Faculty Bibliography

OBJECTIVE: To test the validity of three published algorithms designed to identify incident breast cancer cases using recent inpatient, outpatient, and physician insurance claims data. DATA: The Surveillance, Epidemiology, and End Results (SEER) registry data linked with Medicare physician, hospital, and outpatient claims data for breast cancer cases diagnosed from 1995 to 1998 and a 5 percent control sample of Medicare beneficiaries in SEER areas. STUDY DESIGN: We evaluate the sensitivity and specificity of three algorithms applied to new data compared with original reported results. Algorithms use health insurance diagnosis and procedure claims codes to classify breast cancer cases, with SEER as the reference standard. We compare algorithms by age, stage, race, and SEER region, and explore via logistic regression whether adding demographic variables improves algorithm performance. PRINCIPAL FINDINGS: The sensitivity of two of three algorithms is significantly lower when applied to newer data, compared with sensitivity calculated during algorithm development (59 and 77.4 percent versus 90 and 80.2 percent, p<.00001). Sensitivity decreases as age increases, and false negative rates are higher for cases with in situ, metastatic, and unknown stage disease compared with localized or regional breast cancer. Substantial variation also exists by SEER registry. There was potential for improvement in algorithm performance when adding age, region, and race to an indicator variable for whether the algorithm determined a subject to be a breast cancer case (p<.00001). CONCLUSIONS: Differential sensitivity of the algorithms by SEER region and age likely reflects variation in practice patterns, because the algorithms rely on administrative procedure codes. Depending on the algorithm, 3-5 percent of subjects overall are misclassified in 1998. Misclassification disproportionately affects older women and those diagnosed with in situ, metastatic, or unknown-stage disease. Algorithms should be applied cautiously to insurance claims databases to assess health care utilization outside SEER-Medicare populations because of uneven misclassification of subgroups that may be understudied already

To quantify the house officer's role in end-of-life decisions, the authors abstracted charts for documentation of end-of-life discussions for 100 patients withdrawn from life-sustaining treatment. They assessed the proportion of end-of-life care notes written by house officers, controlling for service, length of stay, outpatient physician involvement, race, and diagnostic category. Patients on the medical service were 22 times more likely to have house officer end-of-life notes than patients on the surgical service (P < 0.00001). Sixty-one percent of medical patients and 10% of surgical patients had a do-not-resuscitate note written by a house officer (P < 0.00001). House officers on the medical service wrote a significantly greater proportion of notes regarding withdrawal of care than surgical house officers (41% vs. 10%, P < 0.00001). This study reveals extensive involvement of medical house officers in primary end-of-life discussions with a complex patient population undergoing withdrawal of life-sustaining therapy. Team structure and professional culture may account for some of the observed differences between the medical and surgical services. These findings have significant implications for the education of house officers on end-of-life communication

OBJECTIVE: We assessed patient views regarding disclosure of first-trimester Down syndrome risk assessment results compared with withholding results until a single, more accurate second-trimester screening result is available. STUDY DESIGN: A prospective, institutional review board-approved, voluntary, anonymous survey was presented to patients with singleton pregnancies who were undergoing first-trimester nuchal translucency and biochemical screening at our institution. Options included immediate (sequential testing) or later disclosure (integrated testing). Descriptions and a comparison of the options were included in the survey. RESULTS: One hundred one women completed the questionnaires; 69.3% of the patients preferred sequential testing, compared with 30.7% of the patients who were either unsure or preferred integrated testing. Older patients and those patients with better background knowledge of screening tests preferred earlier disclosure of screening results (P = .035 and P = .026, respectively). Patients who preferred earlier disclosure also preferred termination of a Down syndrome pregnancy (P = .013). CONCLUSION: Older patients and those patients with a better understanding of screening tests preferred immediate disclosure of first-trimester risk assessment results; these women were more likely to prefer to terminate an affected pregnancy

OBJECTIVE: We sought to determine the prevalence of defined abnormalities on surveillance computed tomography after complete resection for non-small cell lung cancer, as well as the nature and cost of further testing prompted by these abnormalities. We also sought to determine whether resectable metachronous lung cancer can be detected with surveillance scans. METHODS: A retrospective analysis was performed of all patients who presented for follow-up in 2002 after complete resection for non-small cell lung cancer. Data collected included demographics, clinicopathologic features of the initial lung cancer, the number and results of surveillance computed tomographic scans performed in 2002, the attending surgeons' impressions of the surveillance scans, the nature of any abnormalities and further diagnostic testing prompted by these abnormalities, and the nature of any lung cancer detected on surveillance scans, as well as the treatment rendered. The cost of surveillance scanning and associated diagnostics was computed by using Medicare fee schedules. RESULTS: Two hundred thirteen patients met the criteria for inclusion in the study cohort. One hundred sixty-eight surveillance scans were performed in 140 of these patients. One hundred five scans were interpreted as abnormal by the radiologist with regard to pulmonary nodules, adenopathy, or pleural fluid, but the surgeon was suspicious for recurrent or new primary lung cancer in only 32 of 105 scans. Further workup revealed recurrent or new primary lung cancer in 16 of 32 patients, with 6 undergoing resection for localized disease. The cost of the surveillance scans and associated care in the study cohort were 16.6% higher than the cost of care in a hypothetically identical cohort not subjected to surveillance scanning. CONCLUSIONS: Surveillance computed tomography is frequently abnormal after complete resection for non-small cell lung cancer; however, the majority of these abnormalities are not clinically suspicious. Resectable metachronous lung cancer is detected by using surveillance scanning; however, the use of this modality can be associated with increased cost

OBJECTIVE: The objective of this study is to quantify variation and variability in treatment of ductal carcinoma in situ (DCIS) over time and across registries of the Surveillance, Epidemiology, and End Results (SEER) program; to assess diffusion of treatments (breast-conserving surgery [BCS], BCS with radiotherapy, and mastectomy); and to identify correlates of treatment choice. DATA: The linked SEER-Medicare database from 1991 to 1996 includes 2701 women aged 65 and older diagnosed with unilateral DCIS. 1990 census data provide socioeconomic variables at the zip-code level, and the 1999 Dartmouth Atlas of Health Care provides number of radiation oncologists. STUDY DESIGN: Bivariate and multivariate analyses of retrospective cohort data assess factors that explain treatment choice. The multivariate model includes controls for comorbidity, marital status, age, race, education, poverty, rural, and radiation oncologists per 100,000 population. Chi-squared tests assess differences in treatment rates by registry and by year. Diffusion of treatments is analyzed by predicting yearly mean treatment rates and yearly variation in treatment rates across geographic areas and over time. RESULTS: There are significant geographic and temporal differences in treatment rates for DCIS with increasing use of BCS alone. Treatment choice is explained by SEER registry, diagnosis year, marital status, race, age, urban/rural status, educational attainment, and number of radiation oncologists. Variability in treatment of DCIS is increasing during the study period. CONCLUSIONS: Findings indicate that diagnosis year and socioeconomic factors explain treatment choice for DCIS, but unexplained variation at the geographic-region level remains. Increasing variability in treatment implies continued uncertainty about optimal treatment of DCIS

BACKGROUND: There is limited evidence about the extent to which sensitivity analysis has been used in the cost-effectiveness literature. Sensitivity analyses for health-related QOL (HR-QOL), cost and discount rate economic parameters are of particular interest because they measure the effects of methodological and estimation uncertainties. AIM: To investigate the use of sensitivity analyses in the pharmaceutical cost-utility literature in order to test whether a change in economic parameters could result in a different conclusion regarding the cost effectiveness of the intervention analysed. METHODS: Cost-utility analyses of pharmaceuticals identified in a prior comprehensive audit (70 articles) were reviewed and further audited. For each base case for which sensitivity analyses were reported (n = 122), up to two sensitivity analyses for HR-QOL (n = 133), cost (n = 99), and discount rate (n = 128) were examined. Article mentions of thresholds for acceptable cost-utility ratios were recorded (total 36). Cost-utility ratios were denominated in US dollars for the year reported in each of the original articles in order to determine whether a different conclusion would have been indicated at the time the article was published. Quality ratings from the original audit for articles where sensitivity analysis results crossed the cost-utility ratio threshold above the base-case result were compared with those that did not. RESULTS: The most frequently mentioned cost-utility thresholds were $US20,000/QALY, $US50,000/QALY, and $US100,000/QALY. The proportions of sensitivity analyses reporting quantitative results that crossed the threshold above the base-case results (or where the sensitivity analysis result was dominated) were 31% for HR-QOL sensitivity analyses, 20% for cost-sensitivity analyses, and 15% for discount-rate sensitivity analyses. Almost half of the discount-rate sensitivity analyses did not report quantitative results. Articles that reported sensitivity analyses where results crossed the cost-utility threshold above the base-case results (n = 25) were of somewhat higher quality, and were more likely to justify their sensitivity analysis parameters, than those that did not (n = 45), but the overall quality rating was only moderate. CONCLUSIONS: Sensitivity analyses for economic parameters are widely reported and often identify whether choosing different assumptions leads to a different conclusion regarding cost effectiveness. Changes in HR-QOL and cost parameters should be used to test alternative guideline recommendations when there is uncertainty regarding these parameters. Changes in discount rates less frequently produce results that would change the conclusion about cost effectiveness. Improving the overall quality of published studies and describing the justifications for parameter ranges would allow more meaningful conclusions to be drawn from sensitivity analyses

OBJECTIVE: This study examined the relationship of surgeon subspecialty training and interests to in-hospital mortality while controlling for both hospital and surgeon volume. SUMMARY BACKGROUND DATA: The relationship between volume of surgical procedures and in-hospital mortality has been studied and shows an inverse relationship. METHODS: A large Statewide Planning and Research Cooperative System was used to identify all 55,016 inpatients who underwent gastrectomy (n = 6434) or colectomy (n = 48,582) between January 1, 1998 and December 31, 2001. Surgical subspecialty training and interest was defined as surgeons who were members of the Society of Surgical Oncology (training/interest; n = 68) or the Society of Colorectal Surgery (training; n = 61) during the study period. The association of in-hospital mortality and subspecialty training/interest was examined using a logistic regression model, adjusting for demographics, comorbidities, insurance status, and hospital and surgeon volume. RESULTS: Overall mortality for colectomy patients was 4.6%; the adjusted mortality rate for subspecialty versus nonsubspecialty-trained surgeons was 2.4% versus 4.8%, respectively (adjusted odds ratio [OR] = 0.45; 95% confidence interval [CI] = 0.34, 0.60; P < 0.0001). Gastrectomy patients experienced an overall mortality rate of 8.4%; the adjusted mortality rate for patients treated by subspecialty trained surgeons was 6.5%, while the adjusted mortality rate for nonsubspecialty trained surgeons was 8.7% (adjusted OR = 0.70; 95% CI = 0.46, 1.08; P = 0.10). CONCLUSIONS: For gastrectomies and colectomies, risk-adjusted mortality is substantially lower when performed by subspecialty interested and trained surgeons, even after accounting for hospital and surgeon volume and patient characteristics. These findings may have implications for surgical training programs and for regionalization of complex surgical procedures

BACKGROUND: Pneumococcal immunization has been shown to be cost effective, is recommended by the Advisory Committee on Immunization Practices, and is covered by Medicare. Despite that, over 50% of the population aged > or =65 is not vaccinated, leading to significant mortality and morbidity. The objective of this study is to evaluate the costs and the cost utility of immunization in nontraditional settings (community clinics set up to provide influenza and pneumococcal vaccinations) as a strategy to increase pneumococcal immunization rates. METHODS: A cost-utility analysis of public immunization clinics in Monroe County, New York, during the fall of 1998. The study included 1207 adults aged > or =65. Costs of operating the clinics and of vaccine administration were measured. The cost of health sequela and estimates of quality-adjusted life years (QALYs) were obtained from prior studies. Sensitivity analyses were performed to test several important assumptions. RESULTS: Unlike immunizations in physician offices, immunizations in nontraditional settings are not cost saving. Estimates of incremental cost-utility ratios ranged from $4215 per QALY to $12,617 per QALY, depending on the underlying assumptions of the model. CONCLUSIONS: Clinics in nontraditional settings offering pneumococcal immunization have cost-utility ratios near and below those of other recommended vaccines. These results suggest that such clinics should be considered a viable strategy for increasing pneumococcal immunization rates