Diagnosis-related group refinement with diagnosis- and procedure-specific comorbidities and complications

Day Admission Surgery Program in a Prospective Payment System: What Are the Financial Incentives?

Background

Day admission surgery (DAS) is meant to provide a better in-hospital experience for patients and to save costs by reducing the length of stay. However, in a prospective payment system, it may also reduce the reimbursement amount, leading to unintended incentives for hospitals.

Methods

Over a 4-month period in 2021 and based on predefined clinical and logistic criteria, patients from different surgical sub-specialties were identified to follow the institutional DAS program. Revenue-analysis was performed, considering the Swiss diagnosis-related group (SwissDRG) prospective payment policy. Revenue with DAS program was compared to revenue if patients were admitted the day prior surgery (No DAS) using nonparametric pooled bootstrap t-test. All other costs considered identical, an estimation of the average cost spared due to the avoidance of pre-operative hospitalization in the DAS setting was carried out using a micro-costing approach.

Results

Overall, 105 inpatients underwent DAS over the study period, totaling a revenue of CHF 1 209 840. Among them, 25 patients (24%) were low outliers due to the day spared from the DAS program and triggering a mean (SD) financial discount of Swiss Francs (CHF) 4192 (2835), yielding a total amount of CHF 105 435. DAS revealed a mean revenue of CHF 7320 (656), compared to CHF 11 510 (1108) if patients were admitted the day before surgery (No DAS, P = .007).

Conclusion

In a PPS, anticipation of financial penalties when implementing a DAS for all-comers is key to prevent an imbalance of the hospital equation if no financial criteria are used to select eligible patients. Promptly revising workflow to maintain constant fixed costs for a greater number of patients may be a valuable hedging strategy.

Impact of precoding on reimbursement in diagnosis-related group systems: Randomized controlled trial

BACKGROUND

Complete and correct documentation of diagnosis and procedures is essential for adequate health provider reimbursement in diagnosis-related group (DRG) systems. The objective of this study was to investigate whether daily monitoring and semiautomated proposal optimization of DRG coding (precoding) is associated with higher reimbursement per hospitalization day.

MATERIALS AND METHODS

This parallel-group, unblinded, randomized clinical trial randomized patients 1:1 into intervention (precoding) and control groups. Between June 12 and December 6, 2019 all hospitalized patients (1566 cases) undergoing elective or emergency surgery at the department of surgery in a Swiss hospital were eligible for this study. By random sample selection, cases were assigned to the intervention (precoding) and control groups. The primary outcome was the total reimbursement, divided by the length of stay.

RESULTS

Of the 1205 randomized cases, 1200 (precoding group: 602) remained for intention-to-treat, and 1131 (precoding group: 564) for per-protocol analysis. Precoding increased reimbursement per hospitalization day by 6.5% (160 US dollars; 95% confidence interval 31 to 289; P = 0.015). In a regression analysis patients hospitalized 7 days or longer, precoding increased reimbursement per day by 10.0% (246 US dollars; 95% confidence interval -12 to 504; P = 0.021). More secondary diagnoses (mean [SD]: 5.16 [5.60] vs 4.39 [5.34]; 0.77; 95% confidence interval 0.15 to 1.39; P = 0.015) and nonsurgical postoperative complications (mean [SD]: 0.68 [1.45] vs 0.45 [1.12]; 0.23; 95% confidence interval 0.08 to 0.38; P = 0.002) were documented by precoding. No associated was observed regarding the length of stay, total reimbursement, or case mix index. The mean (SD) precoding time effort was 37 (27) minutes per case.

CONCLUSION

Physician-led precoding increases DRG-based reimbursement. Precoding is time consuming and should be focused on cases with a longer hospital stay to increase efficiency.

CONECT-6: a case-finding tool to identify patients with complex health needs

Background

Early identification of patients with chronic conditions and complex health needs in emergency departments (ED) would enable the provision of services better suited to their needs, such as case management. A case-finding tool would ultimately support ED teams to this end and could reduce the cost of services due to avoidable ED visits and hospitalizations. The aim of this study was to develop and validate a short self-administered case-finding tool in EDs to identify patients with chronic conditions and complex health needs in an adult population.

Methods

This prospective development and initial validation study of a case-finding tool was conducted in four EDs in the province of Quebec (Canada). Adult patients with chronic conditions were approached at their third or more visit to the ED within 12 months to complete a self-administered questionnaire, which included socio-demographics, a comorbidity index, the reference standard INTERMED self-assessment, and 12 questions to develop the case-finding tool. Significant variables in bivariate analysis were included in a multivariate logistic regression analysis and a backward elimination procedure was applied. A receiver operating characteristic (ROC) curve was developed to identify the most appropriate threshold score to identify patients with complex health needs.

Results

Two hundred ninety patients participated in the study. The multivariate analysis yielded a six-question tool, COmplex NEeds Case-finding Tool – 6 (CONECT-6), which evaluates the following variables: low perceived health; limitations due to pain; unmet needs; high self-perceived complexity; low income; and poor social support. With a threshold of two or more positive answers, the sensitivity was 90% and specificity 66%. The positive and negative predictive values were 49 and 75% respectively.

Conclusions

The case-finding process is the essential characteristic of case management effectiveness. This study presents the first case-finding tool to identify adult patients with chronic conditions and complex health needs in ED.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-021-06154-4.

A review of the complexity adjustment in the Korean Diagnosis-Related Group (KDRG)

Background:

The Korean Diagnosis-Related Groups (KDRG) was revised in 2003, modifying the complexity adjustment mechanism of the Australian Refined Diagnosis-Related Groups (AR-DRGs). In 2014, the Complication and Comorbidity Level (CCL) of the existing AR-DRG system was found to have very little correlation with cost.

Objective:

Based on the Australian experience, the CCL for KDRG version 3.4 was reviewed.

Method:

Inpatient claim data for 2011 were used in this study. About 5,731,551 episodes, which had one or no complication and comorbidity (CC) and met the inclusion criteria, were selected. The differences of average hospital charges by the CCL were analysed in each Adjacent Diagnosis-Related Group (ADRG) using analysis of variance followed by Duncan’s test. The patterns of differences were presented with R 2 in three patterns: The CCL reflected the complexity well (VALID); the average charge of CCL 2, 3, 4 was greater than CCL 0 (PARTIALLY VALID); the CCL did not reflect the complexity (NOT VALID).

Results:

A total of 114 (19.03%), 190 (31.72%) and 295 (49.25%) ADRGs were included in VALID, PARTIALLY VALID and NOT VALID, respectively. The average R 2 for hospital charge of CCL was 4.94%. The average R 2 in VALID, PARTIALLY VALID and NOT VALID was 4.54%, 5.21%, and 4.93%, respectively.

Conclusion:

The CCL, the first step of complexity adjustment using secondary diagnoses, exhibited low performance. If highly accurate coding data and cost data become available, the performance of secondary diagnosis as a variable to reflect the case complexity should be re-evaluated.

Implications:

Lack of reviewing the complexity adjustment mechanism of the KDRG since 2003 has resulted in outdated CC lists and levels that no longer reflect the current Korean healthcare system. Reliable cost data (vs. charge) and accurate coding are essential for accuracy of reimbursement.

Differences between determinants of in-hospital mortality and hospitalisation costs for patients with acute heart failure: a nationwide observational study from Japan

OBJECTIVES

Although current case-mix classifications in prospective payment systems were developed to estimate patient resource usage, whether these classifications reflect clinical outcomes remains unknown. The efficient management of acute heart failure (AHF) with high mortality is becoming more important in many countries as its prevalence and associated costs are rapidly increasing. Here, we investigate the determinants of in-hospital mortality and hospitalisation costs to clarify the impact of severity factors on these outcomes in patients with AHF, and examine the level of agreement between the predicted values of mortality and costs.

DESIGN

Cross-sectional observational study.

SETTING AND PARTICIPANTS

A total of 19 926 patients with AHF from 261 acute care hospitals in Japan were analysed using administrative claims data.

MAIN OUTCOME MEASURES

Multivariable logistic regression analysis and linear regression analysis were performed to examine the determinants of in-hospital mortality and hospitalisation costs, respectively. The independent variables were grouped into patient condition on admission, postadmission procedures indicating disease severity (eg, intra-aortic balloon pumping) and other high-cost procedures (eg, single-photon emission CT). These groups of independent variables were cumulatively added to the models, and their effects on the models' abilities to predict the respective outcomes were examined. The level of agreement between the quartiles of predicted mortality and predicted costs was analysed using Cohen's κ coefficient.

RESULTS

In-hospital mortality was associated with patient's condition on admission and severity-indicating procedures (C-statistics 0.870), whereas hospitalisation costs were associated with severity-indicating procedures and high-cost procedures (R² 0.32). There were substantial differences in determinants between the outcomes. In addition, there was no consistent relationship observed (κ=0.016, p<0.0001) between the quartiles of in-hospital mortality and hospitalisation costs.

CONCLUSIONS

The determinants of mortality and costs for hospitalised patients with AHF were generally different. Our results indicate that the same case-mix classifications should not be used to predict both these outcomes.

Evaluation of the Performance of Diagnosis-Related Groups and Similar Casemix Systems: Methodological Issues

With the increasing recognition and application of casemix for managing and financing healthcare resources, the evaluation of alternative versions of systems such as diagnosis-related groups (DRGs) has been afforded high priority by governments and researchers in many countries. Outside the United States, an important issue has been the perceived need to produce local versions, and to establish whether or not these perform more effectively than the US-based classifications. A discussion of casemix evaluation criteria highlights the large number of measures that may be used, the rationale and assumptions underlying each measure, and the problems in interpreting the results. A review of recent evaluation studies from a number of countries indicates that considerable emphasis has been placed on the predictive validity criterion, as measured by the R2 statistic. However, the interpretation of the findings has been affected greatly by the methods used, especially the treatment and definition of outlier cases. Furthermore, the extent to which other evaluation criteria have been addressed has varied widely. In the absence of minimum evaluation standards, it is not possible to draw clear-cut conclusions about the superiority of one version of a casemix system over another, the need for a local adaptation, or the further development of an existing version. Without the evidence provided by properly designed studies, policy-makers and managers may place undue reliance on subjective judgements and the views of the most influential, but not necessarily best informed, healthcare interest groups.

Australian diagnosis related groups: Drivers of complexity adjustment

BACKGROUND

In undertaking a major revision to the Australian Refined Diagnosis Related Group (ARDRG) classification, we set out to contrast Australia's approach to using data on additional (not principal) diagnoses with major international approaches in splitting base or Adjacent Diagnosis Related Groups (ADRGs).

METHODS

Comparative policy analysis/narrative review of peer-reviewed and grey literature on international approaches to use of additional (secondary) diagnoses in the development of Australian and international DRG systems.

ANALYSIS

European and US approaches to characterise complexity of inpatient care are well-documented, providing useful points of comparison with Australia's. Australia, with good data sources, has continued to refine its national DRG classification using increasingly sophisticated approaches. Hospital funders in Australia and in other systems are often under pressure from provider groups to expand classifications to reflect clinical complexity. DRG development in most healthcare systems reviewed here reflects four critical factors: these socio-political factors, the quality and depth of the coded data available to characterise the mix of cases in a healthcare system, the size of the underlying population, and the intended scope and use of the classification. Australia's relatively small national population has constrained the size of its DRG classifications, and development has been concentrated on inpatient care in public hospitals.

DISCUSSION AND CONCLUSIONS

Development of casemix classifications in health care is driven by both technical and socio-political factors. Use of additional diagnoses to adjust for patient complexity and cost needs to respond to these in each casemix application.

Analysis of subarachnoid hemorrhage using the Nationwide Inpatient Sample: the NIS-SAH Severity Score and Outcome Measure

Object

Studies using the Nationwide Inpatient Sample (NIS), a large ICD-9–based (International Classification of Diseases, Ninth Revision) administrative database, to analyze aneurysmal subarachnoid hemorrhage (SAH) have been limited by an inability to control for SAH severity and the use of unverified outcome measures. To address these limitations, the authors developed and validated a surrogate marker for SAH severity, the NIS-SAH Severity Score (NIS-SSS; akin to Hunt and Hess [HH] grade), and a dichotomous measure of SAH outcome, the NIS-SAH Outcome Measure (NIS-SOM; akin to modified Rankin Scale [mRS] score).

Methods

Three separate and distinct patient cohorts were used to define and then validate the NIS-SSS and NIS-SOM. A cohort (n = 148,958, the “model population”) derived from the 1998–2009 NIS was used for developing the NIS-SSS and NIS-SOM models. Diagnoses most likely reflective of SAH severity were entered into a regression model predicting poor outcome; model coefficients of significant factors were used to generate the NIS-SSS. Nationwide Inpatient Sample codes most likely to reflect a poor outcome (for example, discharge disposition, tracheostomy) were used to create the NIS-SOM.

Data from 716 patients with SAH (the “validation population”) treated at the authors' institution were used to validate the NIS-SSS and NIS-SOM against HH grade and mRS score, respectively.

Lastly, 147,395 patients (the “assessment population”) from the 1998–2009 NIS, independent of the model population, were used to assess performance of the NIS-SSS in predicting outcome. The ability of the NIS-SSS to predict outcome was compared with other common measures of disease severity (All Patient Refined Diagnosis Related Group [APR-DRG], All Payer Severity-adjusted DRG [APS-DRG], and DRG).

Results

The NIS-SSS significantly correlated with HH grade, and there was no statistical difference between the abilities of the NIS-SSS and HH grade to predict mRS-based outcomes. As compared with the APR-DRG, APSDRG, and DRG, the NIS-SSS was more accurate in predicting SAH outcome (area under the curve [AUC] = 0.69, 0.71, 0.71, and 0.79, respectively).

A strong correlation between NIS-SOM and mRS was found, with an agreement and kappa statistic of 85% and 0.63, respectively, when poor outcome was defined by an mRS score > 2 and 95% and 0.84 when poor outcome was defined by an mRS score > 3.

Conclusions

Data in this study indicate that in the analysis of NIS data sets, the NIS-SSS is a valid measure of SAH severity that outperforms previous measures of disease severity and that the NIS-SOM is a valid measure of SAH outcome. It is critically important that outcomes research in SAH using administrative data sets incorporate the NIS-SSS and NIS-SOM to adjust for neurology-specific disease severity.

An Epidemiological Study of Anemia and Renal Dysfunction in Patients Admitted to ICUs across the United States

The aims of this study were to determine the associations between anemia of critical illness, erythropoietin stimulating agents (ESA), packed red blood cell transfusions and varying degrees of renal dysfunction with mortality, and ICU- and hospital length of stay (LOS). This was a cross-sectional retrospective study of 5,314 ICU patients from USA hospitals. Hospital, patient demographics, and clinical characteristics were collected. Predictors of mortality and hospital and ICU LOS were evaluated using multivariate logistic regression models. The mean ICU admission hemoglobin in this study was 9.4 g/dL. The prevalence of ESA use was 13% and was associated with declining renal function; 26% of the ICU patients in this study received transfusion. ESA utilization was associated with 28% longer hospital LOS (P < 0.001). ICU LOS was increased by up to 18% in patients with eGFR rates of <30 and 30-59 mL/min/1.73 m(2), respectively (P < 0.05) but not in those receiving dialysis. Mortality was significantly associated with renal dysfunction and dialysis with odds ratios of 1.94, 2.66 and 1.40 for the dialysis, and eGFR rates of <30 and 30-59 and mL/min/1.73 m(2), respectively (P < 0.05). These data provide a snapshot of anemia treatment practices and outcomes in USA ICU patients with varying degrees of renal dysfunction.

Relating illness complexity to reimbursement in CKD patients

Background:

Despite significant investments of federal and state dollars to transition patient medical records to an all-electronic system, a chasm still exists between health care quality and payment for it. A major reason for this gap is the difficulty in evaluating health care outcomes based on claims data. Since both payers and patients may not appreciate how illness complexity impacts treatment outcomes, it is difficult to determine fair provider compensation.

Objectives:

Chronic kidney disease (CKD) typifies these problems and is often associated with comorbidities that impact cost, health, and work productivity. Thus, the objective of this study was to evaluate an illness complexity score (ICS) based on a linear regression of select blood values that might assist in predicting average monthly reimbursements in CKD patients. A second objective was to compare the results of this ICS prediction to results obtained by prediction of average monthly reimbursement using CKD stage. A third objective was to analyze the relationship between the change in ICS, estimated glomerular filtration rate (eGFR), and CKD stage over time to average monthly reimbursement.

Methods:

We calculated parsimonious values for select variables associated with CKD patients and compared the ICS to ordinal staging of renal disease. Data from 177 de-identified patients over 13 months was collected, which included 15 blood chemistry observations along with complete claims data for all medical expenses. To test for the relationship between average blood chemistry values, stages of CKD, age, and average monthly reimbursement, we modeled an association through a linear regression function of age, eGFR, and the Z-scores calculated from average monthly values of phosphorus, parathyroid hormone, glucose, hemoglobin, bicarbonate, albumin, creatinine, blood urea nitrogen, potassium, calcium, sodium, alkaline phosphatase, alanine aminotransferase, and white blood cells.

Results:

The results of our study demonstrated that the association between average ICS values throughout the entire study period predicted average monthly reimbursements with an R² value of 0.41. Comparing that value to the association between the average CKD stage and average monthly reimbursement demonstrated an R² value of 0.08. Thus, ICS offers five times greater sensitivity over CKD staging as a measure of illness complexity.

Conclusion:

Sorting the patient population by changes in CKD stage or ICS over the entire study period revealed significant differences between the two scoring methods. Groups scored by ICS demonstrated greater sensitivity by capturing dysfunction in other organ systems and had a better association with reimbursement than groups scored by CKD staging.

Equivalence of two healthcare costing methods: bottom-up and top-down

This paper compares two quite different approaches to estimating costs: a 'bottom-up' approach, represented by the US Department of Veterans Affairs' (VA) Decision Support System that uses local costs of specific inputs; and a 'top-down' approach, represented by the costing system created by the VA Health Economics Resource Center, which assigns the VA national healthcare budget to specific products using various weighting systems. Total annual costs per patient plus the cost for specific services (e.g. clinic visit, radiograph, laboratory, inpatient admission) were compared using scatterplots, correlations, mean difference, and standard deviation of individual differences. Analysis are based upon 2001 costs for 14 915 patients at 72 facilities. Correlations ranged from 0.24 for the cost of outpatient encounters to 0.77 for the cost of inpatient admissions, and 0.85 for total annual cost. The mean difference between costing methods was $707 ($4168 versus $3461) for total annual cost. The standard deviation of the individual differences was $5934. Overall, the agreement between the two costing systems varied by the specific cost being measured and increased with aggregation. Administrators and researchers conducting cost analyses need to carefully consider the purpose, methods, characteristics, strengths, and weaknesses when selecting a method for assessing cost.

Reliability of diagnostic coding in intensive care patients

Introduction

Administrative coding of medical diagnoses in intensive care unit (ICU) patients is mandatory in order to create databases for use in epidemiological and economic studies. We assessed the reliability of coding between different ICU physicians.

Method

One hundred medical records selected randomly from 29,393 cases collected between 1998 and 2004 in the French multicenter Outcomerea ICU database were studied. Each record was sent to two senior physicians from independent ICUs who recoded the diagnoses using the International Statistical Classification of Diseases and Related Health Problems: Tenth Revision (ICD-10) after being trained according to guidelines developed by two French national intensive care medicine societies: the French Society of Intensive Care Medicine (SRLF) and the French Society of Anesthesiology and Intensive Care Medicine (SFAR). These codes were then compared with the original codes, which had been selected by the physician treating the patient. A specific comparison was done for the diagnoses of septicemia and shock (codes derived from A41 and R57, respectively).

Results

The ICU physicians coded an average of 4.6 ± 3.0 (range 1 to 32) diagnoses per patient, with little agreement between the three coders. The primary diagnosis was matched by both external coders in 34% (95% confidence interval (CI) 25% to 43%) of cases, by only one in 35% (95% CI 26% to 44%) of cases, and by neither in 31% (95% CI 22% to 40%) of cases. Only 18% (95% CI 16% to 20%) of all codes were selected by all three coders. Similar results were obtained for the diagnoses of septicemia and/or shock.

Conclusion

In a multicenter database designed primarily for epidemiological and cohort studies in ICU patients, the coding of medical diagnoses varied between different observers. This could limit the interpretation and validity of research and epidemiological programs using diagnoses as inclusion criteria.

Uncomplicated acute renal failure and post-hospital care: a not so uncomplicated illness

BACKGROUND

Although uncomplicated acute renal failure (ARF) is associated with significant hospital resource utilization, its health care requirements following hospital discharge are not well understood. The goal of this study was to characterize the post-hospital care requirements incurred by patients with uncomplicated ARF and to determine its important influencing factors.

METHODS

We obtained hospital case mix data sets for a 2-year period (1999-2000) from the Massachusetts Division of Health Care Finance and Policy. Utilizing DRG and ICD-9-CM codes from 23 Massachusetts hospitals, we identified 2,128 adult patients whose primary reason for hospitalization was uncomplicated ARF. Post-hospital care was defined as the receipt of extended facility care or home health care following hospital discharge.

RESULTS

Nearly 50% of patients hospitalized with uncomplicated ARF required some type of post-hospital care, of whom 27% underwent extended facility care while 22% received home health care. The post-hospital care requirements for uncomplicated ARF were similar to those for serious medical conditions (e.g. heart failure) and exceeded those of many common illnesses (e.g., bronchitis). Advancing age, worsening severity of illness, female gender, and emergency room admission were independently associated with receipt of post-hospital care (p < 0.05). A trend existed between less frequent post-hospital care requirements and hospitalization at academic medical centers compared with non-academic hospitals.

CONCLUSIONS

Uncomplicated ARF is frequently associated with prolonged care following hospitalization. As the health care utilization for ARF becomes better characterized, these post-hospital care resources should not be overlooked.

Assessing the Quality of Surgical Care

Because of better educated patients, more demanding payers, and regulatory agencies, safety and quality have become prominent criteria for evaluating surgical care. Providers are increasingly asked to document these areas, and patients are using this documentation to select surgeons and hospitals. Payers are using the data to direct patients to providers, and potentially to adjust reimbursement rates. Therefore, health care policy makers, health service researchers, and others are aggressively developing and implementing quality indicators for surgical practice. Given the complex interplay of structure, process, and outcomes, assessment of surgical quality presents a daunting task. We must firmly establish the links between these elements to validate current and future metrics, while engendering "buy-in'' on the part of surgeons.

Uncomplicated acute renal failure and hospital resource utilization: a retrospective multicenter analysis

BACKGROUND

Although acute renal failure (ARF) complicating nonrenal organ dysfunction in the intensive care unit is associated with significant mortality and hospital costs, hospital resource utilization attributed to uncomplicated ARF is not well known. The goal of this study is to characterize the costs and lengths of stay (LOSs) incurred by hospitalized patients with uncomplicated ARF and their important determining factors.

METHODS

We obtained hospital case-mix data sets from 23 Massachusetts hospitals for a 2-year period (1999 to 2000) from the Massachusetts Division of Health Care Finance and Policy. A total of 2,252 records of patients hospitalized with uncomplicated ARF were identified. Patient records of other common medical diagnoses were studied for comparison.

RESULTS

Patients hospitalized with uncomplicated ARF incurred median direct hospital costs of 2,600 dollars, median hospital LOS of 5 days, and mortality of 8%. Dialysis was independently associated with significantly greater hospital costs and LOSs for patients with uncomplicated ARF (P < 0.05). Male sex and nonwhite race were associated with significantly lower hospital costs and LOSs, whereas type of hospital had opposing effects on these 2 resource utilization outcomes (P < 0.05). Unadjusted aggregate resource utilization associated with uncomplicated ARF exceeded that of many other common illnesses.

CONCLUSION

Demographic and hospital factors, as well as dialysis therapy, are significant determinants of hospital resource utilization for patients with uncomplicated ARF. Uncomplicated ARF appears to incur greater hospital costs and longer LOSs compared with other common medical conditions. Greater focus should be directed toward further understanding of the factors influencing resource utilization for ARF.

Assessment of nursing management and utilization of nursing resources with the RAFAELA patient classification system - case study from the general wards of one central hospital

BACKGROUND

RAFAELA is a new Finnish PCS, which is used in several University Hospitals and Central Hospitals and has aroused considerable interest in hospitals in Europe.

AIMS AND OBJECTIVES

The aim of the research is firstly to assess the feasibility of the RAFAELA Patient Classification System (PCS) in nursing staff management and, secondly, whether it can be seen as the transferring of nursing resources between wards according to the information received from nursing care intensity classification.

METHODS

The material was received from the Central Hospital's 12 general wards between 2000 and 2001. The RAFAELA PCS consists of three different measures: a system measuring patient care intensity, a system recording daily nursing resources, and a system measuring the optimal nursing care intensity/nurse situation. The data were analysed in proportion to the labour costs of nursing work and, from that, we calculated the employer's loss (a situation below the optimal level) and savings (a situation above the optimal level) per ward as both costs and the number of nurses.

RESULTS

In 2000 the wards had on average 77 days below the optimal level and 106 days above it. In 2001 the wards had on average 71 days below the optimal level and 129 above it. Converting all these days to monetary and personnel resources the employer lost 307,745 or 9.84 nurses and saved 369,080 or 11.80 nurses in total in 2000. In 2001 the employer lost in total 242,143 or 7.58 nurses and saved 457,615 or 14.32 nurses. During the time period of the research nursing resources seemed not have been transferred between wards.

CONCLUSIONS

RAFAELA PCS is applicable to the allocation of nursing resources but its possibilities have not been entirely used in the researched hospital. The management of nursing work should actively use the information received in nursing care intensity classification and plan and implement the transferring of nursing resources in order to ensure the quality of patient care.

RELEVANCE TO CLINICAL PRACTICE

Information on which units resources should be allocated to is needed in the planning of staff resources of the whole hospital. More resources do not solve the managerial problem of the right allocation of resources. If resources are placed wrongly, the problems of daily staff management and cost control continue.

Recalibration of the pediatric risk of admission score using a multi-institutional sample

STUDY OBJECTIVE

Case-mix adjustment is a critical component of quality assessment and benchmarking. The Pediatric Risk of Admission (PRISA) score is composed of descriptive, physiologic, and diagnostic variables that provide a probability of hospital admission as an index of severity. The score was developed and validated in a single tertiary pediatric hospital emergency department (ED) after exclusion of children with minor injuries and illnesses. We provide a multi-institutional recalibration and validation of the PRISA score and test its performance in 4 additional EDs, including patients with minor injuries and illnesses.

METHODS

Masked, photocopied, randomly selected medical records of ED patients from 2000 were abstracted and were used to test the performance (discrimination and calibration) of the original PRISA score. This sample differed from the original PRISA sample by including 5 hospitals and including patients with minor injuries and minor illnesses. Independent variables included components of acute and chronic history, physiologic variables, and 3 ED therapies. The dependent variable was hospital admission. PRISA was then recalibrated as needed by using an 80% development sample and a 20% validation sample. Area under the curve and the Hosmer-Lemeshow goodness-of-fit test were used to measure, respectively, discrimination and calibration of the PRISA score after recalibration. We then applied the recalibrated PRISA score to secondary outcomes to test construct validity. We reasoned that a valid measure of ED severity should also be associated with the secondary outcomes of mandatory admissions (admissions using > or =1 inpatient resources) and ICU admissions.

RESULTS

The recalibrated PRISA score performed well in all deciles of predicted probability of admission. The area under the curve was 0.81 and the calibration was good (Hosmer-Lemeshow 10.658; df=8; P=.222) for the development sample, and the area under the curve was 0.785 with excellent calibration (Hosmer-Lemeshow 8.341; df=9; P=.500) for the validation sample. The overall development sample had 423.9 admissions predicted and 423 observed; the validation sample had 112.1 predicted and 110 observed.

CONCLUSION

The PRISA score has been recalibrated and performs well in EDs of tertiary pediatric hospitals. Comparison with this benchmark may allow individual EDs to improve their performance and may provide insight into best practices.

Hospitalists and an innovative emergency department admission process

After treatment in an emergency department (ED), patients often wait several hours for hospital admission, resulting in dissatisfaction and increased wait times for both admitted and other ED patients. We implemented a new direct admission system based on telephone consultation between ED physicians and in-house hospitalists. We studied this system, measuring admission times, length of stay, and mortality. Postintervention, admission times averaged 18 minutes for transfer to the ward compared to 2.5 hours preintervention, while pre- and postintervention length of stay and mortality rates remained similar.

Costs of childhood physical abuse: comparing inflicted and unintentional traumatic brain injuries

OBJECTIVES

To estimate the effect of early childhood abuse (ie, inflicted injury) on medical costs of head trauma.

METHODS

Abstracts of patient records were drawn from the annual 1993-2000 Colorado state-mandated hospital discharge database maintained by the Colorado Hospital Association. The 2 dependent variables were total charges (TC) and length of stay. Our key independent variable was the nature of injury, ie, inflicted or unintentional; other independent variables were age, severity level, death, and trauma designation of the hospital. Comparisons of variables between patients with inflicted and unintentional head trauma were performed using Student's t tests or chi2 statistics. Ordinary least squares regression was used to estimate the marginal and total effects of inflicted injury on TC and LOS.

RESULTS

Of the 1097 head trauma patients <3 years old, 814 had unintentional and 283 had inflicted head trauma. Head trauma was defined using the Centers for Disease Control definition of traumatic brain injury. Patients with inflicted injuries were younger and had a higher average severity level and overall mortality rate than did patients with unintentional head trauma. The regression models showed that, controlling for age and severity, patients with inflicted head trauma stayed in the hospital 52% longer (2 days), and had a mean total bill 89% higher (4232 dollars more) than did patients with unintentional head trauma.

CONCLUSIONS

The findings from multivariate models of TC and length of stay corroborate the simpler univariate findings of earlier studies. By focusing on the impact of those cases of child abuse that lead to a specific, severe clinical entity (traumatic brain injury), we isolated a significant economic impact of abuse on health care expenditures for injury.

Predicting outcome after appendicectomy

AIM

To validate an intraoperative appendicitis severity score (IASS) and examine outcome following emergency appendectomy.

METHODS

A prospective study was undertaken, enrolling consecutive patients undergoing emergency appendicectomy. Data were obtained independently on preoperative Alvarado scores, IASS (0-3: 0 no inflammation, 1 engorged appendix/no peritonitis, 2 peritoneal reaction/exudate or 3 evidence of perforation/abscess) and postoperative outcome parameters.

RESULTS

There were 149 patients identified with a mean age of 20.7 years. There was no association between Alvarado score and length of hospital stay, septic complication, patient sex or duration of symptoms (p>0.05). IASS was found to be an independent risk factor for septic complication, wound infection (p<0.05) and length of hospital stay (p<0.001). There was no correlation between preoperative duration of symptoms or time until surgery and intraoperative score.

CONCLUSIONS

This simple scoring system can identify patients more likely to suffer morbidity following emergency appendicectomy. Specifically, this system identifies patients who have a high risk of sepsis and therefore could be of use when comparing healthcare performance.

Age and laboratory costs for hospitalized medical patients

OBJECTIVE

To examine the hypothesis that older hospitalized patients have higher laboratory costs than younger patients in the same severity-adjusted diagnosis-related group (DRG).

DESIGN

We obtained hospital case mix data sets (1995-1997) from the Massachusetts Division of Health Care Finance and Policy. We selected discharge abstracts from 4 medical DRGs, at 5 large academic hospitals (n = 15,265) and 5 midsized community hospitals (n = 10,540), for analysis. We converted laboratory and blood product charges to direct costs using the department-specific ratio of cost to charges. We adjusted diagnostic groups for severity of comorbid conditions and complications using the refined DRG method.

MAIN OUTCOME MEASURES

Hospital length of stay (LOS), laboratory direct cost (LDC) per hospitalization, LDC per hospital day, and ratio of LDC to total direct cost.

RESULTS

Hospital LOS was longer for older patients in all comparisons. Laboratory direct cost per hospitalization was higher for older patients in some DRGs, but lower in other DRGs. Laboratory direct cost per hospital day was almost always less for older patients than for younger patients, both at academic and community hospitals. Data stratification by gender, admission status, and principal diagnosis yielded substantially the same pattern of cost differences observed within the larger data set.

CONCLUSIONS

Older medical patients have longer hospital stays and generally higher costs. These patients also have a significantly decreased rate of laboratory resource consumption over the course of hospitalization (LDC per hospital day), as well as lower laboratory costs as a proportion of total costs. Age-specific differences in LOS and cost parameters were essentially unchanged after controlling for several potential sources of bias.

Outcomes research: getting started, defining outcomes, a framework, and data sources

As key providers of care and as participants in systems delivering care, APNs are in ideal positions to explore, identify, measure, and evaluate patient outcomes. This article has provided an overview of some of the basic knowledge and skills required to undertake outcome studies. Through the application of these basic principles, the beginning APN can start to explore patient outcomes via quality improvement activities. With mastery and expansion of these principles, the expert APN can easily progress into outcomes research studies, which provide substantive contributions to the outcomes research agenda.

A typology for provider payment systems in health care

A typology to classify provider payment systems from an incentive point of view is developed. We analyse the way, how these systems can influence provider behaviour and, a fortiori, contribute to attain the general objectives of health care, i.e. quality of care, efficiency and accessibility. The first dimension of the typology indicates whether there is a link between the provider's income and his activity. In variable systems, the provider has an ability to influence his earnings, contrary to fixed systems. The second dimension indicates whether the provider's payments are related to his actual costs or not. In retrospective systems, the provider's own costs are the basis for reimbursement ex post whereas in prospective systems payments are determined ex ante without any link to the real costs of the individual provider. These different characteristics are likely to influence provider behaviour in different ways. Furthermore the most frequently used criteria to determine the provider's income are discussed: per service, per diem, per case, per patient and per period. Also a distinction is made between incentives at the level of the individual provider (micro-level) and the sponsor (macro-level). Finally, the potential interactions when several payment systems are used simultaneously are discussed. This typology is useful to classify and compare different types of payment systems as prevailing in different countries, and provides a useful framework for future research of health care payment systems.

Effect of Disease Complications on Hospital Costs

Background: To test the hypothesis that complications increase the use of resources in managing patients in hospitals, we examined the costs of managing patients with the same disease with and without complications.

Methods: We used a database developed by the University HealthSystems Consortium that contains the costs of managing more than 1 million patients in 60 University hospitals. We created a simplified database of the costs of 457 445 patients in 111-paired diagnosis-related groups (DRGs) that were classified as either having or not having complications and/or comorbidities. Costs were calculated from the ratio of costs to charges within the individual hospitals.

Results: The median costs of managing patients with complications were higher than those for managing patients without complications, confirming the appropriateness of the dual classification. Notably, these extra costs were largely incurred through increased length of stay. Of note, the cost per day for DRGs with complications and/or comorbidities was most often less than that for the corresponding uncomplicated conditions. Although accommodation costs generally were the largest single component of total costs for both complicated and uncomplicated conditions, in only 31 DRGs (15 with complications, 16 without) did they account for more than one-half the total costs. Laboratory and drug costs were higher for complicated conditions, but as a proportion of total costs were comparable for complicated and uncomplicated conditions.

Conclusions: Complications in patients are associated with increased hospital costs, although the costs per day of hospitalization are often less than in patients without such complications.

Relationship of hospital size, case volume, and cost for coronary artery bypass surgery: analysis of 12,774 patients operated on in Massachusetts during fiscal years 1995 and 1996

OBJECTIVE

This study investigates the relationship between the cost of coronary artery bypass graft surgery and both hospital size and case volume.

METHODS

Retrospective administrative and cost data were obtained for all 12,774 patients who underwent isolated coronary bypass surgery at 12 Massachusetts hospitals during 1995 and 1996. Hospitals were stratified by number of operating beds into 3 groups (group I, <250 beds; group II, 250-450 beds; group III, >450 beds). Total (diagnosis-related groups 106 + 107) annual coronary bypass cases per hospital varied from 271 to 913 (mean 532). Univariate and multivariable analyses were used to study the relationship between the direct and total cost and a number of patient (age, sex, acuity class, payer) and hospital (bed capacity, annual case volume per diagnosis-related group, cardiothoracic residency) predictor variables. For each hospital, we also studied the relationship between changes in coronary bypass case volume and the corresponding changes in average cost from 1995 to 1996.

RESULTS

Scatterplots revealed a broad range of mean direct cost of coronary bypass surgery among hospitals with comparable case volumes. When annual cases were analyzed as continuous variables, there was no linear relationship of case volume with direct or total cost of coronary bypass (r = -0.05 to +0.08) for any diagnosis-related group or year. When hospital bed capacity and case volume were grouped into strata and studied by analysis of variance, there was no evidence of an inverse relationship between these variables and cost. In multivariable analysis, patient acuity class and diagnosis-related group were the most important predictors of cost. Beds and case volume met inclusion criteria for most models but added little to the "explanation" of variability R(2), often less than 1%. Finally, substantial interhospital differences were noted in the magnitude and direction (direct vs inverse) of their 1995 to 1996 change in volume versus change in cost.

CONCLUSIONS

Within the range of hospital size and case volume represented in this study, there is no evidence that either variable is related to the cost of performing coronary bypass surgery. Massachusetts hospitals appear to function on different segments of different average cost curves. It is not possible to predict the relative cost of coronary bypass grafting at a given hospital based primarily on volume.

Laryngeal cancer cost analysis: association of case-mix and treatment characteristics with medical charges

OBJECTIVES

To examine the relationship of various pretreatment case-mix characteristics and treatment modalities with medical charges incurred during diagnosis, treatment, and 2-year follow-up for patients with laryngeal cancer.

DESIGN

Retrospective chart review and billing record analysis.

METHODS

The charts and billing records of patients diagnosed with laryngeal cancer at the University of Iowa Hospitals and Clinics (UIHC) between January 1, 1991 and December 31, 1994 were reviewed. The independent variables included various pretreatment patient-mix and tumor characteristics (age, AJCC TNM clinical stage, smoking history, ASA class, and comorbidity as defined by Kaplan-Feinstein grade) as well as type of treatment. The dependent variables included total physician, office, and university hospital-based charges incurred during the pretreatment evaluation and 0- to 3-, 3- to 12, and 12- to 24-month billing periods after the initiation of cancer-directed therapy. Total 1-year and 2-year charges were also evaluated. Univariate and multivariate analyses were used to investigate the relationships between dependent and independent variables and to develop models predictive of management charges during the individual and total billing periods.

RESULTS

Pretreatment charges showed no significant associations (P < .05) with any of the independent variables. Multiple regression analyses indicated that comorbidity, stage, and initial treatment modality were significant variables in one or more of the models predicting charges incurred during the 0- to 3-month, 3- to 12-month, total 1-year, and total 2-year billing periods. The models yielded R2 values for the total 1- and 2-year billing periods of 0.5246 and 0.5055, respectively.

CONCLUSIONS

This work supports continued study of measures that may result in earlier detection of laryngeal cancer as a potential means of reducing management charges. These results also indicate that a more accurate method of stratifying the disease severity of laryngeal cancer patients for reimbursement purposes would include measurements of the severity of the index disease as well as comorbid diseases.

[Chronic comorbidity and homogeneity in diagnostic related groups]

OBJECTIVE

[corrected] One of the ways to compare the efficiency of different hospitals and services is to evaluate Diagnostic Related Groups (DRGs), with the hypothesis that patients in the same RDG will present homogeneous behavior with respect to length of stay. The object of this study was to evaluate in the context os the National Health System the internal variability of specific DRGs in terms of the patients' comorbidity.

METHODS

On the basis of various comorbidity scores measured with the Charlson index (ChI), we analyzed length of stay, inhospital mortality and emergency readmissions at 30 and 365 days in 106.673 hospitalizations (excluding subjects younger than 17 years of age, and obstetrics and psychiatric patients) in 12 hospitals, and in 17 DRGs selected on the basis of their greater frequency and comorbidity.

RESULTS

In the aggregated analysis, length of stay (from 8.5 days in patients with no comorbidity to 17.0 days in patients with scores higher than 4) and inhospital mortality rates (from 3.7% in patients with no comorbidity to 17.6% in patients with highest score) increased significantly with each level of the Charlson index. The readmission rate at 30 days rose from 4.7% to 10.9% also in step with increases in comorbidity scores. Readmissions at one year varied from 14.8% in patients with scores of 0 to 35.2% in patients with scores of 3-4, and dropped to 27.9% in patients with scores higher than 4. When analysing different DRGs, 8 of the 17 groups studied showed a significantly higher length of stay with increased comorbidity scores. Some DRGs also showed intra-group variability with respect to mortality and readmission, particularly at 365 days.

CONCLUSIONS

Some DRGs show significant internal variability in terms of comorbidity that may be generating a false worse evaluation of the efficiency of hospitals that treat patients with higher comorbidity.

A comparative study of the costliness of Manitoba hospitals

OBJECTIVES

In light of ongoing discussions about health care policy, this study offered a method of calculating costs at Manitoba hospitals that compared relative costliness of inpatient care provided in each hospital.

RESEARCH DESIGN

This methodology also allowed comparisons across types of hospitals-teaching, community, major rural, intermediate and small rural, as well as northern isolated facilities.

MEASURES

Data used in this project include basic hospital information, both financial and statistical, for each of the Manitoba hospitals, hospital charge information by case from the State of Maryland, and hospital discharge abstract information for Manitoba. The data from Maryland were used to create relative cost weights (RCWs) for refined diagnostic related groups (RDRGs) and were subsequently adjusted for Manitoba length of stay. These case weights were then applied to cases in Manitoba hospitals, and several other adjustments were made for nontypical cases. This case mix system allows cost comparisons across hospitals.

RESULTS

In general, hospital case mix costing demonstrated variability in hospital costliness, not only across types of hospitals but also within hospitals of the same type and size.

CONCLUSIONS

Costs at the teaching hospitals were found to be considerably higher than the average, even after accounting for acuity and case mix.

Modeling of high-cost patient distribution within renal failure diagnosis related group

Modeling by mixed-distribution was proposed in order to analyze heterogeneity of costs and length of stays within Diagnosis Related Groups (DRGs). A mixed-distribution model based on Weibull distributions was applied to 791 discharge abstracts of French DRG no. 450 (Health Care Financing Administration 3 DRG no. 316 "Renal failure") from a national database. Three subgroups of cost and length of stay were identified. Except for age, clinical criteria significantly linked with the long-stay subgroup were the same as those associated with the high-cost subgroup: acute renal failure, intensive care, infectious complications, and vascular investigations. The identification of factors associated with high costs, based on the proposed model, will allow physicians to understand more accurately how their choice of specific procedures influences hospital costs.

The ICD-9-based illness severity score: a new model that outperforms both DRG and APR-DRG as predictors of survival and resource utilization

OBJECTIVE

This project is designed to develop and validate a predictive model that is a useful benchmarking and quality of care assessment tool based on International Classification of Diseases, Ninth Revision (ICD-9), diagnoses and procedures. This model, the ICD-9-Based Illness Severity Score (ICISS), was developed from the Agency for Health Care Policy Research's Health Care Utilization Project database and is used to predict hospital survival, hospital length of stay, and hospital charges of injured patients admitted to University of North Carolina Hospitals. The study also compared the outcome predictions of ICISS with those of the long-established diagnosis-related groups (DRG) and the 3M product APR-DRG systems.

METHODS

We performed a retrospective study of 9,483 trauma patients at University of North Carolina Hospitals. A model was developed to predict survival, length of stay, and hospital charges. The accuracy of the model of survival was assessed using the area under the receiver-operating characteristics curve; the adjusted R2 statistic was used to judge the proportion of variation described by the models of length of stay and hospital charges.

RESULTS

ICISS proved to be superior to both DRG and APR-DRG in predicting survival of trauma patients: the area under the receiver-operating characteristics curve for prediction of hospital survival was 0.957 for ICISS, 0.707 for DRG, and 0.808 for APR-DRG. ICISS also outperformed DRG and APR-DRG in predicting hospital length of stay and hospital charges: the adjusted R2 for the ICISS length of stay model was 0.57, compared with the DRG length of stay model with adjusted R2 of 0.31 and the APR-DRG length of stay model with adjusted R2 of 0.35. The adjusted R2 for the ICISS hospital charges model was 0.67, compared with the DRG and APR-DRG hospital charges model R2 of 0.46 and 0.51, respectively (p < 0.001 in all cases).

CONCLUSION

This study demonstrates that an ICD-9-based predictive model (ICISS) can markedly outperform both DRG and APR-DRG as a predictor of survival, hospital length of stay, and hospital charges.

Fitting the distributions of length of stay by parametric models

OBJECTIVES

The purpose of this study was to assess the adequacy of three widely used models--Lognormal, Weibull, and Gamma--for describing the distribution of length of stay. This is a fundamental step in the development of outliers resistant (robust) methods for the statistical analysis of this kind of data, where the main objective is to determine measures of average and total resource consumption of groups of patients. Current practice uses several types of trimming rules, many of which are based on the Lognormal model, although theoretical and experimental bases are still insufficient.

METHODS

The three models were adjusted using robust procedures based on M-estimators to approximately 5 million stays grouped by Diagnosis-Related Groups (DRGs). The resulting 3,279 samples were collected in five European countries during 3 years.

RESULTS

Most of the distributions observed could be fitted with one of these models. The descriptions provided by the Gamma and the Weibull models were similar, and the Gamma model could be omitted. The casemix description provided by the Log-normal-Weibull family was, for certain countries, significantly better than the one provided by the single Lognormal model. Often, for a given DRG and a given country, length of stay distributions could be described with the same model during several years. A given DRG, however, usually had to be described by means of different models for different countries.

CONCLUSIONS

Practical and conceptual consequences of the results are discussed. They can be extended to the analyses of other consumption variables used in health services. Statistical procedures for casemix description, including current rules of trimming, should be improved by means of more flexible families of models.

Cost-identification analysis in oral cavity cancer management

The objectives of this study were to investigate potential relationships between pretreatment patient-mix characteristics, treatment modalities, and costs generated during the pretreatment work-up, treatment, and 1-year follow-up periods for patients with oral cavity cancer (OCC). Another objective was to identify potential areas for cost reduction and improved resource allocation in the management of OCC patients. Using a retrospective cohort of 73 patients with OCC, pretreatment patient-mix characteristics and treatment modalities were evaluated in relation to university-based charges incurred during the pretreatment evaluation, treatment, and 1-year follow-up periods. Simple regression and stepwise multiple regression analyses were used to develop predictive models for cost based on independent variables, including age, AJCC TNM clinical stage, smoking history, American Society of Anesthesiologists (ASA) class, comorbidity as defined by the Kaplan-Feinstein grade and treatment modality. The dependent measurements included all physician, office, and hospital charges incurred at the University of Iowa Hospitals and Clinics during the pretreatment evaluation, treatment, and follow-up periods, as well as the total pretreatment through 1-year follow-up management costs. Independent variables that were identified as being significantly associated with treatment costs included T classification, N classification, TNM stage, unimodality versus multimodality treatment, and the Kaplan-Feinstein comorbidity grade. Age, smoking status, and ASA class were not significantly associated with costs. The majority of the OCC management costs were incurred during the treatment period. The most substantial decreases in management costs for OCC will be realized through measures that allow identification and treatment of disease at an early stage, in which single-modality treatment may effectively be used. Resource allocation for OCC should support the investigation of measures through which the diagnosis and treatment of OCC at the earliest possible stage is facilitated. The presence of comorbid illness is a significant component in the determination of management costs for OCC and should be included in analyses of resource allocation for OCC. The singular diagnosis of OCC encompasses a wide range of patient illness severity, and diagnosis-related reimbursement schemes for OCC treatment should optimally differentiate between early and advanced stage disease.

Case-mix adjustment using administrative databases: a paradigm to guide future research

One of the most persistent problems in the field of quality assessment remains how to remove the confounding effect of different institutions providing care to patients with dissimilar severity of illness and case complexity. The authors review the literature to determine whether risk adjustment systems based on administrative data are inherently inferior to systems that depend on primary data collection and conclude that they are not. In light of the potential competence of risk adjustment systems based on administrative data, the authors identify those systems that are best supported by theory and evidence. Data elements that have been found most explanatory of medical outcomes are also identified. On the basis of an evaluation of the performance of various risk adjustment approaches, the authors propose a paradigm that could serve to unify and direct future studies.

Judging hospitals by severity-adjusted mortality rates: the influence of the severity-adjustment method

OBJECTIVES

This research examined whether judgments about a hospital's risk-adjusted mortality performance are affected by the severity-adjustment method.

METHODS

Data came from 100 acute care hospitals nationwide and 11880 adults admitted in 1991 for acute myocardial infarction. Ten severity measures were used in separate multivariable logistic models predicting in-hospital death. Observed-to-expected death rates and z scores were calculated with each severity measure for each hospital.

RESULTS

Unadjusted mortality rates for the 100 hospitals ranged from 4.8% to 26.4%. For 32 hospitals, observed mortality rates differed significantly from expected rates for 1 or more, but not for all 10, severity measures. Agreement between pairs of severity measures on whether hospitals were flagged as statistical mortality outliers ranged from fair to good. Severity measures based on medical records frequently disagreed with measures based on discharge abstracts.

CONCLUSIONS

Although the 10 severity measures agreed about relative hospital performance more often than would be expected by chance, assessments of individual hospital mortality rates varied by different severity-adjustment methods.

Does severity explain differences in hospital length of stay for pneumonia patients?

OBJECTIVES

In the USA, the role of patient severity in determining hospital resource use has been questioned since Medicare adopted prospective hospital payment based on diagnosis-related groups (DRGs). Exactly how to measure severity, however, remains unclear. We examined whether assessments of severity-adjusted hospital lengths of stay (LOS) varied when different measures were used for severity adjustment.

METHODS

The complete study sample included 18,016 patients receiving medical treatment for pneumonia at 105 acute care hospitals. We studied 11 severity measures, nine based on patient demographic and diagnosis and procedure code information and two derived from clinical findings from the medical record. For each severity measure, LOS was regressed on patient age, sex, DRG, and severity score. Analyses were performed on trimmed and untrimmed data. Trimming eliminated cases with LOS more than three standard deviations from the mean on a log scale.

RESULTS

The trimmed data set contained 17,976 admissions with a mean (S.D.) LOS of 8.9 (6.1) days. Average LOS ranged from 5.0-11.8 days among the 105 hospitals. Using trimmed data, the 11 severity measures produced R-squared values ranging from 0.098-0.169 for explaining LOS for individual patients. Across all severity measures, predicted average hospital LOS varied much less than the observed LOS, with predicted mean hospital LOS ranging from about 8.4-9.8 days.

DISCUSSION

No severity measure explained the two-fold differences among hospitals in average LOS. Other patient characteristics, practice patterns, or institutional factors may cause the wide differences across hospitals in LOS.