals in the same region (14). The data in this study strongly suggest that the diseases that are most expensive to manage are not likely candidates for treatment in community hospitals
This study identifies the major sources contributing to the overall costs incurred in managing patients with most of the diseases treated in hospitals. Such an approach enables a hospital to compare its experience with that of its peers, even allowing a hospital to make a decision to embark on new programs based on solid background information. The information derived from 60 hospitals eliminates biases that might occur when only the experience of 1 hospital is considered. The data do reflect the experience of teaching hospitals, which have higher expenses per patient, which has been attributed to sicker patients, more aggressive treatment, higher quality care, and patient care subsidizing teaching or inefficiency; therefore, the data may not be directly exportable to other hospitals (15). With the continuing efforts in every hospital to contain costs, e.g., by reducing LOS, the data presented here may reflect a transient situation only, but it is unlikely that the relative costs of treating the different diseases will change radically. Nevertheless, it would be interesting to reassess the costs at periodic intervals.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 The following additional Tables are included in a data supplement with this article at Clinical Chemistry Online (www.clinchem.org/content/vol46/issue7/): Supplemental Tables: S1. DRGs with highest ratios of median to mean LOS. S2. DRGs with lowest ratios of median to mean LOS. S3. DRGs with highest (A) and lowest (B) ratios of mean to median total costs. S4. DRGs with median surgical costs >$5000. S5. DRGs (n = 20) with highest median total costs/day with surgical costs excluded. S6. DRGs with highest (A) and lowest (B) proportions of costs attributable to accommodation. S7. Total costs and relative weights of high-complexity DRGs. S8. DRGs with highest (A) and lowest (B) ratios of costs to relative weights. S9. Aggregate component costs of surgical DRGs with and without surgical and transplant-specific costs.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults2DiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  D. S. Young, B. S. Sachais, and L. C. Jefferies Effect of Disease Complications on Hospital Costs Clin. Chem., January 1, 2002; 48(1): 140 - 149. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. S. Young, B. S. Sachais, and L. C. Jefferies Comparative Costs of Treating Adults and Children within Selected Diagnosis-related Groups Clin. Chem., January 1, 2002; 48(1): 150 - 160. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. S. Young, B. S. Sachais, and L. C. Jefferies Laboratory Costs in the Context of Disease Clin. Chem., July 1, 2000; 46(7): 967 - 975. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
