Contemporary Practice of Anemia Treatment Among Dialysis Patients in the United States.

Introduction: The treatment of anemia is a major activity in the care of patients undergoing maintenance hemodialysis (HD). The comparative effectiveness of new pharmacologic treatments, relative to erythropoiesis-stimulating agents (ESAs), should be anticipated on the bases of controlled trials and current practice. We describe the contemporary practice of anemia treatment in a national cohort of patients undergoing maintenance HD. Methods: We analyzed the United States Renal Data System (USRDS) data to identify adult patients undergoing in-facility HD in 2016 to 2019. Using the Consolidated Renal Operations in a Web-Enabled Network (CROWNWeb) dataset, we identified hemoglobin and ESA utilization (agent and cumulative dose) during each patient-month, as well as intravenous (IV) iron utilization, ferritin, and transferrin saturation. We compared ESA dosing during the study era to dosing in the Normal Hematocrit Cardiac Trial (NHCT), conducted in the 1990s. We assessed ESA hyporesponsiveness by estimating the prevalence of the following: (i) high erythropoietin resistance index (ERI) and (ii) either 3 or 6 consecutive months with hemoglobin <10 g/dl. Results: Nearly two-thirds of patient-months had hemoglobin of 10.0 to 11.9 g/dl. Mean ESA utilization was 76.7% per month, with increasing use of pegylated epoetin beta. ESA dosing was stable; epoetin alfa dosing was slightly lower than in the low-target arm of the NHCT. The prevalence of ESA hyporesponsiveness was 22.2% if defined by high ERI, but only 2.1% to 6.0% if defined by 3 to 6 consecutive months with hemoglobin <10 g/dl. Median transferrin saturation was 22.3% with high ERI and persistently low hemoglobin. Conclusion: Hemoglobin and ESA dosing distributions are stable, with epoetin alfa dosing below the low-target arm of the NHCT. Persistently low hemoglobin occurs infrequently and may reflect iron depletion.

Measuring the predictability of life outcomes with a scientific mass collaboration.

How predictable are life trajectories? We investigated this question with a scientific mass collaboration using the common task method; 160 teams built predictive models for six life outcomes using data from the Fragile Families and Child Wellbeing Study, a high-quality birth cohort study. Despite using a rich dataset and applying machine-learning methods optimized for prediction, the best predictions were not very accurate and were only slightly better than those from a simple benchmark model. Within each outcome, prediction error was strongly associated with the family being predicted and weakly associated with the technique used to generate the prediction. Overall, these results suggest practical limits to the predictability of life outcomes in some settings and illustrate the value of mass collaborations in the social sciences.

Quantifying a rare disease in administrative data: the example of calciphylaxis.

BACKGROUND: Calciphylaxis, a rare disease seen in chronic dialysis patients, is associated with significant morbidity and mortality. As is the case with other rare diseases, the precise epidemiology of calciphylaxis remains unknown. Absence of a unique International Classification of Diseases (ICD) code impedes its identification in large administrative databases such as the United States Renal Data System (USRDS) and hinders patient-oriented research. This study was designed to develop an algorithm to accurately identify cases of calciphylaxis and to examine its incidence and mortality. DESIGN, PARTICIPANTS, AND MAIN MEASURES: Along with many other diagnoses, calciphylaxis is included in ICD-9 code 275.49, Other Disorders of Calcium Metabolism. Since calciphylaxis is the only disorder listed under this code that requires a skin biopsy for diagnosis, we theorized that simultaneous application of code 275.49 and skin biopsy procedure codes would accurately identify calciphylaxis cases. This novel algorithm was developed using the Partners Research Patient Data Registry (RPDR) (n = 11,451 chronic hemodialysis patients over study period January 2002 to December 2011) using natural language processing and review of medical and pathology records (the gold-standard strategy). We then applied this algorithm to the USRDS to investigate calciphylaxis incidence and mortality. KEY RESULTS: Comparison of our novel research strategy against the gold standard yielded: sensitivity 89.2%, specificity 99.9%, positive likelihood ratio 3,382.3, negative likelihood ratio 0.11, and area under the curve 0.96. Application of the algorithm to the USRDS identified 649 incident calciphylaxis cases over the study period. Although calciphylaxis is rare, its incidence has been increasing, with a major inflection point during 2006-2007, which corresponded with specific addition of calciphylaxis under code 275.49 in October 2006. Calciphylaxis incidence continued to rise even after limiting the study period to 2007 onwards (from 3.7 to 5.7 per 10,000 chronic hemodialysis patients; r = 0.91, p = 0.02). Mortality rates among calciphylaxis patients were noted to be 2.5-3 times higher than average mortality rates for chronic hemodialysis patients. CONCLUSIONS: By developing and successfully applying a novel algorithm, we observed a significant increase in calciphylaxis incidence. Because calciphylaxis is associated with extremely high mortality, our study provides valuable information for future patient-oriented calciphylaxis research, and also serves as a template for investigating other rare diseases.