We’ve been frequent critics of using Direct Observation to measure and impact hand hygiene performance (more here). Still, we hear from some hospitals who report that they can’t invest in an electronic hand hygiene reminder system this year…so what should they do?
Here are three steps to improve the accuracy of Direct Observation.
First, do what you can to minimize the Hawthorne Effect. This effect says that people will modify their behavior when they know they’re being observed[1]. In this context, reported hand hygiene rates are often two to three times higher than reality. This is because “secret shoppers” don’t stay secret for long, and providers clean their hands more often when they know an observer is present. (Here’s a recent study confirming this.)
So, minimize the Hawthorne Effect by not telling staff that they will be observed. Use a wide variety of observers that are rotated out and do their work inconspicuously. Expect your reported hand hygiene rates to drop…but they’ll be more accurate and better reflect reality.
Second, increase your sample size. The average non-critical care room has around 1000 hand hygiene opportunities a week, just looking at room entrances and exits. The average ICU room has closer to 2000 opportunities a week, based on our customer data. So, if you have a 30 bed ICU and only 30 observations a month…do the math. You’re capturing 0.0125% of the data, which is practically worthless.
If you’re going to do Direct Observation, do it right. Invest in the people resources to capture at least 5% of the data (room entrances and exits) to ensure that your findings are more accurate.
Third, train your observers to guard against bias. First, they need to record data continuously as their only focus. Asking someone to make note of ten hand hygiene opportunities while they perform their regular duties is begging for mistakes and bias. This is one form of recall bias, which says we’re more likely to notice when someone does clean their hands than when they don’t.
Observation bias says we’re likely to see what we expect to see. If an observer thinks that physicians are less likely to clean their hands than nurses, that’s what they’ll be more likely to see and report. There are also recording errors, performance bias, observer demographics, social hierarchies, complexity of observation, predictability of observer response, uniform application errors, and more.
Training is critical to ensure that observers are as unbiased as possible (knowing that eliminating all bias in humans is, unfortunately, impossible).
It is possible to improve Direct Observation…unfortunately, this doesn’t necessarily mean you’re increasing hand hygiene nor reducing HAIs. By its very nature, observers can’t remain anonymous and also correct behavior. The only way to both collect complete, accurate, unbiased data AND change behavior in the moment is to use an electronic hand hygiene reminder system.
If you’d like to explore how our system typically doubles hand hygiene performance rates…and has reduced HAIs by between 45% and 81% in 100% of customers following our process for 6 months…download our whitepaper to discover the process. Or here’s a brief video about how the system works.
[1] Srigley, J.A., et al., Quantification of the Hawthorne effect in hand hygiene compliance monitoring using an electronic monitoring system: a retrospective cohort study. BMJ Qual Saf, 2014. 23(12): p. 974-80.
