Identifying potentially dangerous prescribing patterns among patients taking multiple medicines
People living with several conditions are often prescribed a range of drugs, each with the potential to cause side-effects. If those effects are mistaken for symptoms of a new condition, another medicine may be prescribed instead of the original treatment being reviewed.
Researchers in Canada have identified 24 common examples of this sequence, known as a ‘prescribing cascade’, with the potential to cause harm. These cascades can leave patients taking unnecessary additional medicines and increase the risk of further side-effects.
One example involves non-steroidal anti-inflammatory drugs, which are widely prescribed for pain but can increase blood pressure. A patient may subsequently be prescribed medication for hypertension without the connection to the original pain treatment being recognised.
The study, published in The BMJ, examined population-level prescribing data from Ontario. An international panel of specialists in internal medicine, geriatric medicine and clinical pharmacology had previously identified 65 potentially inappropriate prescribing cascades. Researchers assessed these according to the prevalence of the initial medicine, the frequency with which it was followed by a second drug and the strength of the association between them.
This produced a priority list of 24 cascades commonly found within the population and considered capable of causing harm.
Risks for older women
Older people are particularly vulnerable because they are more likely to live with several conditions and take multiple medicines. The growing number of treatments can make it harder for patients and clinicians to connect a new symptom with a drug prescribed weeks, months or even years earlier.
The risks may be greater for mature women. Women tend to live with more chronic conditions over their lifetimes, receive more drug treatments and experience more adverse drug events than men, according to the research team.
AI-enabled clinical decision support could help clinicians detect these patterns before another medicine is prescribed. A system capable of examining a patient’s medication history and known adverse effects could alert the clinician when a suspected new condition may instead be connected to an existing treatment.
Its effectiveness would depend on access to complete and accurate information. Medication records may be distributed across primary care, hospitals, pharmacies and private providers, while over-the-counter treatments may not appear in the patient’s clinical record.
Systems would also need to distinguish meaningful risks from routine combinations. Excessive or poorly prioritised warnings could contribute to alert fatigue, causing clinicians to overlook the cases requiring urgent attention.
AI should therefore support rather than replace clinical judgement. The researchers also highlighted the importance of integrating pharmacists more closely into the prescribing process. Combining their expertise with carefully designed decision-support tools could enable health systems to review a patient’s complete treatment history and intervene before one prescription leads unnecessarily to another.
Have some thoughts on the topic? Share them with the community here