This was a unique study out of a single paediatric care ED in Toronto, Canada. The authors were obviously of the opinion that they were admitting too many children to hospital that could otherwise go home based on low pulse ox readings. Does a single number change our practice?
The study design involved a little deception. (Who doesn't like a little deception?) Infants with moderate bronchiolitis were randomized in a double blind fashion to getting an accurate pulse oximetry measurement vs. one that displayed a result 3% points higher than the real measurement. In case you are wondering, they had the manufacturer alter the machines and they bribed the ethics committee (okay... I made that last bit up).
The primary outcome was hospitalization within 72 hours. There were other common sense secondary outcome measurements.
213 infants were randomized and there was no loss to follow-up. 41% of patients with the true oximetry group were hospitalized as compared to 25% in the altered oximetry group. Absolute difference 16% or NNT to prevent a hospitalization is 7. The overall strategy seemed pretty safe. Wow, pretty compelling results.
But before you consider throwing away your pulse oximeter, consider the following. The mean true pulse ox reading in all of the patients enrolled was about 97%. Very few of these infants were really that sick with a mean respiratory rate about 50. Only 28 infants in this study had true readings below 94%. There probably is no problem when you are considering a pulse ox of 97% being inaccuragely displayed as 100%. So this study is clearly not powered to investigate safety.
Nevertheless, I think this study is quite interesting and emphasizes a key point. Are we too reliant on single numbers when it comes to patient treatment decisions? The answer is probably yes.
Schuh S, Freedman S, Coates A, et al. Effect of Oximetry on Hospitalization in Bronchiolitis. A Randomized Clinical Trial. JAMA. 2014;312(7):712-718.