Medication reviews

Deep learning classification of drug-related problems from pharmaceutical interventions issued by hospital clinical pharmacists during medication prescription review: a large-scale descriptive retrospective study in a French university hospital

OBJECTIVES

Pharmaceutical interventions are proposals made by hospital clinical pharmacists to address sub-optimal uses of medications during prescription review. Pharmaceutical interventions include the identification of drug-related problems, their prevention and resolution. The objective of this study was to exploit a newly developed deep neural network classifier to identify drug-related problems from pharmaceutical interventions and perform a large retrospective descriptive analysis of them in a French university hospital over a 3-year period.

METHODS

Data were collected from prescription support software from 2018 to 2020. A classifier running in Python 3.8 and using Keras library was then used to automatically categorise drug-related problems from pharmaceutical interventions according to the coding of the French Society of Clinical Pharmacy.

RESULTS

2 930 656 prescription lines were analysed for a total of 119 689 patients. Among these prescription lines, 153 335 (5.2%) resulted in pharmaceutical interventions (n=48 202 patients; 40.2%). Pharmaceutical interventions were predominantly observed in patients aged 65 years or older (n=26 141 patients out of 53 186; 49.1%) and in patients taking five or more medications (44 702 patients out of 93 419; 47.8%). The most frequently identified types of drug-related problems associated with pharmaceutical interventions were 'Non-conformity to guidelines or contra-indication' (n=88 523; 57.7%), 'Overdosage' (16 975; 11.1%) and 'Improper administration' (13 898; 9.1%). The most frequently encountered drugs were: paracetamol (n=10 585; 6.9%), esomeprazole (6031; 3.9%), hydrochlorothiazide (2951; 1.9%), enoxaparin (2191; 1.4%), tramadol (1879; 1.2%), calcium (2073; 1.3%), perindopril (1950; 1.2%), amlodipine (1716; 1.1%), simvastatin (1560; 1.0%) and insulin (1019; 0.7%).

CONCLUSIONS

The deep neural network classifier used met the challenge of automatically classifying drug-related problems from pharmaceutical interventions from a large database without mobilising significant human resources. The use of such a classifier can lead to alerting caregivers about certain risky practices in prescription and administration, and triggering actions to improve patients' therapeutic outcomes.

Harmonizing the definition of medication reviews for their collaborative implementation and documentation in electronic patient records: A Delphi consensus study

BACKGROUND

Medication review practices have evolved internationally in a direction in which not only physicians but also other healthcare professionals conduct medication reviews according to agreed practices. Collaborative practices have increasingly highlighted the need for electronic joint platforms where information on medication regimens and their implementation can be documented, kept updated, and shared.

OBJECTIVE

The aim of this study was to harmonize the definition of medication reviews and create a unified conceptual basis for their collaborative implementation and documentation in electronic patient records (definition appellation: collaborative medication review).

METHODS

The study was conducted using the Delphi consensus survey with three interprofessional expert panel rounds in September-December 2020. The consensus rate was set at 80%. Experts assessed the proposed definition of collaborative medication review based on an international and national inventory of medication review definitions. The expert panel (n = 41) involved 12 physicians, 13 pharmacists, 10 nurses, and six information management professionals. The range of response rates for the rounds was 63-88%.

RESULTS

The experts commented on which of the pre-selected items (n = 75) characterizing medication reviews should be included in the definition of collaborative medication review. The items were divided into the following five themes and 51 of them reached consensus: 1) Actions included in the collaborative medication review (n = 24/24), 2) Settings where the review should be conducted (n = 5/5), 3) Situations where the review should be considered as needed and carried out (n = 10/11), 4) Prioritization of top five benefits to be achieved by the review and 5) Prioritization of top five patient groups to whom the review should be targeted.

CONCLUSIONS

A strong interprofessional consensus was reached on the definition of collaborative medication review. The most challenging was to identify individual patient groups benefiting from the review.

Deep learning application to automated classification of recommendations made by hospital pharmacists during medication prescription review

PURPOSE

Recommendations to improve therapeutics are proposals made by pharmacists during the prescription review process to address suboptimal use of medicines. Recommendations are generated daily as text documents but are rarely reused beyond their primary use to alert prescribers and caregivers. If recommendation data were easier to summarize, they could be used retrospectively to improve safeguards for better prescribing. The objective of this work was to train a deep learning algorithm for automated recommendation classification to valorize the large amount of recommendation data.

METHODS

The study was conducted in a French university hospital, at which recommendation data were collected throughout 2017. Data from the first 6 months of 2017 were labeled by 2 pharmacists who assigned recommendations to 1 of the 29 possible classes of the French Society of Clinical Pharmacy classification. A deep neural network classifier was trained to predict the class of recommendations.

RESULTS

In total, 27,699 labeled recommendations from the first half of 2017 were used to train and evaluate a classifier. The prediction accuracy calculated on a validation dataset was 78.0%. We also predicted classes for unlabeled recommendations collected during the second half of 2017. Of the 4,460 predictions reviewed, 67 required correction. When these additional labeled data were concatenated with the original dataset and the neural network was retrained, accuracy reached 81.0%.

CONCLUSION

To facilitate analysis of recommendations, we have implemented an automated classification system using deep learning that achieves respectable performance. This tool can help to retrospectively highlight the clinical significance of daily medication reviews performed by hospital clinical pharmacists.

Developing and piloting a self-assessment tool for medication review competence of practicing pharmacists based on nationally set competence criteria

BACKGROUND

New competence requirements have emerged for pharmacists as a result of changing societal needs towards more patient-centred practices. Today, medication review competence can be considered as basic pharmaceutical competence. Medication review specific competence criteria and tools for self-assessing the competence are essential in building competences and a shared understanding of medication reviews as a collaborative practice. The aim of this study was to develop and pilot a self-assessment tool for medication review competence among practicing pharmacists in Finland.

METHODS

The development of the self-assessment tool was based on the national medication review competence criteria for pharmacists established in Finland in 2017 and piloting the tool among practicing pharmacists in a national online survey in October 2018. The pharmacists self-assessed their medication review competence with a five-point Likert scale ranging from 1 for "very poor/not at all" to 5 for "very good".

RESULTS

The internal consistency of the self-assessment tool was high as the range of the competence areas' Cronbach's alpha was 0.953-0.973. The competence areas consisted of prescription review competence (20 items, Cronbach's alpha 0.953), additional statements for medication review competence (11 additional items, Cronbach's alpha 0.963) and medication review as a whole, including both the statements of prescription review and medication review competence (31 items, Cronbach's alpha 0.973). Competence items closely related to routine dispensing were most commonly self-estimated to be mastered by the practicing pharmacists who responded (n = 344), while the more clinical and patient-centred competence items had the lowest self-estimates. This indicates that the self-assessment tool works logically and differentiates pharmacists according to competence. The self-assessed medication review competence was at a very good or good level among more than half (55%) of the respondents (n = 344).

CONCLUSION

A self-assessment tool for medication review competence was developed and validated. The piloted self-assessment tool can be used for regular evaluation of practicing pharmacists' medication review competence which is becoming an increasingly important basis for their contribution to patient care and society.

Prevalence and factors associated with potentially inappropriate prescriptions among older patients at hospital discharge

RATIONALE, AIMS AND OBJECTIVES

Explicit criteria have been used worldwide to identify suboptimal prescribing such as potentially inappropriate prescriptions (PIPs). The objective of our study was to determine prevalence, types and factors associated with PIPs in older people discharged from an Albanian hospital.

METHOD

Retrospective, cross-sectional study conducted among patients aged 60 years and more discharged from the Cardiology and Internal Medicine departments of the University Hospital Center 'Mother Theresa' Tirana during 2013. PIPs were identified by using Beers (2012 update) and STOPP criteria (2008 and 2014 versions). Chi-square analysis and Student Test were performed. Crude and adjusted odds ratios with their 95% confidence intervals were estimated by logistic regression analysis.

RESULTS

Medical files for 319 patients were assessed. The median number of drugs prescribed was 7.8 (SD 2.2). PIPs prevalence at hospital discharge was 34.5% (95% CI 27.5-42.2%; 110 patients) according to both Beers and STOPP version 1 criteria. STOPP version 2 identified 201 (63.0%) patients with at least one PIP (95% CI 55.2-70.2%; 312 PIPs). The drugs more frequently involved in PIPs were aspirin, spironolactone, benzodiazepines, digoxin and methyldopa. The odds of having a PIP were higher in patients discharged from Internal Medicine (P < 0.005). The PIP index was 0.056%, 0.054% and 0.125% respectively for Beers, STOPP 2008 and STOPP 2014 criteria. A significant positive correlation was found between the number of prescribed drugs and PIP occurrence.

CONCLUSIONS

Our study found that between one and two out of three older patients has at least one PIP among the treatment prescribed at hospital discharge, depending on the tool used for detection. The high frequency of PIPs suggests the urgent need for interventions to reduce them.