Skirting in abstracts of dermatology articles: A preliminary study

Sir,

Skirting is a term often used in scientific literature to refer to the tendency of the authors to avoid liability by using indirect words such as “can,” “may” etc in place of terms that are more definitive.¹ For example, instead of saying “hepatoxicity is associated with itraconazole,” by resorting to skirting one says “hepatotoxicity may be associated with itraconazole.” While the second sentence is probably safer from the point of view of the speaker in the sense that it exonerates him from any further responsibility, excessive use of this literary device often confuses the reader and dilutes the importance of the research findings. However, very little, if any, research has been done on the prevalence of skirting in the dermatology literature.²

Excessive skirting was defined as the occurrence of the words “can” or “may” three or more times individually or more than five times taken together. Moderate skirting was defined as the occurrence of “can” or “may” two to three times individually or four to five times taken together. The past tenses of these words “could” or “might” were not taken into account since they might reflect genuine usage in the context of a study. Since many randomised control trials and other observational studies are often reported in past tenses, these verbs are often used in a genuine context and do not represent skirting. One might say, “study participants could withdraw from the study if they did not agree to participate further in the course” or “the participants could cross over to either arm of the study”. Since such usages do not come under the ambit of skirting, the past tenses have been omitted from this study.

The aim of this study was to find out the prevalence of skirting in the dermatology literature. All types of articles enlisted in PubMed for which full abstracts are available free of cost and satisfying the search strategy were included. Only English language abstracts were considered. A PubMed search was conducted using the keywords “itraconazole” and “tinea.” Filters were applied for the English language search.

The search returned 997 results. Only those articles for which abstracts were available free of cost and in the English language were further taken into consideration. A total of 753 abstracts were available. After duplicate removal and removal of unrelated articles, only 25 articles were deemed fit. Their abstracts were thus screened for skirting.

The results were analysed using Microsoft Excel version XP, and statistical tests were conducted using SPSS 7. The chi-square test was used to evaluate significance, with a cut-off P < 0.05 taken as a significant level.

The study included 14 review articles (including systematic reviews), seven original articles, three case reports and one randomised control trial. Excessive skirting was seen in five out of 25 articles, giving a rate of 20%. Moderate skirting was seen in a further three articles, that is, in 12%. Together, some degree of skirting was seen in 32% while the remaining 68% did not show skirting behaviour.

Of the articles showing excessive skirting, all five were review articles while among those showing moderate skirting, all three were review articles. Interestingly, none of the case reports, randomised control trials or observational studies did show any evidence of skirting.

Skirting leads to some degree of confusion and also dilutes the importance of research findings. While a certain degree of skirting is inevitable, given the rise of litigations in the current scenario, efforts must be made to avoid skirting, to the extent possible. This is especially true for review articles and guidelines, where the use of “can,” “may,” etc. might lead to a considerable degree of confusion. A threshold regarding the use of “can” or “may” has been defined in the study since one or two such instances of can or may usually be unavoidable and do not represent skirting.

One might argue that the conclusions are always based on results derived from Fisherian and sometimes, Bayesian statistics. Whether the hypotheses are rejected or not, based upon probabilistic assumptions, “can” and “may” in fact, maybe more accurate descriptors than absolute statements. However, as nearly all of the present-day evidence-based medicine is based on Bayesian statistics, one will not be able to make any definitive conclusion and the whole process of therapy will be stalled if one applies statistical principles too rigidly. For example, when one gets a hemoglobin level of seven g/dL in a patient, one immediately begins therapy with blood transfusion or intravenous iron as deemed fit. One fully knows that there is still a five percent chance that the hemoglobin level is normal but the test result is erroneous (since most lab values are normally distributed and therefore based on 95% confidence intervals; which means, for any value, no matter how carefully a test is done, there is a five percent chance of error. If one implements Bayesian statistics too rigidly and gets entangled in a meshwork of “can” and “may” one will never be able to deliver care.

Another example would be finding a haemorrhage in the computed tomography scan of patients who has suddenly lost consciousness due to head trauma. Bayesian statistics would still tell us that there is a five percent chance that the blood seen in a computed tomography scan is not actual blood, but a false positive result. Yet medically everyone knows such patients need immediate treatment and one usually progresses towards key hole evacuation of haemorrhage.

The crux is that such excessive use of “can” and “may” by a too rigid interpretation of Bayesian statistics should be discouraged since that leads us nowhere–we do not expect authors to completely avoid “can” or “may”, but rather to minimise their usage, beyond a certain threshold, since that leads to considerable confusion in decision making. More concrete statements wherever possible is to be encouraged. For example, “tranexamic acid can improve melasma” is better re-written as “tranexamic acid is often useful in improving melasma”.

No previous study on skirting could be found in the dermatologic literature.

A possible solution to the use of skirting is to declare at the start of the article or to put a disclaimer at the end that biological diversity implies that exceptions always exist and that the authors should not be held liable for any harm arising out of anything contained within the article.

Skirting was excessively prevalent in 20% of the articles while moderate skirting was prevalent in 12% of the articles. Most of the skirting occurred in the case of review articles. More large-scale studies are needed to find out the exact prevalence of such phenomenon in the dermatology literature.