An Ontario study involving almost 2.3 million older people has identified 24 high-priority prescribing cascades, where treatment for the suspected side effect of one medicine can lead to another drug being prescribed.

Common medicines used to manage pain, cardiovascular disease, dementia and other long-term conditions may contribute to chains of additional prescribing among older adults, according to a major Canadian study. The population-level retrospective cohort study, published in The BMJ on 10 September 2026, examined 65 previously identified potentially inappropriate prescribing cascades. Of these, 24 met all three of the researchers’ criteria for prioritisation: widespread use of the initial medicine, a sufficiently frequent subsequent prescription, and evidence that the two drugs were started in a sequence consistent with a possible cascade. The research was led by Paula A Rochon, a professor and geriatrician at Sinai Health and the University of Toronto. It involved 2,297,942 community-dwelling adults aged 66 or older in Ontario, 54.3 per cent of whom were female. The study population was defined as people alive on 1 January 2022, while new use of the medicine combinations was assessed from 1 January 2022 to 31 December 2023.

The pattern examined by the researchers is known as a potentially inappropriate prescribing cascade, or PIPC. It can occur when an adverse effect caused by an initial medicine is interpreted as a separate medical condition and a second drug is prescribed to manage it. The second medicine may therefore be unnecessary and could expose the patient to further adverse effects. However, the word “potentially” is important. The study analysed dispensing sequences in health administrative data; it did not establish that every individual combination represented an inappropriate clinical decision.

An example discussed in the source material is the use of non-steroidal anti-inflammatory drugs, or NSAIDs. These medicines can raise blood pressure. If the change is treated solely as newly developed hypertension, an antihypertensive drug might be added without first considering whether the original pain medicine contributed to the increase. For healthcare professionals, the finding reinforces the need to ask not only which medicines a person currently takes, but when each treatment began, why it was prescribed and whether a later symptom emerged after a medication change.

Study identifies 24 priority cascades

The researchers began with a list of 65 cascades developed through a 2025 modified Delphi consensus process. That earlier work involved 12 specialists in internal medicine, geriatric medicine and clinical pharmacology from eight countries. The list has also been endorsed by the European Geriatric Medicine Society, according to the paper. For the new analysis, each potential cascade was assessed separately using linked Ontario health data analysed at ICES, an independent, not-for-profit health research institute.

Three measures were used to determine priority:

at least 5 per cent of the study population had received the initial drug during the baseline period;

at least 1 per cent of new users of the initial medicine subsequently received the second medicine within one year; and

the adjusted sequence ratio was greater than 1, with the lower boundary of the 95 per cent confidence interval also exceeding the relevant threshold.

The sequence ratio compared how often the suspected initiating medicine came before the subsequent medicine with how often the order was reversed. The researchers adjusted this measure to account for underlying changes in prescribing patterns over time. Of the 65 candidate cascades, 24, or 37 per cent, met all three criteria. Cardiovascular medicines were the most prevalent classes associated with the beginning of a prioritised cascade, with prevalence values ranging from 19.0 to 65.9 per cent.

Iron supplements, statins and dementia medicines

The most frequently observed prioritised sequence was iron supplement followed by a laxative, with an incidence of 11.9 per cent. This is biologically plausible because constipation is a recognised problem associated with oral iron therapy, although the administrative data alone could not confirm the reason for each laxative prescription. The second-highest incidence was a statin, formally an HMG-CoA reductase inhibitor, followed by a pain reliever, at 10.9 per cent. A cholinesterase inhibitor followed by a sleep medicine had an incidence of 10.3 per cent. Cholinesterase inhibitors are commonly used in the management of dementia symptoms.

Frequency was only one part of the assessment. The strongest temporal association was seen for corticosteroid followed by an antipsychotic, with an adjusted sequence ratio of 2.55. This was followed by laxative to antidiarrhoeal treatment, at 2.53, and cholinesterase inhibitor to antiemetic treatment, at 2.24. These figures indicate that the initial-first order occurred more often than would be expected after adjusting for background prescribing trends. They do not, by themselves, demonstrate that the first medicine caused the second prescription.

Older adults are more likely to live with multiple long-term conditions and to take several medicines concurrently. This makes it difficult to distinguish between a new illness, progression of an existing condition and an adverse drug effect. The study focused on community-dwelling people rather than residents of long-term care homes at cohort entry. People aged 110 or older, those with incomplete data and those who had not made an Ontario Drug Benefit claim during the preceding 12 months were among the groups excluded under the study methodology.

The analysis also considered differences by sex. Most adjusted sequence ratios were similar between female and male participants. The principal exception involved antidepressants in the SSRI or SNRI classes followed by treatment for overactive bladder. The association was stronger among male older adults, with an adjusted sequence ratio of 1.21, than among female older adults, for whom the estimate was 1.02 and its confidence interval crossed 1. This is more limited than saying that mature women generally faced greater cascade risk. The paper reports some sex differences in medicine prevalence, but its main results do not establish that women experienced more of the 24 prioritised cascades overall.

Better reviews rather than automatic deprescribing

The study’s findings could help clinicians focus medication reviews on combinations that appear sufficiently common and temporally associated to warrant closer examination. This does not mean that the second medicine should automatically be stopped. Both treatments may be clinically justified. Instead, the sequence should prompt questions about the timing of symptoms, the indication for each prescription and whether modifying the first treatment might be safer than adding another. Pharmacists could have an important role in these reviews because they can examine prescription chronology, possible adverse effects and the continuing need for each medicine as part of multidisciplinary care.

The researchers also suggest that the prioritised list could support targeted interventions that alert healthcare providers to potentially problematic prescribing patterns. Their paper says the results could inform prescribing and deprescribing initiatives, drug-safety policy, educational materials and future tools for professionals and patients. Automated clinical decision support is one possible application, although the study itself did not test an alert system or demonstrate that alerts prevent harm. Any such system would need to avoid presenting every drug sequence as an error and instead provide clinicians with appropriately contextualised prompts for review.

Important study limitations

The study was observational and based on records of medicines dispensed, rather than detailed clinical assessments of individual patients. It could establish the order and timing of prescriptions, but not the clinical reason for every prescription or whether the patient experienced the suspected adverse effect. Over-the-counter treatments were not included. Information on race and gender was unavailable, limiting the researchers’ ability to examine intersecting demographic differences. The study also used prespecified thresholds for prioritisation, including a 5 per cent initial-drug prevalence threshold for which the authors said they were unaware of an established published recommendation.

Despite these qualifications, the size of the population and the systematic comparison of 65 candidate cascades provide a valuable basis for identifying prescribing patterns that deserve attention. The central message is not that multiple medicines are inherently inappropriate. For many older people, several treatments are necessary and beneficial. The risk arises when a medication-related symptom is treated in isolation and the chronology behind it is missed. Reconstructing that sequence may help clinicians prevent avoidable polypharmacy while preserving treatments that remain necessary.