(continued)
Title: Surgical and Nonsurgical management of sciatica secondary to a lumbar disc herniation.
Publication: Spine 2001;26(10):1179-87.
Authors: Atlas SJ.
Grant No.: HS06344/HS08194/HS09804.
Principal Investigator: Richard Deyo/Richard Deyo/Robert Keller.
Project Officer: Mary Cummings.
Abstract: STUDY DESIGN: A prospective cohort study. OBJECTIVE: To assess 5-year outcomes for patients with sciatica caused by a lumbar disc herniation treated surgically or nonsurgically. SUMMARY OF BACKGROUND DATA: There is limited knowledge about long-term treatment outcomes of sciatica caused by a lumbar disc herniation, particularly the relative benefits of surgical and conservative therapy in contemporary clinical practice. METHODS: Eligible, consenting patients recruited from the practices of orthopedic surgeons, neurosurgeons, and occupational medicine physicians throughout Maine had baseline interviews with mailed followup questionnaires at 3, 6, and 12 months and annually thereafter. Clinical data were obtained at baseline from a physician questionnaire. Outcomes included patient-reported symptoms of leg and back pain, functional status, satisfaction, and employment and compensation status. RESULTS: Of 507 patients initially enrolled, 5-year outcomes were available for 402 (79.3 percent) patients: 220 (80 percent) treated surgically and 182 (78.4 percent) treated nonsurgically. Surgically treated patients had worse baseline symptoms and functional status than those initially treated nonsurgically. By 5 years 19 percent of surgical patients had undergone at least one additional lumbar spine operation, and 16 percent of nonsurgical patients had opted for at least one lumbar spine operation. Overall, patients treated initially with surgery reported better outcomes. At the 5-year followup, 70 percent of patients initially treated surgically reported improvement in their predominant symptom (back or leg pain) versus 56 percent of those initially treated nonsurgically (P < 0.001). Similarly, a larger proportion of surgical patients reported satisfaction with their current status (63 percent vs. 46 percent, P < 0.001). These differences persisted after adjustment for other determinants of outcome. The relative advantage of surgery was greatest early in followup and narrowed over 5 years. There was no difference in the proportion of patients receiving disability compensation at the 5-year followup. The least symptomatic patients at baseline did well regardless of initial treatment, although function improved more in the surgical group. CONCLUSIONS: For patients with moderate or severe sciatica, surgical treatment was associated with greater improvement than nonsurgical treatment at 5 years. However, patients treated surgically were as likely to be receiving disability compensation, and the relative benefit of surgery decreased over time.
Title: Low back pain.
Publication: New England Journal of Medicine 2001 Feb;344(5):363-70.
Authors: Deyo RA et al.
Grant No.: HS09804.
Principal Investigator: Robert Keller.
Project Officer: Mary Cummings.
Abstract: For patients with nonspecific low back pain, a precise pathoanatomical diagnosis is often impossible, which leads to various imprecise diagnoses (e.g., sprain or strain). The natural history of low back pain if favorable, and patients need this reassurance. The favorable natural history may partly explain the proliferation of unproved treatments that may seem to be effective. The use of plain radiography can be limited to patients with clinical findings suggestive of underlying systemic disease, and more advanced imaging can be reserved for potential candidates for surgery. The role of imaging in other situations is limited because of the poor association between symptoms and anatomical findings. Bed rest is not recommended for the treatment of low back pain or sciatica, and a rapid return to normal activities is usually the best course. Back exercises are not useful for the acute phase but help to prevent recurrences and treat chronic pain. Surgery is appropriate for a small proportion of patients with low back symptoms; it is most successful for those with sciatica or pseudoclaudication that persists after nonsurgical therapy has been tried.
Title: Use of plain radiography to screen for cervical spine injuries.
Publication: Annals of Emergency Medicine 2001 Jul;38(1):1-7.
Authors: Mower, WR et al.
Grant Number: HS08239.
Principal Investigator: William Mower.
Project Officer: Lynn Bosco.
Abstract: STUDY OBJECTIVE: Standard radiographic screening may fail to reveal any evidence of injury in some patients with spinal injury. The purposes of this investigation were to document the efficacy of standard radiographic views and to categorize the frequencies and types of injuries missed on plain radiographic screening of the cervical spine. METHODS: All patients with blunt trauma selected for radiographic cervical spine imaging at 21 participating institutions underwent a standard 3-view series (cross-table lateral, anteroposterior, and odontoid views), as well as any other imaging deemed necessary by their physicians. Injuries detected with screening radiography were then compared with final injury status for each patient, as determined by review of all radiographic studies. RESULTS: The study enrolled 34,069 patients with blunt trauma, including 818 patients (2.40 percent of all patients; 95 percent confidence interval [CI] 2.40 percent to 2.40 percent) having a total of 1,496 distinct cervical spine injuries. Plain radiographs revealed 932 injuries in 498 patients (1.46 percent of all patients; 95 percent CI 1.46 percent to 1.46 percent) but missed 564 injuries in 320 patients (0.94 percent of all patients; 95 percent CI 0.94 percent to 0.94 percent). The majority of missed injuries (436 injuries in 237 patients [representing 0.80 percent of all patients]; 95 percent CI 0.80 percent to 0.80 percent) occurred in cases in which plain radiographs were interpreted as abnormal (but not diagnostic of injury) or inadequate. However, 23 patients (0.07 percent of all patients; 95 percent CI 0.05 percent to 0.09 percent) had 35 injuries (including 3 potentially unstable injuries) that were not visualized on adequate plain film imaging. These patients represent 2.81 percent (95 percent CI 1.89 percent to 3.63 percent) of all injured patients with blunt trauma undergoing radiographic evaluation. CONCLUSION: Standard 3-view imaging provides reliable screening for most patients with blunt trauma. However, on rare occasions, such imaging may fail to detect significant unstable injuries. In addition, it is difficult to obtain adequate plain radiographic imaging in a substantial minority of patients.
Title: Test performance of the individual Nexus "low-risk" clinical screening criteria for cervical spine injury.
Publication: Annals of Emergency Medicine 2001 Jul;38(1):22-5.
Authors: Panacek EA et al.
Grant No.: HS08239.
Principal Investigator: William Mower.
Project Officer: Lynn Bosco.
Abstract: BACKGROUND: The National Emergency X-Radiography Utilization Study (NEXUS) recently validated the ability of a decision instrument to define a population with an extremely low risk of cervical spine injury (CSI) after blunt trauma. It is unclear whether each of the five individual criteria is necessary for the decision instrument to maintain its high sensitivity. METHODS: NEXUS was a prospective observational study at 21 emergency departments, which enrolled all patients with blunt trauma for whom cervical spine radiographs were ordered. In this substudy, we examined the NEXUS database to determine the contribution of each of the five individual low-risk clinical criteria to the overall sensitivity of the decision instrument. RESULTS: All but 8 of 818 patients with CSI, and all but 2 of 578 patients with significant CSI, were identified by using the decision instrument. A substantial number of patients with CSI (236/818 [29 percent]) and patients with significant CSI (175/578 [30 percent]) met only 1 of the 5 nonlow-risk criteria, and each of the 5 criteria was the only indicator of nonlow-risk status in at least 8 patients with CSI and at least 5 patients with significant CSI. CONCLUSION: Because each of the 5 low-risk criteria was the only marker of nonlow-risk status in at least a few patients with significant CSI, modification of the overall NEXUS decision instrument by eliminating any one of the criteria would markedly reduce sensitivity and make the instrument unacceptable for clinical use.
Title: Use of Flexion-Extension radiographs of the cervical spine in blunt trauma.
Publication: Annals of Emergency Medicine 2001 Jul;38(1):8-11.
Authors: Pollack CV et al.
Grant No.: HS08239.
Principal Investigator: William Mowe.r
Project Officer: Lynn Bosco.
Abstract: STUDY OBJECTIVE: Flexion-extension (F/E) radiographs of the cervical spine are often used in patients with blunt trauma when the evaluating physician remains concerned about bony or ligamentous injuries despite negative or nondiagnostic standard radiographs. The use of this approach has never been addressed in a large prospective study. We sought to determine the clinical factors associated with ordering F/E views and the incidence of diagnostic F/E films in patients with a normal 3-view cervical spine series. METHODS: Patients with blunt trauma selected for radiographic cervical spine imaging at 21 participating institutions in the National Emergency X-Radiography Utilization Study project underwent standard 3-view (cross-table lateral, anteroposterior, and odontoid views) series, as well as any other imaging deemed necessary by their physicians. Injuries detected by means of screening radiography were then compared with final injury status for each patient, as determined by review of all radiographic studies. Patients who underwent F/E views were analyzed separately. RESULTS: Of 818 patients ultimately found to have cervical spine injury, 86 (10.5 percent) underwent F/E testing. Two patients sustained stable bony injuries detected only on F/E views. Four other patients had a subluxation detected only on F/E views, but all had other injuries apparent on routine cervical spine imaging. CONCLUSION: F/E imaging adds little to the acute evaluation of patients with blunt trauma. Other approaches, including magnetic resonance imaging, computed tomography, or delayed F/E, in the presence of specific clinical concerns would seem to provide a more reasonable approach to adjunctive imaging.
Title: Lessons from a trial of acupuncture and massage for low back pain: patient expectations and treatment effects.
Publication: Spine 2001 Jul 1;26(13):1418-24.
Authors: Kalauokalani D et al.
Grant No.: HS09351.
Principal Investigator: David Cherkin.
Project Officer: Mary Cummings.
Abstract: STUDY DESIGN: A subanalysis of data derived from a randomized clinical trial was performed. OBJECTIVE: To evaluate the association of a patient's expectation for benefit from a specific treatment with improved functional outcome. SUMMARY OF BACKGROUND DATA: Psychosocial factors, ambiguous diagnoses, and lack of a clearly superior treatment have complicated the management of patients with chronic low back pain. The authors hypothesized that patient expectation for benefit from a specific treatment is associated with improved functional outcomes when that treatment is administered. METHODS: In a randomized trial, 135 patients with chronic low back pain who received acupuncture or massage were studied. Before randomization, study participants were asked to describe their expectations regarding the helpfulness of each treatment on a scale of 0 to 10. The primary outcome was level of function at 10 weeks as measured by the modified Roland Disability scale. RESULTS: After adjustment for baseline characteristics, improved function was observed for 86 percent of the participants with higher expectations for the treatment they received, as compared with 68 percent of those with lower expectations (P = 0.01). Furthermore, patients who expected greater benefit from massage than from acupuncture were more likely to experience better outcomes with massage than with acupuncture, and vice versa (P = 0.03). CONCLUSIONS: The results of this study suggest that patient expectations may influence clinical outcome independently of the treatment itself. In contrast, general optimism about treatment, divorced from a specific treatment, is not strongly associated with outcome. These results may have important implications for clinical trial design and recruitment, and may help to explain the apparent success of some conventional and alternative therapies in trials that do not control for patient expectations. The findings also may be important for therapy choices made in the clinical setting.
Title: Acupuncture for chronic low back pain: diagnosis and treatment patterns among acupuncture evaluating the same patient.
Publication: South Medical Journal 2001 May;94(5):486-92.
Authors: Kalauokalani D et al.
Grant No.: HS09989/HS09565/HS09351.
Principal Investigator: David Cherkin.
Project Officer: Cummings.
Abstract: BACKGROUND: There is increasing need to examine the effectiveness of acupuncture and other alternative therapies for common conditions. However, little attention has focused on the variability in acupuncturists' assessment, diagnosis, and treatment patterns. METHODS: Seven office-based acupuncturists practicing Traditional Chinese Medicine evaluated the same patient with chronic low back pain and provided data regarding principal assessment techniques, diagnoses, and therapeutic recommendations. RESULTS: A high diagnostic agreement existed among five of seven acupuncturists. However, recommended treatments included varying numbers and locations of acupuncture points. Recommendations varied between 5 and 14 points requiring 7 to 26 needles, since many points were intended for bilateral application. Of 28 acupuncture points selected, only 4 (14 percent) were prescribed by two or more acupuncturists. Most recommended various forms of adjuvant heat. CONCLUSIONS: Seven acupuncturists agreed considerably in the diagnoses for the same patient with chronic low back pain, but treatment recommendations varied substantially. Clinicians and researchers must recognize treatment recommendation variations and the challenges they present for study design and interpretation.
Title: Distribution and patterns of blunt traumatic cervical spine injury.
Publication: Annals of Emergency Medicine 2001 Jul; 38(1):17-21.
Authors: Goldberg W et al.
Grant No.: HS08239.
Principal Investigator: William Mower.
Project Officer: Lynn Bosco.
Abstract: STUDY OBJECTIVE: Previous studies of cervical spine injury involve individual institutions or special populations. There is currently little reliable information regarding natural cervical spine injury patterns after blunt trauma. This substudy of the National Emergency X-Radiography Utilization Study project was designed to accurately assess the prevalence, spectrum, and distribution of cervical spine injury after blunt trauma. METHODS: We prospectively enrolled all patients with blunt trauma undergoing cervical spine radiography at 21 diverse institutions. Injury status was determined by review of all radiographic studies obtained on each patient. For each individual injury, we recorded which specific films revealed the injury, the level and location of injury on each vertebra, and the age and sex of the patient. RESULTS: Of 34,069 enrolled patients with blunt trauma, 818 (2.4 percent) individuals had a total of 1,496 distinct cervical spine injuries to 1,285 different cervical spine structures. The second cervical vertebra was the most common level of injury (286 [24.0 percent] fractures, including 92 odontoid fractures), and 470 (39.3 percent) fractures occurred in the 2 lowest cervical vertebrae (C6 and C7). The vertebral body, injured in 235 patients, was the most frequent site of fracture. Nearly one third of all injuries (29.3 percent) were considered clinically insignificant. CONCLUSION: Cervical spine injuries occur in a small minority of patients with blunt trauma who undergo imaging. The atlantoaxial region is the most common site of injury, and the sixth and seventh vertebrae are involved in over one third of all injuries. Other spine levels are much more commonly involved than has previously been appreciated. A substantial minority of radiographically defined cervical spine injuries are of little clinical importance.
Title: Epidemiology of cervical spine injury victims.
Publication: Annals of Emergency Medicine 2001 Jul;38(1):12-6.
Authors: Lowery DW et al.
Grant No.: HS08239.
Principal Investigator: William Mower.
Project Officer: Lynn Bosco.
Abstract: STUDY OBJECTIVE: We sought to characterize demographics and injury patterns among patients undergoing emergency department cervical spine radiography for blunt traumatic injury. METHODS: All patients with blunt trauma undergoing cervical spine radiography at 21 centers were enrolled in this prospective, observational study. Patients' date of birth, age, sex, and ethnicity were noted before cervical spine radiography. RESULTS: Demographic factors associated with cervical spine injury, present in 818 of 33,922 patients, included the following: age of 65 years or older (relative risk [RR] 2.09; 95 percent confidence interval [CI] 1.77 to 2.59); "other" ethnicity (RR 1.79, 95 percent CI 1.46 to 2.19); male sex (RR 1.72, 95 percent CI 1.48 to 2.00); and white ethnicity (RR 1.50, 95 percent CI 1.31 to 1.72). Hispanic ethnicity (RR 0.64, 95 percent CI 0.51 to 0.79), female sex (RR 0.58, 95 percent CI 0.50 to 0.67), black ethnicity (RR 0.55, 95 percent CI 0.45 to 0.66), and age of less than 18 years (RR 0.39, 95 percent CI 0.27 to 0.55) were associated with reduced risk of cervical spine injury. CONCLUSION: Among patients undergoing ED cervical spine radiography, cervical spine injury is more common among the elderly, male subjects, and patients of white or "other" ethnicity. Because cervical spine injury occurs in patients in all demographic categories, however, this information cannot be used to select individual patients who should or should not undergo imaging.
Title: Constructs of burden of illness in older patients with breast cancer: A comparison of measurement methods.
Publication: Health Serv Res 2001 Dec;36(6 Pt 1):1085-107.
Authors: Mandelblatt JS et al.
Grant No.: HS08395.
Principal Investigator: Jeanne Mandelblatt.
Project Officer: Mary Cummings.
Abstract: OBJECTIVE. The burden of illness can influence treatment decisions, but there are limited data comparing the performance of different illness burden measures. We assessed the correlations between five previously validated measures of illness burden and global health and physical function and evaluated how each measure correlates with breast cancer treatment patterns in older women. DATA SOURCE: A cohort of 718 women > 67 years with early-stage breast cancer formed the study group. STUDY DESIGN/DATA COLLECTION METHODS: The study made a cross-sectional comparison of illness burden measures (Charlson index, Index of Co-existent Diseases, cardiopulmonary burden of illness, patient-specific life expectancy, and disease counts) and physical function and self-rated global health status. Data were collected from records and patient interviews. PRINCIPAL FINDINGS: All of the measures were significantly correlated with each other and with physical function and self-rated health (p < .001). After controlling for age and stage, life expectancy had the largest effect on surgical treatment, followed by self-rated physical function and health; life expectancy was also independent of physical function. For instance, women with higher life expectancy and better self-rated physical function and health were more likely to receive breast conservation and radiation than sicker women. Women with higher physical functioning were more likely to receive adjuvant chemotherapy than women with lower functioning. CONCLUSIONS: Several measures of illness burden were associated with breast cancer therapy, but each measure accounted for only a small amount of variance in treatment patterns. Future work is needed to develop and validate measures of burden of illness that are feasible, comprehensive, and relevant for diverse clinical and health services objectives.
Title: Measuring and Predicting Surgeons' Practice Style for Breast Cancer Treatment in Older Women.
Publication: Medical Care 2001;39(3):228-42.
Authors: Mandelblatt JS et al.
Grant Number: HS08395.
Principal Investigator: Jeanne Mandelblatt.
Project Officer: Mary Cummings.
Abstract: BACKGROUND: Few measures exist to assess physicians' practice style, and there are few data on physicians' practice styles and patterns of care. OBJECTIVES: To use clinical vignettes to measure surgeons' "propensity" for local treatments for early-stage breast cancer and to describe factors associated with propensity. RESEARCH DESIGN AND SUBJECTS: A cross-sectional mailed survey with telephone followup of a random sample of 1,000 surgeons treating Medicare beneficiaries in fee-for-service settings. MEASURES: Outcome measures include treatment propensity, self-reported practice, and actual treatment received by the surgeons' patients. RESULTS: Propensities were significantly associated with actual treatment, controlling for covariates. Area Medicare fees were the strongest predictor of propensity, followed by region, attitudes, volume, and gender. For instance, after other factors were considered, surgeons practicing in areas with the highest breast-conserving surgery (BCS) fees were 8.61 (95 percent CI 2.26-32.73) times more likely to have a BCS propensity than surgeons in areas with the lowest fees. Surgeons with the strongest beliefs in patient participation in treatment decisions were nearly 6 times (95 percent CI 1.67-20.84) more likely to have a BCS propensity than surgeons with the lowest such beliefs, controlling for covariates. Male surgeons were also independently more likely to have a mastectomy propensity than female surgeons. CONCLUSIONS: Surgeons' propensities explain some of the observed variations in breast cancer treatment patterns among older women. Standardized scenarios provide a practical method to measure practice style and could be used to evaluate physician contributions to shared decision making, practice patterns, costs and outcomes, and adherence to guidelines.
Title: Medicare fees and small area variations in breast-conserving surgery among elderly women.
Publication: Medical Care Research and Review Sep 2001;58(3):334-60.
Authors: Hadley J et al.
Grant No.: HS08395.
Principal Investigator: Jeanne Mandelblatt.
Project Officer: Mary Cummings.
Abstract: This study used data from Medicare files, the American Hospital Association's Annual Survey of Hospitals, and the 1990 census to investigate whether Medicare fees for breast-conserving surgery (BCS) and mastectomy (MST) affected the rate of BCS across 799 3-digit ZIP code areas in 1994. The full model, which was based on the conceptual framework of the supply of and demand for different treatments, explained 51 percent of the variation in BCS rates. Medicare fees were statistically significant and had the hypothesized effects: a 10 percent higher BCS fee was associated with a 7 to 10 percent higher BCS rate, while a 10 percent higher MST fee was associated with a 2 to 3 percent lower proportion receiving BCS. Other significant economic variables were proximity to a radiation therapy hospital, a teaching hospital or a cancer center, and the percentage of elderly women with incomes below the poverty rate, which were negatively related to the BCS rate. Variations in age, race, and metropolitan populations had small or insignificant effects. The single most important variable was the percentage of cases with one or more comorbidities.
Title: Does enrollment in a randomized clinical trial lead to a higher cost of routine care?
Publication: Am Heart J 2002 Jan;143(1):140-4.
Authors: Weiss JP et al.
Grant No.: HS08362.
Principal Investigator: Mark Hlatky.
Project Officer: Daniel Stryer.
Abstract: BACKGROUND: Reimbursement for the routine care of patients enrolled in clinical trials is controversial. Our objective was to determine the added medical costs, if any, associated with enrollment in a randomized clinical trial. METHODS: We analyzed data from the Myocardial Infarction Triage and Intervention (MITI) Trial (1988-1991) and the registry of all patients admitted to 19 Seattle area coronary care units (1988-1993). The major trial entry criteria were age 35 to 71 years, symptom duration 15 minutes to 6 hours, and acute myocardial infarction on electrocardiogram. The trial group consisted of 264 of 324 randomized patients who received thrombolytics and had available cost data. From 11,932 registry patients, we identified a control group who met trial entry criteria but who were not enrolled because of logistic barriers or presentation outside the trial enrollment period, 335 of whom received thrombolytics and had available cost data. The groups were compared for total cost for initial hospitalization, with and without multivariable adjustment for baseline characteristics. RESULTS: Total hospital cost was not different between trial patients (median $11,516) and control subjects (median $14,200)(trial/control mean cost ratio 0.91 [95percent CI 0.82-1.02]). Participation in the trial had an insignificant effect on costs in the multivariable model (cost ratio 1.04, 95percent CI 0.95-1.16). Significant predictors of cost included hospital of admission, length of stay, and coronary revascularization procedures. CONCLUSION: Participation in the MITI randomized trial had no effect on the cost of routine care.
Title: Potential cost-effectiveness of prophylactic use of the implantable cardioverter defibrillator or amiodarone after myocardial infarction.
Publication: Annals of Internal Medicine 2001 Nov 20;135(10):870-83.
Authors: Sanders GD et al.
Grant No.: HS08362.
Principe Investigator: Mark Hlatky.
Project Officer: Daniel Stryer.
Abstract: BACKGROUND: Clinical trials have shown that implantable cardioverter defibrillators (ICDs) improve survival in patients with sustained ventricular arrhythmias. OBJECTIVE: To determine the efficacy necessary to make prophylactic ICD or amiodarone therapy cost-effective in patients with myocardial infarction. DESIGN: Markov model-based cost utility analysis. DATA SOURCES: Survival, cardiac death, and inpatient costs were estimated on the basis of the Myocardial Infarction Triage and Intervention registry. Other data were derived from the literature. TARGET POPULATION: Patients with past myocardial infarction who did not have sustained ventricular arrhythmia. TIME HORIZON: Lifetime. PERSPECTIVE: Societal. INTERVENTIONS: ICD or amiodarone compared with no treatment. OUTCOME MEASURES: Life-years, quality-adjusted life-years (QALYs), costs, number needed to treat, and incremental cost-effectiveness. RESULTS OF BASE-CASE ANALYSIS: Compared with no treatment, ICD use led to the greatest QALYs and the highest expenditures. Amiodarone use resulted in intermediate QALYs and costs. To obtain acceptable cost-effectiveness thresholds (</=$75,000/QALY), ICDs had to reduce arrhythmic death by 50percent and amiodarone had to reduce total death by 7percent in patients with depressed ejection fraction. RESULTS OF SENSITIVITY ANALYSIS: For moderate efficacies, in patients with ejection fractions less than or equal to 0.3, 0.31 to 0.4, and greater than 0.4, the cost-effectiveness of amiodarone compared with no therapy was $43,100/QALY, $66,500/QALY, and $132,500/QALY, respectively, and the cost-effectiveness of ICD compared with amiodarone was $71,800/QALY, $195,700/QALY, and $557,900/QALY, respectively. CONCLUSIONS: Use of ICD or amiodarone in patients with past myocardial infarction and severely depressed left ventricular function may provide substantial clinical benefit at an acceptable cost. These results highlight the importance of clinical trials of ICDs in patients with low ejection fractions who have had myocardial infarction.
Title: Comparison of use of medications after acute myocardial infarction in the Veterans Health Administration and Medicare.
Publication: Circulation 2001 Dec 11;104(24): 2898-904.
Authors: Petersen LA et al.
Grant No.: HS08071.
Principal Investigator: Barbara McNeil.
Project Officer: Mary Cummings.
Abstract: BACKGROUND: There is concern that care provided in the Veterans Health Administration (VA) may be of poorer quality than non-VA health care. We compared use of medications after acute myocardial infarction in the VA with that in non-VA healthcare settings under fee-for-service (FFS) Medicare financing. METHODS AND RESULTS: We used clinical data from 2486 VA and 29 249 FFS men >65 years old discharged with a confirmed diagnosis of acute myocardial infarction from 81 VA hospitals and 1530 non-VA hospitals. We reported odds ratios (ORs) for use of thrombolytics, beta-blockers, ACE inhibitors, or aspirin among ideal candidates adjusted for age, sample design (hospital academic affiliation, availability of cardiac procedures, and volume), and within-hospital clustering. Ideal VA candidates were more likely to undergo thrombolytic therapy at arrival (OR [VA relative to Medicare] 1.40 [1.05, 1.74]) or to receive ACE inhibitors (OR 1.67 [1.12, 2.45]) or aspirin (OR 2.32 [1.81, 3.01]) at discharge and equally likely to receive beta-blockers (OR 1.09 [1.03, 1.40]) at discharge. CONCLUSIONS: Ideal candidates in VA were at least as likely as those in FFS to receive medical therapies of known benefit for acute myocardial infarction.
Title: Variability in cost of coronary bypass surgery in New York State: potential for cost savings.
Publication: Am Heart J 2002 Jan;143(1):130-9.
Authors: Cowper PA et al.
Grant No.: HS06503.
Principal Investigator: Elizabeth DeLong.
Project Officer: Heddy Hubbard.
Abstract: OBJECTIVE: Previous analyses of variability in bypass resource use have not focused on hospital-level variation or adequately explored the influence of patient risk. We combined a clinical database with claims data to fully characterize patient level and hospital level variability in bypass surgery cost and length of stay in New York State and explored the extent to which lower cost is associated with worse quality of care. METHODS: By use of 1992 clinical and claims data, we identified by multivariable regression which patient characteristics influence bypass cost and length of stay. Hospital was then incorporated as a random variable in mixed linear models to determine its impact on resource use. The relationship between risk-adjusted in-hospital mortality and cost was then explored. RESULTS: In the 21 hospitals for which cost data were available, mean leveled cost (exclusive of professional fees and noncomparable costs) was $15,713, with a mean length of stay of 14 days (n = 12,087). One fifth of the variation in resource use was explained by baseline patient risk. After adjustment for patient risk, hospital explained an additional 42percent of variation in cost and an additional 8percent of variation in length of stay. Among hospitals, risk-adjusted cost varied almost 3-fold and risk-adjusted length of stay varied 50percent. There was no association between cost and in-hospital mortality. CONCLUSIONS: As of 1992, there was considerable interhospital variability in bypass surgery cost after patient baseline risk was accounted for. This suggests that reductions in bypass cost could be achieved by normalizing clinical practice.