Evaluating structured care for diabetes: can calibration on margins help to avoid overestimation of the benefits? An illustration from French diabetes provider networks using data from the ENTRED Survey.


Chevreul, K; Brunn, M; Cadier, B; Nolte, E; Durand-Zaleski, I; (2014) Evaluating structured care for diabetes: can calibration on margins help to avoid overestimation of the benefits? An illustration from French diabetes provider networks using data from the ENTRED Survey. Diabetes care, 37 (7). pp. 1892-9. ISSN 0149-5992 DOI: https://doi.org/10.2337/dc13-2141

Full text not available from this repository.

Abstract

While there is growing evidence on the effectiveness of structured care for diabetic patients in trial settings, standard population level evaluations may misestimate intervention benefits due to patient selection. In order to account for potential biases in measuring intervention benefits, we tested the impact of calibration on margins as a novel adjustment method in an evaluation context compared with simple poststratification. We compared the results of a before-after evaluation on HbA1c levels after 1 year of enrollment in a French diabetes provider network (DPN) using an unadjusted sample and samples adjusted by simple poststratification to results obtained after adjustment via calibration on margins to the general diabetic population's characteristics using a national cross-sectional sample of diabetic patients. Both with and without adjustment, patients in the DPN had significantly lower HbA1c levels after 1 year of enrollment. However, the reductions in HbA1c levels among the adjusted samples were 22-183% lower than those measured in the unadjusted sample, regardless of the poststratification method and characteristics used. Compared with simple poststratification, estimations using calibration on margins exhibited higher performance. Evaluations of diabetes management interventions based on uncontrolled before-after experiments may overestimate the actual benefit for patients. This can be corrected by using poststratification approaches when data on the ultimate target population for the intervention are available. In order to more accurately estimate the effect an intervention would have if extended to the target population, calibration on margins seems to be preferable over simple poststratification in terms of performance and usability.

Item Type: Article
Faculty and Department: Faculty of Public Health and Policy > Dept of Health Services Research and Policy
Research Centre: ECOHOST - The Centre for Health and Social Change
PubMed ID: 24784830
Web of Science ID: 338020400023
URI: http://researchonline.lshtm.ac.uk/id/eprint/2255428

Statistics


Download activity - last 12 months
Downloads since deposit
0Downloads
236Hits
Accesses by country - last 12 months
Accesses by referrer - last 12 months
Impact and interest
Additional statistics for this record are available via IRStats2

Actions (login required)

Edit Item Edit Item