Predictors of obstructive sleep apnea in snorers: statistical errors.

To the Editor: I have read with great interest the recently published article by Ibrahim and colleagues1 in the Annals of Saudi Medicine and I appreciated the authors’ efforts and work. However, I would like to make few comments on it because of the apparent flaws in results.

First, the authors mentioned in the abstract that 39% of the females had obstructive sleep apnea (OSA) with a mean of respiratory disturbance index (RDI) of 27.8±26.5. In Table 1, it was shown that 17 of 41 females were RDI positive, which constituted 41.5% and not 39%.

Second, in the results text, the authors stated that “OSA was seen more in non-Qatari (n=84) subjects versus in Qatari subjects (n=38). The sum of both groups was 122, not 126 study participants, who were diagnosed as OSAso unless there were 4 participants of unknown nationality, there is an error. Third, the authors again contradict what was shown in Table 1 where they mentioned that females (of the OSA group) had a non-significant higher mean age of 51.8±8.2 years compared to 48.7±9.4 years in males. The table showed that figure 51.8±8.2 years belonged to females in the RDI negative group and not the OSA group, whereas the second figure did not exist at all. Moreover, the statistical analysis in the table compared between the age of OSA group and the negative group among males and females separately.

Fourth, in the discussion, on page 424, the authors stated that “the frequency of PSG diagnosed OSA. was 72.4%”, whereas the correct figure was previously mentioned by them as 66% (126 cases of 191 participants).

Fifth, in the discussion, the authors more than once, used the word “several studies”, whilst mentioning only a single study (as in line 3 and line 8 in the second column of page 424). Moreover, they ignored explaining an important finding in their study, which was the nonsignificant association of age with the degree of severity of OSA as shown in both the bivariate and multivariate analysis in Tables 2 and 3.

Finally, I would like to stress that statistical input to medical research is widely recommended but inconsistently obtained. Therefore, authors should involve statisticians in all the steps of their research and recognize them by either authorship or acknowledgment.2