to the editor: I have several comments on the Medicine by the Numbers department on wound irrigation.¹ In the cited study on chronic wounds, the relative risk of infection in wounds treated with tap water compared to sterile saline irrigation (0.16) was not stated, but with a 95% confidence interval of 0.01 to 2.96, it was not statistically significant and did not warrant a number needed to treat (NNT) calculation.² This study also showed that there were twice as many wounds healed in the normal saline group than in the tap water group (16 vs. 8, respectively). Although this was not statistically significant either, the two groups were dissimilar in several important ways. The wounds in the normal saline group were older (216 days vs. 82 days), smaller (323 mm² vs. 503 mm²), and shallower (0.123 cm vs. 0.188 cm) than those in the tap water group; any of these factors could have affected the infection or healing rates.

At the present time, the best that can be said is that the risks and benefits of tap water irrigation are similar to those of normal saline irrigation and either may be the same as not irrigating at all.

in reply: I'd like to thank Dr. Pisarik for his response to our article. His letter discusses a study of 43 patients mentioned in the caveats section of our publication. As mentioned, the study addressed chronic wounds and found a statistically nonsignificant decrease in chronic wounds in the tap water group vs. the normal saline group. Dr. Pisarik is correct that an NNT calculation is not warranted for nonsignificant findings. The differences in wound demographics between the two groups was another limitation of this small study.

to the editor: The review by Drs. Chao and Runde¹ provided interesting and clinically useful information about a topic that is highly relevant to family physicians. However, we have a concern regarding the authors' potentially misleading interpretation of the study results drawn from a Cochrane review.²

The authors calculated a number needed to treat (NNT) of 36 to prevent one acute wound infection among the tap water group compared with the sterile saline group without including the confidence interval (CI). They concluded that using tap water for acute wound irrigation has benefits greater than harms based on nonsignificant differences from a Cochrane review (relative risk = 0.66; 95% CI, 0.42 to 1.04).² Typically, NNT is not applied to a finding that lacks statistical significance: the CI for the relative risk crosses 1.0. Such an application of the NNT only as a point estimate, especially when there is no significant difference, could easily be misleading. The importance of reporting the CIs with the NNT value has been highlighted previously.³ Including CIs helps us more easily determine and recognize the accuracy of estimated differences.

We agree with the authors' conclusion that tap water is a reasonable alternative to sterile saline for cleansing acute wounds. Potential benefits include cost of supplies, physician workload, and the risk of body fluid contamination with splatter.⁴ Nevertheless, given the current evidence, it is an overstatement to conclude that it is superior to sterile saline for preventing wound infections.

in reply: Drs. Ie and Wilson correctly note that the CI for a decrease in wound infection rates in the tap water group crossed 1.0 (relative risk = 0.66; 95% CI, 0.42 to 1.04) and, as such, the result is not statistically significant. In this case, we agree that an NNT calculation is not appropriate. This error was rooted in the fact that we used an outdated version of the Cochrane review during preparation of our article. In the originally published version of the Cochrane, the authors reported a treatment effect for tap water that was statistically significant (relative risk = 0.6; 95% CI, 0.40 to 0.99). An astute reader subsequently identified a transcription error in their analysis, which, when corrected, resulted in the treatment effect for tap water becoming nonsignificant. When the correct calculation was incorporated in the final prepublication review of this Medicine By The Numbers, I did not note the significance of the change and left the NNT calculation in place.

In a similar vein, Dr. Pisarik notes that we also calculated an NNT for tap water use in chronic wound irrigation for a result that was not statistically significant (NNT = 9; 95% CI, 0.01 to 2.96). In this case, we noted the lack of significance but wanted to include the calculation by way of comparison to our primary finding.

We are grateful to these physicians for highlighting the error and have corrected the online version of the review so it does not include either of these NNT calculations. Although these corrections are important, they do not have any substantive impact on the overall conclusion of the review: “Given the lack of adverse events and the affordability of tap water, tap water should be considered as preferable to normal saline for cleansing of acute wounds.”