Tom Nolan’s research reviews—13 July 2023

Pulling up the anchor

Is there a name for a cognitive bias for cognitive biases? The best publication in recent weeks has to be this cross sectional study that explores cognitive bias in clinical decision making. Anchoring bias occurs when we focus on a single—often initial—piece of information when formulating a diagnosis. For example, when you pick up the notes for the next patient who feels breathless and has a history of congestive heart failure (CHF), you’re less likely to look for a pulmonary embolism if it says “CHF” in the presenting complaint box than if it doesn’t.

That might seem a basic error, but, according to this study, it happens. In the 4.1% of 108 019 patients with CHF presenting to an emergency department with shortness of breath where CHF was mentioned in the section of the triage notes describing the reason for visit, it was associated with a 4.6% reduction in testing for pulmonary embolism and a 15 minute delay before testing for pulmonary embolism. Being aware of my bias towards this research, I should probably mention that these differences were small, it seems unlikely that they would lead to any significant harms, and less testing for pulmonary embolism may not always be such a bad thing.

JAMA Intern Med doi:10.1001/jamainternmed.2023.2366

Charting the overdose epidemic in the US

In the US the drug overdose epidemic continues to worsen. There’s been a sharp rise in fentanyl-associated overdose deaths outside hospital in recent years, as described in correspondence in the New England Journal of Medicine. Between 2016 and 2021, out-of-hospital deaths from fentanyl increased from 46 per million to 178 per million, while death rates from other substances have remained flat. By comparison, in 2021, 58 deaths involving fentanyl were recorded in England and Wales—around one death per million people—fewer than from oxycodone, paracetamol, or diazepam.

N Engl J Med doi:10.1056/NEJMc2304991

Inpatient treatment of asymptomatic high blood pressure

The main goal of checking blood pressure in inpatients is to monitor for hypotension. When it comes to raised blood pressure, there are no clinical trials assessing the benefits and harms of treating asymptomatic (that is, without signs of end organ damage) raised blood pressure in hospitalised patients. A retrospective cohort study looked at data from the Veterans Health Administration in the US (97.5% male, mean age 74 years) to see if patients who had high blood pressure readings in the first 48 hours of an admission, and were given either a new oral or any intravenous antihypertensive treatment, fared better or worse than those who weren’t given treatment. Those given treatment were slightly more likely to experience the composite primary outcome of inpatient mortality, acute kidney injury, stroke, myocardial injury, B-type natriuretic peptide elevation, or transfer to an intensive care unit. With this being an observational study, we need randomised trials to work this all out, but the authors argue that their findings “do not support the pharmacologic treatment of elevated blood pressures without evidence of acute end organ damage in hospitalised older adults.”

JAMA Intern Med doi:10.1001/jamainternmed.2023.1667

Guidelines not tramlines

They say that guidelines aren’t tramlines, but sometimes the gap between guidelines and practice is so great you have to wonder if anyone is listening to the guide anymore. When it comes to statin treatment, “prescriber or patient hesitation” as a new study carefully puts it, is likely to be a major factor behind the finding that nine out of 10 people in a representative sample of 25 000 people with coronary artery disease in the US are not achieving European Society of Cardiology guideline targets for low density lipoprotein (LDL)-cholesterol levels. I’m not sure how this compares with rates in the UK, but it does seem that the idea of intensive lipid lowering treatment, including adding second line agents to statins, still hasn’t caught on—and may never do so.

JAMA doi:10.1001/jama.2023.8646

Testing times ahead

People in the UK may be enjoying their summer holidays, but it won’t be long before we’re all coughing, snotting, and wheezing our way into the viral season. Alongside the bed crises, drug shortages, and mask mandates, we'll be back to asking again and again: “Have you taken a covid test?” But should we add a follow up question: “How many have you done?” A prospective study that enrolled people between October 2021 and January 2022 asked participants to do a rapid antigen covid test every 48 hours for 15 days. They found that taking a second test (two days later) improved sensitivity in symptomatic participants to 93.4% (95% CI 90.4% to 95.9%).

Ann Intern Med doi:10.7326/M23-0385