Re: Prevalence, severity, and nature of preventable patient harm across medical care settings: systematic review and meta-analysis
This is a timely and interesting piece, but unfortunately, it has conclusions that are, generally, not supported by the data presented.
Firstly, the headline figure - 1 in 20 patients suffer from preventable harm - is a classic example of conflating individual risk with population risk. The vast majority of patients in healthcare are in primary care, with a much smaller number in secondary (outpatient) care, then a smaller number again admitted, with an even smaller number requiring surgery or intensive care. As the authors note, the bulk of the risk of preventable harm applies in those two fields, so one cannot simply apply this risk across the whole healthcare setting.
Could the authors repeat the analysis excluding those two specialities, and provide a risk estimate that reflects the vast majority of patients?
Another problem relates to timing, an issue the authors briefly mention. Clearly, the 1 in 20 figure might make sense for an inpatient admission, but how this is measured in outpatients is less clear. Is this 1 in 20 risk applied across each individual consultation (this seems unlikely, suggesting that most GP's would cause preventable death or disability a few times a week, which is clearly wrong).
So how long is this? Is this across a patients whole exposure to healthcare? In which case, the 1 in 20 figure might seem more realistic, but given most multimorbid patients (those at the most risk of preventable harm) have hundreds, if not thousands of consultations, this makes the risk of preventable harm per consultation extremely low, limiting the potential benefit of any individual intervention.
For example, a prescribing system that leads to a ten second increase in consultation time for a 10% relative reduction in risk of harm per consultation might make sense if each consultation had a 1 in 20 risk of preventable harm, but not if the risk was 1 in 20,000 (assuming 1,000 consultations per lifetime).
Without knowing the denominator (risk per consultation? per admission? per year?), the interpretation of this figure is likely to lead to a significant overestimate of the benefit of harm prevention programmes, assuming a risk per consultation or process that is much higher than reality.
SpR in Microbiology and Infectious Disease
NIHR Academic Clinical Fellow
Competing interests: No competing interests