Commentary

Acute sinusitis is associated with substantial expenditures on antibiotics.¹ Because of increasing antibiotic resistance, there is pressure for accurate diagnosis to prevent inappropriate prescribing. Whether the review by Lindbæk and Hjortdahl provides such information is moot. Firstly, their search strategy was restricted to Medline and to English language studies. The authors suggest that they were unlikely to find any other studies because 2 other recent reviews had not.^2–,³ However, these reviews were also language restricted, 1 to English and the other to Northern European languages. Indeed, 1 of the previous reviews identified 2 English language reports in general practice populations that were not used by Lindbæk and Hjortdahl.² Secondly, the review does not report the processes used to minimise bias during data extraction, which introduces another uncertainty as there is a discrepancy between the description of characteristics of 1 study included in this and an earlier review. Thirdly, although the authors provide useful information about the association of purulent rhinorrhoea, pain in teeth, and elevated erythrocyte sedimentation rate with the diagnosis of purulent sinusitis, they only report positive likelihood ratios (+LRs). A negative likelihood ratio helps to reveal the probability of a false negative, giving information on how good a sign is at excluding patients without the condition. +LRs are only half the picture.

What does this review contribute to clinical practice? A 1999 systematic review found that antibiotic treatment is more effective than placebo for treating acute maxillary sinusitis, and that the penicillins are as effective as modern antibiotics.⁴ However, 62% of the placebo group improved or healed, which lead the Cochrane Consumer Network to conclude that antibiotics may help some patients, but will not make a major difference to most. Delayed prescription may be an option for both diagnosis and treatment.