Medication assessment tool to detect care issues from routine data: a pilot study in primary care.

Background: Medication assessment tools (MATs) may be implemented in routine electronic data sources in order to identify patients with opportunities for optimisation of medication therapy management (MTM) and follow-up by a multi-disciplinary team.

Objective: (1) To demonstrate the use of a MAT for cardiovascular conditions (MAT CVC) as a means of profiling potential opportunities for MTM optimisation in primary care and (2) to assess the performance of MAT CVC in identifying actual opportunities for better care.

Setting: Members of a pharmacotherapy discussion group, i.e. two single-handed general practitioners (GPs), three GP partners, and community pharmacists (CPs) from each of two community pharmacies, in a rural part of the Netherlands.

Methods: MAT CVC comprises 21 medication assessment criteria, each of which is designed to detect a specific care issue and to check whether it is 'addressed' by provision of guideline recommended care or 'open' in the presence ('open explained') or absence ('open unexplained') of pre-specified explanations for guideline deviations. (1) Relevant data was extracted from linked GP and CP electronic records and MAT CVC assessment was conducted to profile the population of CVC patients registered with both, participating CPs and GPs, in terms of 'open unexplained' care issues. (2) A purposive sample of patients with 'open unexplained' care issues was reviewed by each patient's GP.

Main Outcome Measures: Number and proportion of 'open unexplained' care issues per MAT CVC criterion and per patient. The number of patients with MAT CVC detected 'open unexplained' care issues to be reviewed (NNR) in order to identify one that requires changes in MTM.

Results: In 1,876 target group patients, MAT CVC identified 6,915 care issues, of which 2,770 (40.1 %) were 'open unexplained'. At population level, ten MAT CVC criteria had particularly high potential for quality improvement. At patient level, 1,277 (68.1 %) target group patients had at least one 'open unexplained' care issue. For patients with four or more 'open unexplained' care issues, the NNR was 2 (95 % CI 2-2).

Conclusion: The study demonstrates potential ways of using MA TCVC as a key component of a collaborative MTM system. Strategies that promote documentation and sharing of explanations for deviating from guideline recommendations may enhance the utility of the approach.