Premium
Individualizing HbA 1c targets for patients with diabetes: impact of an automated algorithm within a primary care network
Author(s) -
Berkowitz S. A.,
Atlas S. J.,
Grant R. W.,
Wexler D. J.
Publication year - 2014
Publication title -
diabetic medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.474
H-Index - 145
eISSN - 1464-5491
pISSN - 0742-3071
DOI - 10.1111/dme.12427
Subject(s) - medicine , mcnemar's test , algorithm , diabetes mellitus , medicaid , population , observational study , primary care , health care , family medicine , statistics , computer science , mathematics , environmental health , economic growth , economics , endocrinology
Abstract Aims To develop glycaemic goal individualization algorithms and assess potential impact on a healthcare system and different segments of the population with diabetes. Methods A cross‐sectional observational study of patients with diabetes in a primary care network age > 18 years with an HbA 1c measured between 1 January and 31 December 2011. We applied diabetes guidelines to create targeted algorithms 1 and 2, which assigned HbA 1c goals based on age, duration of diabetes (< 15 years or < 10 years), diabetes complications and Charlson co‐morbidity score (< 6 or < 4) [targeted algorithm 2 was designed to assign more patients a goal < 64 mmol/mol (8.0%) than targeted algorithm 1]. Each patient's HbA 1c was compared with these targeted goals and to the ‘standard’ goal < 53 mmol/mol (7.0%). Agreement was tested using McNemar's test. Results Overall, 55.7% of 12 199 patients would be considered controlled under the ‘standard’ approach, 61.2% under targeted algorithm 1 and 67.5% under targeted algorithm 2. Targeted algorithm 1 reclassified 1213 (23.6%) patients considered uncontrolled under the standard approach to controlled, P < 0.001. Targeted algorithm 2 reclassified 1844 (35.2%) patients, P < 0.001. Compared with those controlled under the standard goal, there was no significant difference in the proportion of those controlled using targeted goals who had Medicaid, had less than a high school diploma or received primary care in a federally qualified health centre. Conclusions Two automated targeted algorithms would reclassify one quarter to one third of patients from uncontrolled to controlled within a primary care network without differentially affecting vulnerable patient subgroups.