Could “normal affect variation” be a new physical examination finding that predicts severity of illness?
Dr Jeff Kline thinks it might be. He’s the preeminent PE researcher who brought us the PERC rule. He has conducted this pilot study that suggests we might be on to something.
This was a prospective study of a convenience sample of patients presenting to the ED with dyspnea and chest pain. 50 patients looked at 3 stimulus slides on a laptop computer that recorded their facial expressions. Did their faces move when viewing the slides?
Two independent blinded observers reviewed the recordings and scored the change in facial expression using a coding system. The main analysis compared the results of the score to the patient’s presence of disease (predefined yes/no serious cardiopulmonary disease.)
How well did the change in facial expression perform?
It seemed to be ok.
The expression of surprise had the largest difference between yes/no disease with an area under the receiver operating curve (AUC) of 0.75. But given the low numbers, the confidence intervals were rather wide (95% CI 0.52 to 0.87).
Perhaps the biggest limitation to the study beyond the small numbers was the gold standard. How do we quantify “how sick” someone is? The authors acknowledge several patients did not meet their predefined determination of “serious cardiopulmonary disease” but were likely sick due to other pathology.
To be fair, Dr Kline points out that we consciously or unconsciously use affect variation as part of our gestalt assessment anyway. How many times have you heard an experienced clinician say, “the patient just looks sick…” But this study is a step towards standardisation and/or quantification of this implicit process.
Could artificial intelligence be the next step… perhaps. Dr Kline co-founded the company BreathQuant Medical Systems Inc that has many patents for medical devices. Perhaps we’ll have “affect variationometer” soon.
Kline JA, Neumann D, Huang MA, et al. Decreased facial expression variability in patients with serious cardiopulmonary disease in the emergency care setting. Emerg Med J 2015;32:3-8. [link to article]