There is a high prevalence of potentially hazardous prescribing among some GP practices, a new research study published in the British Medical Journal has found. One of the authors, Dr Jill Stocks, explains the findings.
Safe prescribing of medication is one of the most essential roles of GPs, yet our team’s research has found both potentially hazardous prescribing among some GP practices, as well as wide variations between GP practices that need to be addressed.
Adverse drug events account for around 7% of hospital admissions in the UK; around half of these are judged to be preventable. One in 20 prescription items is associated with a clinically important error and one in 550 is associated with a serious error. Some prescribing patterns that should generally be avoided are well-known and have been described by prescribing safety indicators.
A prescribing safety indicator identifies a patient who is prescribed a medication that they usually should not take, because of an existing health condition or prescription.
Prescribing safety indicators are intended to pick up oversights or errors in prescribing. However, occasionally, there may be specific reasons for the need for the medication – prescribing safety indicators are not always an error and may be the best option for an individual patient, but their occurrence should be the exception rather than the rule. The intended role of an indicator is to prompt a medication review to ensure the potentially hazardous prescription is in the patient’s overall best interests.
In our paper we analysed anonymised patient data from 526 general practices which contributed to the Clinical Practice Research Datalink to measure how often prescribing safety indicators were happening. Some potentially hazardous prescribing happened quite often: for example, around 10% of patients across all 526 practices with a history of peptic ulceration or gastro-intestinal bleed were prescribed a drug that could increase the risk of a gastro-intestinal bleed (such as aspirin) without co-prescription of a gastro-protective medication to reduce that risk.
Other types of potentially hazardous prescribing happened less often – for example, less than 1% of women who were current smokers and aged over 35 years were prescribed a combined hormonal contraceptive; a pattern of potentially hazardous prescribing that could result in further increased risk of venous thromboembolism.
Combining these prescribing safety indicators we found that around 5% of patients received a potentially hazardous prescription. More important, though, was the difference between practices in how often potentially hazardous prescribing occurred, which ranged from 1% to 15%. This raises the question, can prescribing safety be improved by intervening in practices with higher occurrences of potentially hazardous prescribing? It also shows how anonymised electronic health data could be used to identify potential problems in health care delivery and improve patient safety in the future.
These are the facts, but what might be done to encourage practices with more potentially hazardous prescribing to improve prescribing safety? Is it right to use financial or other incentives to improve safety? A commentary on our paper makes the point that “Striving for excellence is something we would all readily sign up to, but it does sail dangerously close to the idea of league tables for prescribing safety.”
There is a precedent for financially incentivising good practice in the Quality Outcomes Framework. This has turned out to be controversial with arbitrary changes to the targets year on year and a lack of objective evaluation.
Even so, some Scottish GPs were amenable to the concept of small payments per record review of patients at risk of potentially hazardous prescribing. One difficulty, however, is that incentive schemes usually operate at the practice level, whereas prescribing tends to be the responsibility of the individual GP. This is supported by a recent Scottish study that found the frequency of occurrence of potentially hazardous prescribing varied equally between GPs in the same practice and those in different practices. This suggests that interventions should target all practices, not just those with overall higher levels. The identity of individual prescribers is not reliably recorded in electronic prescribing systems.
Are there other approaches to reducing the prevalence of potentially hazardous prescribing? Medication reviews may be helpful. The PINCER trial found that clinical pharmacists working with colleagues in primary care could reduce potentially hazardous prescribing, compared to simply feeding back to practices the identity of patients at risk of potentially hazardous prescribing.
Whether such simple feedback alone reduced potentially hazardous prescribing was not evaluated; it was judged unethical to identify patients at risk of potentially hazardous prescribing without informing the practices. Consequently a trial comparing practices receiving feedback with usual practice (i.e. no feedback) was not done. Unethical it may be, but being unaware of which patients receive a potentially hazardous prescription (as defined by these prescribing safety indicators) is the norm in primary care.
The DQIP trial in Scotland is currently investigating the effectiveness of weekly updated feedback of targeted prescribing at practice level in reducing potentially hazardous prescribing. This trial should provide a more robust answer to the question of whether feeding back the information about which patients are at risk of potentially hazardous prescribing will reduce the overall number at risk.
If it is true that “once problems are identified all health professions will pull together to make the necessary improvements”, then there is an urgent need to inform practices about which patients are receiving potentially hazardous prescriptions. This could be done at individual practice level – the software to identify patients receiving potentially hazardous prescriptions is available at low cost. Within a group of practices – for example, a Clinical Commissioning Group (CCG) – the aim could be to reduce the variation in potentially hazardous prescribing between practices. (Currently the pharmacist-led PINCER intervention is being rolled out in 17 CCGs.)
Appropriate statistical measures of variation are available to do this: no targets, no naming and shaming, just teamwork to reduce practice variation and the number of patients at-risk, as recommended in the 2015 NICE medicines optimisation guidance.