What are the risks of adverse outcomes associated with polypharmacy?

Problem

Polypharmacy, the use of multiple medications, is increasing in prevalence. Many drug treatments, indicated for individual conditions, carry similar potential risks and are often used in combination. The degree of risk associated with these combinations is often unclear. This study aims to assess risk associated with taking multiple medications with similar adverse side effect profiles.

Approach

We used data from SAIL databank (n = 1,175,826 aged ≥40, using primary care records) and UK Biobank (n = 502,533, age 37-73, using baseline assessment data). The Scottish Polypharmacy Guideline was used to identify medications associated with six potential adverse outcomes (fall or fracture, constipation, urinary retention, CNS depression, bleeding, and acute kidney injury (AKI)). Outcomes were identified using Hospital events, identified using ICD-10 codes from linked Hospital Episode Statistics. For each individual, we then calculated the number and type of medications associated with each adverse side effect taken at baseline. Cause specific Cox-Proportional Hazards models were used to assess the risk of each outcome associated with number of medications with similar adverse side effect profiles (0, 1, 2, ≥3). We also tested each combination of 2 medications taken by >100 people. Multi-state Weibull models used to predict 5-year risk associated with number of medications across a range of ages. All models adjusted for age, sex, deprivation and multimorbidity count.

Findings

Increasing numbers of medications with similar adverse side effect profiles was associated with falls (hazard ratio 1.13 (1.11-1.16) for ≥3 medications compared to 0 in SAIL), faecal impaction (1.17 (1.13-1.21)), urinary retention (2.21 (2.05-2.38)), delirium (1.86 (1.81-2.07)), bleeding (1.93 (1.81-2.07)) and AKI (2.97 (2.86-3.09)). Effect sizes were similar for UK Biobank. For each outcome, the risk increased with age, meaning that the difference in absolute risk associated with medications was higher for older people. For example, at age 50 the 5-year AKI risk associated with 1 and 3 or more medications linked with AKI was 1.2% (1.0-1.4) and 3.6% (3.4-3.8), respectively. At age 80 the risk was 6.8% (6.5-7.0) for 1 medication and 12.9% (12.4-13.5) for 3 or more medications. When comparing pairs of medication with similar adverse outcomes, the range of risk associated with different combinations of medications was wide (e.g. HR 4.17 (2.93-5.93) for delirium with opiates and benzodiazepines in combination, while many other combinations had null effect).

Consequences

Taking multiple medications with similar adverse side effect profiles is associated with higher risk of related outcomes. However, in absolute terms, this risk is far greater in older people. Different combinations of medication are also associated with markedly different levels of risk. These is therefore potential for alerts within electronic healthcare systems to incorporate data on magnitude of risk and better inform prescribing decisions.