Nationwide Solution for Drug Interactions

Drug interaction checker as clinical decision support

Electronic prescribing can include clinical decision support (CDS) alerts about drug-drug interactions (DDI) and, therefore, enhance the drug-use process from prescribing to dispensing.

Electronic prescribing improves the quality of information regarding the patient’s drugs (active medication list) and integrates the latest evidence to e-prescription.

Clinicians study pharmacokinetics and pharmacodynamics, but without the help of technology, it is impossible to remember tens of thousands of combinations of interacting drugs. Another problem is the exponential growth of research: medical knowledge doubles every 73 days (Densen et al., 2011). How to always be up-to-date?

A clinical decision support tool for drug interactions can be integrated into electronic health records (EHR) or pharmacy point-of-sale systems (POS). This article describes nationwide adoption, whereas drug interaction service is integrated into a centralized e-prescription system. Estonia had implemented a national e-prescription system in 2010, and Lithuania did the same in 2016.

Precondition for implementation

A critical precondition to consider is the adoption rate of e-prescription. It is essential to include all drugs used by a patient for automatic drug interactions analysis to work correctly; otherwise, the service will create false recommendations and has low value for users.

In Estonia, the e-prescription adoption rate of e-prescriptions was 99% before implementation, and in Lithuania, the adoption rate was close to 90% before the performance.

Evidence for using clinical decision support systems

In recent years, clinical decision support systems have focused on health technology assessments (HTA). The evidence demonstrating the effectiveness of using clinical decision support systems is solid.

• Evidence on effectiveness – integrating the drug-drug interaction database as clinical decision support into primary health care records was associated with a 17% decrease of serious interactions per prescribed drug-drug pair (Andersson et al. 2012).
• Evidence on effectiveness– integration of the drug-drug interaction clinical decision support into e-prescription had a significant impact on prescribing habits, resulting in 28% of cases to changes in prescription by a clinician (Doogue et al. 2017).
• Evidence on economic impact– implementation of drug-related clinical decision support brought a reduction of medications on average 0.82 drugs/patient (57% one drug, 11% two drugs) (Rieckert et al. 2019).

Feasibility study

Estonia implemented an e-prescription system in 2010. The wide use of e-prescription created secondary data for research purposes.

In 2014, a feasibility study “Pharmacokinetic drug-drug interactions among Estonian elderly” was done by Jana Lass, Mait Raag, and Alar Irs. A retrospective cross-sectional descriptive drug utilization study considered the Estonian Health Insurance Fund’s prescription data and was evaluated with the drug interactions database Inxbase.

All prescription medicines dispensed to Estonian older adults (>65 years old) from January 1 to December 31, 2013, were recorded. Patients who had more than two prescriptions dispensed during a predefined prescription window (≤90 days) were included. Prescription data were linked to the Inxbase database to yield the frequency of interacting combinations dispensed in the population during this period.

According to a nationwide study: 7% of over 65-year patients had at least one type D drug interaction, and 35% had at least one type C interaction.

The most common pairs of D category interactions were propafenone-metoprolol, warfarin diclofenac, and warfarin-tramadol, which occurred in 2182, 1354, and 1070 individuals, respectively. D category interactions occurred most with the following medicines: warfarin (28%), metoprolol (7.4%), propafenone (7%), and clopidogrel (6.9%). 10 927 individuals received one pair of C or D-category interacting medicines, while 2 142 received two pairs. One person had eight pairs of interacting medications at the same period.

Read more about the database: Drug interactions database.

Checklist for implementation

Active medication list – is the active medication list available that covers all drugs used by the patient, or the list includes only some of the drugs (e.g., half of the prescriptions are still on paper). We recommend using drug interactions as a nationwide service only when the e-prescription adoption rate is below 90%.

Prescribed/purchased – is there are available data only about prescriptions or also about purchases? Adherence studies have shown that almost 20% of prescribed drugs will not be bought. 

OTC drugs – is the electronic data about over-the-counter medications (OTC) available? Studies have shown that 20% of interactions are related to OTC drugs such as NSAIDs.

Drug package database – is there is an available database that covers all drugs registered in a specific country? How often is this database updated? How this data can be used – can this data be accessed via an application programming interface (API)?

Classification system – is there implemented ATC classification system by WHO?

Drug formulation – is a list of drug formulations available that cover all drug formulations registered in a specific country? How often is this data updated? How this data can be used – can this data be accessed via an application programming interface (API)?

Multisource solution – if an integrated solution uses more than one source, how will you solve the issue of duplicates? For example, some prescriptions are in central systems and some in local electronic health records (EHR).

Health registry data – There are available local health workers registry data for authentication and single sign-on (SSO) services. How this data can be used – can this data be accessed via an application programming interface (API)?

Read more compliance issues:

Implementing clinical decision support in European Union

Value of secondary data

Before implementing drug interactions related to clinical decision support, any healthcare system can only speculate on the prevalence of drug interactions. Some may even claim that in their hospital or region, this problem doesn’t exist.

After implementing clinical decision support, any healthcare system can have detailed secondary data about the prevalence of interactions. In Estonia, the national drug interaction service was available from June 2016. Here is secondary data collected for one month period from the Estonian Health Insurance Fund:

• In one month, 1 206 752 e-prescriptions were dispensed
• The e-prescription adoption rate is 99.5% in this month
• Drug interaction system gave 358 986 alerts, incl. 41 458 D-level alerts
• A total of +2500 interacting combinations was analyzed
• Prevalence: 2.6% of patients had at least one D-level interaction, and 12.8% had at least one C-level interaction