The sinusitis debate.

Abstract

To the Editor—I appreciated the delicious (intended?) irony that Garbutt’s placebo-controlled study of sinusitis treatment, published in Pediatrics,1 arrived in the same AAP wrapper as Wald’s article in Pediatrics in Review.2Wald restates her nostrums: diagnose pediatric sinusitis clinically, without sinus films or computed tomography scanning, in children with nasal discharge or cough that persists longer than 10 to 14 days. Treat these patients with amoxicillin, she advises.But Garbutt et al convincingly show that neither amoxicillin nor amoxicillin-clavulanate offers any benefit compared with placebo in routine sinusitis, as Wald would diagnose it clinically.I’ve always agreed in principle with Nelkin, whose “Requiem for the Common Cold” satirically suggests that treating “sinusitis” is just the latest, lousy excuse for dispensing antibiotics for upper respiratory tract infections.3 Treating colds with antibiotics is an embarrassing and, thanks to recent studies, poorly kept secret.4,5 If you can prescribe antibiotics after 10 to 14 days, the reasoning goes, why not save the hassle of a return visit or phone call and use amoxicillin after 5 to 7 days for “early sinusitis.” Parents certainly lead the way down this slippery slope when they insist that “last time the cold turned into sinusitis.”I wish that I could say I always practice in a high-minded manner, eschewing antibiotics and engaging parents in lengthy discussions about proper antibiotic use that may leave them confused and dissatisfied. Thanks to Garbutt et al, I now have greater impetus to do the right thing.To the Editor—I read with great interest the article by Garbutt et al.1 The study was undertaken to establish whether there is any clinical benefit to antimicrobial treatment of children who are diagnosed by clinical criteria (without images) to have acute uncomplicated sinusitis. As it is known that many children with acute bacterial sinusitis improve spontaneously, the challenge for the clinician is to identify the subset of children who are most likely to benefit from antimicrobial therapy.2 The question posed by the authors is especially important in light of recent recommendations made by several organizations to diagnose acute uncomplicated sinusitis on clinical grounds alone without the use of images.3,4 The rationale for the abandonment of imaging is that plain radiographs of the paranasal sinuses are technically difficult to perform, particularly in very young children. Correct positioning may be difficult to achieve, and therefore the radiographic images may both overestimate and underestimate the presence of abnormalities within the paranasal sinuses.3,5,6 Furthermore, the performance of images dramatically increases both the cost and complexity (general unavailability of radiographs in the primary care setting) of making a diagnosis of sinusitis.Garbutt evaluated children between the ages of 1 and 18 years with respiratory symptoms of between 10 and 28 days. No images were performed. Children were randomized to receive either low-dose antibiotic or placebo. No differences were observed in outcome, either in the timing or frequency of recovery, between children treated with antibiotics and those treated with placebo.Previous studies have clearly demonstrated that the paranasal sinuses of children with persistent respiratory symptoms (nasal discharge or cough or both for >10 but <30 days that are not improving) and significantly abnormal radiographs (complete opacification, mucosal swelling of at least 4 mm, or an air-fluid level) are infected with a high density of bacteria.7 A similar placebo-controlled trial, reported in 1986, evaluated children with both respiratory symptoms and abnormal radiographs and demonstrated that children with acute sinusitis who were treated with an antibiotic recovered more quickly and more often than children receiving placebo.2 This same study showed that a history of persistent respiratory symptoms predicted significantly abnormal radiographs in 88% of children ≤6 years old. However, in older children (>6 years old), only 70% with a history of persistent symptoms had abnormal films.2Differences in study design between the Wald and Garbutt studies are highlighted in an effort to understand the differences in outcome: 1) Entry criteria for the Wald study were specifically stated as nasal discharge or cough or both for at least 10 but no longer than 30 days that did not seem to be improving. Although the latter criterion (not improving) is implied in the Garbutt study, it is not specifically stated in their “Methods” section. Respiratory symptoms that have begun to resolve, although persisting beyond 10 days, do not merit antibiotic therapy. 2) There may have been a difference in the severity of illness of the children included in each study. The Garbutt study excluded children with a temperature of >39°C, facial swelling, or facial pain. Wald excluded children with severe headache or periorbital swelling. Exclusion of sicker children, those most likely to benefit from treatment, may bias results in favor of no difference. It would be important to examine the symptom scores from Garbutt’s article to determine if there is a degree of severity of symptoms that predicts response to antibiotic treatment. 3) Confirmatory radiographs were not performed in the Garbutt study. In the Wald study, normal radiographs eliminated 20% of the initial study cohort overall and 30% of the children >6 years old. Accordingly, the inclusion of an older age group in the Garbutt study (mean age: 8 years), compared with the Wald study (mean age: 6 years), likely resulted in the inclusion of a substantial number of children without sinusitis. 4) Children in the Garbutt study were permitted to use symptomatic therapies while those in the Wald study were not. Use of symptomatic therapies may obscure the effect of antibiotic treatment. 5) The final issue that may influence the interpretation of Garbutt’s findings is antibiotic selection and dose. Current recommendations for treatment of children with uncomplicated sinusitis vary, depending on the previous history of antibiotic exposure (in the last 1–3 months), attendance at day care, and age. Many of the children in their study might have qualified for high-dose amoxicillin/clavulanate to overcome either penicillin-resistant Streptococcus pneumoniae or β-lactamase-producing Haemophilus influenzae and Moraxella catarrhalis. Amoxicillin at 40 mg/kg/day in 3 divided doses and amoxicillin/clavulanate at 45 mg/kg/day in 2 divided doses, appropriate in 1986, may have been inadequate in 1999.The authors have investigated a very important question. In planning the next examination of this issue, we must standardize our approach in the following way: To the Editor—I read with interest the article by Garbutt et al.1 I commend the authors for an excellent study design with utilization of validated outcomes instruments for assessment of the children with sinusitis. However, as the ancient Greeks were fond of noting in their powerful characters, a hamartia or fatal flaw exists: the validity of the diagnosis of acute sinusitis used in this study.Although the authors have attempted to use clinical guidelines to diagnose acute sinusitis, it is important to point out that physicians need to rely on clinical judgment when using any set of guidelines; injudicious application of guidelines will likely result in overdiagnosis of sinusitis. Indeed, the diagnosis of acute bacterial rhinosinusitis (ABRS) in adults and children is a difficult one. Most physicians acknowledge that duration of upper respiratory tract symptoms beyond 10 days increases the likelihood of ABRS,2 but transition from viral infection to bacterial infection may occur at any time and is unpredictable. If one chooses as a primary focus symptom duration for the establishment of the diagnosis of ABRS, it is instructive to note that for patients with confirmed viral rhinosinusitis, 20% to 30% will continue to exhibit cough and nasal drainage beyond 14 days of infection.3 With the average child experiencing 3 to 8 acute viral illnesses a year and ABRS complicating only roughly 0.5% to 2% of these viral rhinosinusitis episodes,4,5 we see that establishment of the diagnosis of ABRS based primarily on symptom duration will vastly overestimate the number of cases of true ABRS. Without either maxillary sinus puncture with positive cultures or, better, using minimally invasive endoscopic examination with positive cultures, the diagnosis of acute sinusitis in this article is suspect at best. Divergent diagnoses such as viral rhinosinusitis, nasopharyngitis, or adenoiditis will all be mistaken for acute sinusitis.Although we note that injudicious application of clinical guidelines will likely overdiagnose ABRS, similarly troubling is the incomplete application of guidelines leading to an underdiagnosis of the disease. In this article the authors chose to manipulate guidelines to exclude children with more significant symptoms such as fever, facial swelling, and facial pain, all of which are common features of true ABRS. Indeed, features such as these have been suggested to be “major factors” and useful indicators of ABRS, as suggested in the guidelines of the Rhinosinusitis Task Force of the American Academy of Otolaryngology/Head and Neck Surgery.6Unfortunately, with the diagnosis suspect, the foundation of this article begins to crumble. The title of the article is simply incorrect. Acute sinusitis has not been confirmed in these children, simply upper respiratory illness, and the implication that true acute bacterial sinusitis does not warrant antibiotics is not supported by the data presented. With the serious risks of ABRS including meningitis, brain abscess, and vision-threatening orbital infection, such implication is not only disingenuous but also downright dangerous. What is established by this article is that children with runny nose and cough lasting 10 to 28 days without associated facial pain or swelling or fever do not require antibiotics, an assertion with which few would disagree.In Reply—We thank Drs Harris, Wald, Senior, and Shores for their letters. We are delighted to provide Dr Harris and others with evidence to support the judicious use of antimicrobials. Data from an ongoing study support Dr Harris’s supposition that antimicrobial treatment for acute sinusitis is commonly initiated when symptoms have been present for <10 days. Chart reviews of children diagnosed with acute sinusitis by 29 primary care pediatricians revealed that 146 (42%) of 339 children in whom symptom duration was recorded had symptoms for 7 days or less. All were treated with an antimicrobial. Results from our study1 would suggest that antimicrobial treatment is unlikely to offer any clinical benefit to the majority of these patients.Drs Wald, Senior, and Shores raise concerns about the diagnostic criteria used to define our study population. The goal of our study was to determine the effectiveness of treatment with antimicrobials with known efficacy against likely bacterial pathogens in the primary care management of children with acute sinusitis. Therefore, the study population was comprised of patients with a clinical diagnosis of acute sinusitis made by their primary care pediatrician. Other than symptom duration (10–28 days) and severity (S5 score of at least 1), diagnostic criteria were not specified to participating pediatricians. However, a survey to assess the frequency of use of accepted diagnostic criteria was completed by study pediatricians after all study patients were enrolled. All reported use of the following diagnostic criteria all the time, or most of the time—failure of symptoms to improve, symptom duration, presence of specific symptoms, and symptom severity.We agree that a definitive diagnosis of acute bacterial sinusitis can be made only with a positive sinus culture, and that use of clinical diagnostic criteria will necessarily lead to overdiagnosis. However, sinus aspiration and endoscopic examination are neither readily available nor recommended for use in the primary care management of children with acute rhinosinusitis. Drs Senior and Shore’s suggestion that patient selection criteria include a positive culture result are neither feasible nor appropriate to answer our study question. Likewise, radiological confirmation of sinus disease was not a study entry requirement as sinus radiograph is not recommended for use in primary care and is inaccurate for diagnosis of acute bacterial sinus infection.2Patients with fulminant sinusitis2 were excluded from the study, as patient management decisions are different for this population and require separate study. This clinical presentation is uncommon in community practice, and there is little debate concerning the use of antimicrobials. We were interested in investigating the more controversial area of management of patients with the typical presentation of persistent, nonspecific, upper respiratory symptoms. This said, many patients in our study had severe symptoms (52% had an S5 score of ≥2 [possible range 0–3]), 25% had missed school or day care for at least 2 days, and 68% had used a symptomatic treatment before the initial office visit. Ninety-four patients (58%) complained of facial pain or headaches.Dr Wald suggests several reasons for the discrepancy between our findings and those from her study.3 The first difference is the population studied. Following current management recommendations,2 we did not require radiological confirmation of disease. This increases the generalizability of our study findings to patients managed by primary care pediatricians. All study patients met the clinical criteria proposed by Dr Wald—nasal discharge, daytime cough, or both persistent for at least 10 days and not improving. We included patients aged 1 to 18 years, as pediatricians routinely treat patients in this age group. The second difference is that use of symptomatic therapies may obscure any antimicrobial benefit. We think use of symptomatic treatments was not responsible for our finding of no clinical benefit with antimicrobial treatment, as the proportion of children who used additional treatments was the same in all treatment groups. We allowed additional symptomatic treatment because this comprised usual care, and study of antimicrobials used alone would not answer the clinically relevant question. We thought that use of symptomatic treatments would be unlikely to introduce a systematic bias. Elimination of such treatment could adversely affect patient accrual or result in illicit use, precluding assessment of any potential bias. Finally, Dr Wald suggests that the dosing of amoxicillin may have been inadequate to treat resistant bacteria. This would be of most concern for Streptococcus pneumoniae, the most common bacterial pathogen in acute sinusitis and the least likely to spontaneously resolve.4 Current treatment recommendations are to use amoxicillin, 40 mg/kg/day, unless risk factors for penicillin resistant S pneumoniae are present (recent antimicrobial use, daycare attendance, and age <2 years).4 Patients with recent antimicrobial use were excluded from the study. We repeated the analysis excluding patients with the other 2 risk factors (n = 38, evenly distributed among treatment groups), and our results were the same: no clinical benefit of antimicrobial treatment for children with clinically diagnosed acute sinusitis.Dr Wald’s suggested criteria for future studies may be appropriate for investigating the pathophysiologic response of bacterial infections to antimicrobials. However, to understand the optimal management of patients in community settings, we advocate instead including all patients for whom there is clinical controversy (including older children) and excluding those for whom there is not (such as those with fulminant infection). We also support testing the effectiveness of antibiotics in the doses commonly used, and with reasonable, safe, and effective co-treatments that are often used in community practice.We suggest that future research in this area have 2 main foci. First, we should improve our ability to detect the relatively few children who will benefit from antimicrobial treatment, and determine their optimal treatment. Second, we need to identify effective treatments to relieve symptoms in all children. In both cases, study patients need to be representative of the patient population in whom the results will be applied, namely children attending the pediatrician’s office with symptoms of rhinosinusitis.