Excessive noise in the operating room can lead to surgical site infections [SSI] according to a study published in the British Journal of Surgery. The authors measured sound levels in 35 cases of open colon and hepatobiliary surgery and found that SSI was statistically significantly more likely to occur in cases in which the noise level was higher. They suggested that the distraction associated with noise might be the root cause. The abstract states “Demographic parameters and duration of operation were not significantly different between patients with, or without SSI.”
Although one always likes to see research that confirms one’s biases, I have to question the methods and conclusions of this work. Here are the problems:
Some of the parameters assessed by a nurse observer in the operating room, such as positive general impression, lack of focused conversation and non-patient-related conversation, were a little vague at best.
The study involved too few patients and was underpowered. It is well-known that the longer an operation takes, the more likely an SSI will occur. The median duration of surgery was 135 minutes longer in the 6 patients who developed SSIs. While this was not statistically significant, the p value was 0.056, which is about as close as it gets to the magical p value of <0.05.
Colorectal surgery, which is far more apt result in an SSI, was done 5/6 patients in the SSI group but only 12 of the 29 who did not have an infection. Again this was not statistically significant, but the p value was 0.088.
Just a few more patients in each group for both duration of surgery and number having colorectal surgery would have made the differences significant.
The median peak sound levels above baseline for the two groups [SSI vs. No SSI] were only 4 decibels [dB] different. While this was statistically significant, it is not clear that a rise of 4 dB is actually important.
The paper says that non-patient-related conversations created statistically significantly higher dB levels. For some reason, this portion of the paper only included 29 of the original 35 patients. A p value of 0.024 [statistically significant] was given for the comparison using chi square as the test. When I run the same test on their numbers, I get 0.061 [not statistically significant]. Because the numbers are so small, the correct test for this comparison is Fisher’s exact test, which also yields a statistically insignificant value of p of 0.052.
Several basic principles of critical analysis have been discussed but the most important is—read the whole paper, not just the abstract.