1. Given a clinical diagnosis of acute bacterial rhinosinusitis, what are the comparative efficacies of the antibiotics in resolving symptoms and preventing complications or recurrence?

We identified a total of 39 randomized controlled trials including 15,739 patients from 1997 to 2004 that studied antibiotic comparisons in treatment of acute bacterial sinusitis (Tables 1 & 2). With the exception of 5 studies that did not provide the information, all the studies were either funded by pharmaceutical companies or had authors associated with the pharmaceutical industry. Twelve of the studies consisted of subjects from the United States. Ten out of 39 studies included subjects less than 18 years old. No study was explicitly restricted to the pediatric population, although one study consisted only of subjects whose age ranged from 6 months to 17 years. Sample size of the studies ranged from 40 to 1,798. Of 13,220 patients whose per-protocol results were reported in the studies, less than 3 percent received placebo. All studies stated explicit requirements of clinical signs and symptoms of acute sinusitis for entry into the studies. In addition, 33 studies included results from either radiography or computed tomography (CT) scan of the sinuses as part of their eligibility criteria. Four studies are considered to be superior in methodological, reporting and data quality (A). Twenty-two studies are considered moderate (B) and nineteen studies low quality (C). The low quality studies often had limitations due to high dropout rate and incomplete reporting of data.

Table 1

Placebo-controlled antibiotic trials for the treatment of acute bacterial rhinosinusitis from 7/1997 to 8/2004.

Table 2

Antibiotic-comparison trials for the treatment of acute bacterial rhinosinusitis from 7/1997 to 8/2004.

The antibiotics studied are listed in Tables 1 & 2. The classes of antibiotics studied consist of penicillins, cephalosporins, macrolides, azalides, ketolides, quinolones, carbapenems and tetracyclines. There are a total of 112 comparisons reported in the 39 trials (Table 4); 7 compare antibiotics to placebos; 5 compare various antibiotics (including 2 placebos) to amoxicillin, 22 compare various antibiotics (including 1 placebo) to amoxicillin/ clavulanate; 10 compare various antibiotics to cefuroxime. In contrast to the previous evidence report, there was no comparison against trimethoprim/sulfamethoxazole.

Duration of treatment varied between 3 days and 4 weeks. Twenty-six of the studies included at least one antibiotic that was prescribed for 10 days. Primary outcome assessment took place anywhere from 3 days to more than 4 weeks after the initiation of treatment.

Results of Meta-analyses

We performed 13 different meta-analyses to answer Research Question 1 regarding the comparative efficacy of different antibiotics on treatment failure and recurrence rates (Table 5). All meta-analyses were reported using per-protocol data from the primary studies. Meta-analyses were also performed substituting per-protocol data with modified intention-to-treat data available in 5 studies; results were similar and are not reported here. None of the cumulative meta-analyses ordered by study methodological quality demonstrated any alteration of treatment effect trend by the addition of studies with lower quality scores to those with higher quality scores. As a result, studies of all methodological quality are included in these meta-analyses.

Table 5

Per-protocol meta-analyses of treatment failure & recurrence with different antibiotics.

Placebo-controlled trials. There were 5 trials (7 comparisons, total of 780 enrolled patients) comparing antibiotics to placebo. All of these trials recruited patients from a primary care setting. Four of the 5 trials used an antibiotic in the penicillin class, while the fifth trial compared azithromycin to placebo. Overall, antibiotics were more effective than placebo, reducing the risk of clinical failure by about 25% to 30% 7 to 14 days after treatment initiation (risk ratio [RR] 0.69, 95% confidence interval [CI] 0.53–0.89, Table 5). Nevertheless, symptoms improved or were cured in 65% of patients without any antibiotic treatment at all (95% CI 40–91%).

Antibiotic comparison trials. Five studies, involving a total of 3033 patients, compared various quinolones to cefuroxime. Except for one study, all antibiotics were given for 10 days. In the four studies that reported data for outcome assessment between 11 to 26 days after initiation of treatment, there was a non-statistically significant trend suggesting that quinolones were superior to cefuroxime in reducing clinical failure (RR 0.68, 95% CI 0.44 to 1.04).

There were four studies involving a total of 2765 patients, which showed that amoxicillin/clavulanate, when compared to antibiotics in the cephalosporin class, was 41% more effective in reducing clinical failure 10 to 25 days after treatment initiation (RR 1.41, 95% CI 1.08 to 1.82). In absolute terms, this means treating 100 patients with antibiotics in the cephalosporin class will lead to 3.5 more failures (95% CI 0.9 to 6.0) as compared to amoxicillin/clavulanate. However, data from four studies involving a total of 2797 patients did not show a significant difference in recurrence rates between amoxicillin-clavulanate and cephalosporins (RR 1.10, 95% CI 0.83 to 1.45) 24–45 days after treatment initiation.

There was no consistent trend observed when comparing amoxicillin/clavulanate, cephalosporins or quinolones to the group encompassing macrolides, azalides and ketolides.

1a. Is there evidence that duration of antibiotic treatment in acute bacterial rhinosinusitis affects efficacy?

There are eight studies that reported data on comparisons of the effect of treatment duration on outcomes (Table 6). One study that compared 10 days with 5 days of amoxicillin/clavulanate 500 mg tid reported a statistically non-significant 27% reduction in clinical failure rate³. Another study of ceftibuten concluded that 20 days of treatment may be more effective than either 10 days or 15 days regimen (0% failure rate vs. 8% vs. 8%, respectively)⁴; however, the study did not report the actual number of patients who completed the study. Two studies of 10 days vs. 5 days of telithromycin reported that the clinical failure rate between the two treatment durations were comparable.⁵, ⁶ The studies on gemifloxacin (5 days vs. 7 days)⁷, azithromycin (3 days vs. 6 days)⁸ and gatifloxacin (5 days vs. 10 days)⁹ showed therapeutic equivalence of the 2 durations. One study compared nasal smear findings for certain periods after different durations of antibiotic treatment and concluded that at least 2 weeks of antibiotics would be an appropriate treatment duration for acute maxillary sinusitis because the average nasal smear score (derived from number of neutrophils) was significantly different beginning from study day 21 between the 7-day-antibiotic group and the other groups. ¹⁰

Table 6

Comparing different durations of treatment in sinusitis.

2. What adverse effects are reported for antibiotics used for acute bacterial rhinosinusitis?

Thirty-four of the comparative trials and five additional non-comparative trials reported adverse events. Descriptions of adverse events were diverse among studies. Almost all reported probable treatment-related adverse events, but many of them did not state the criteria for determining whether an event was considered likely treatment-related or not. Several studies also graded the severity of events; however, no study gave clear criteria for the grading scale. Other studies reported serious adverse events only. Few studies differentiated between severe (a description of degree) and serious (life-threatening, disabling or requiring prolonged hospitalization) events. Virtually all studies examined only short-term adverse events with assessment ending at the conclusion of patient follow-up 2 months or less after treatment initiation.

The overall percentage of subjects who reported at least one adverse event varied from 3% to 88% (Table 7). In general, an average adverse event rate of 15% to 40% of subjects was observed for the different classes of antibiotic. The adverse event rate for placebo was in this range, as well. Severe adverse events were rare, occurring in 0 to 7.7% of subjects. Severe adverse events included diarrhea, abdominal pain and nausea on amoxicillin/clavulanate; abdominal pain, diarrhea, constipation, urticaria, headache/dizziness, loss of appetite/disorientation/insomnia and vaginitis/monilia on levofloxacin; headache, asthenia, diarrhea, vomiting, dizziness and agitation on moxifloxacin; vaginitis, headache, nausea, diarrhea, arthralgia, increased cough and dyspnea on cefuroxime; diaphoresis/rash and loss of appetite/disorientation/insomnia on clarithromycin; and increased coagulation test on faropenem. Very few specific serious adverse events were reported by studies. Reported serious adverse events included diplopia on amoxicillin/clavulanate; myocardial infarction, lumbar disk lesion and neuropathy on levofloxacin; asthma on sparfloxacin; allergic reaction, facial/tongue edema, hepatitis, asthma and convulsion on gatifloxacin; maxillary antral abscess, convulsions, and collapse during local anesthesia on clarithromycin; facial edema on ciprofloxacin; amblyopia, ischemic heart disease and maxillary sinus surgery on cefuroxime; and tachycardia on moxifloxacin. Discontinuation due to adverse events was uncommon with fewer than 10% of subjects removed from any trial due to adverse events.

Table 7

Summary ranges of percent of patients experiencing adverse events with placebo and each antibiotic.

It is difficult to compare the rates of occurrence of particular adverse events by antibiotic class given the heterogeneity among studies. Overall, the most common events involved the gastrointestinal system, specifically reports of nausea or vomiting, diarrhea and abdominal pain. Central nervous system adverse events, mostly complaints of headaches, were also common. Some subjects in each antibiotic class reported skin disorders, such as rash and photosensitivity. Taste perversion seemed to be a problem specific to clarithromycin administration with anywhere between 8% and 21% of study subjects reporting this complaint.

Cardiovascular problems were a particular concern to investigators of quinolones given their association with prolongation of the QT_c interval on electrocardiogram. Nevertheless, cardiac-related adverse event rates were low with quinolones, as well as for all classes of antibiotics. In the study conducted on 10,822 subjects with sinusitis by Faich et al. ¹¹, an independent safety committee was convened to search for a link between moxifloxacin and cardiac-related events. Investigators asked patients specifically about symptoms suggestive of a possible cardiac event such as chest pain or tachycardia; however, electrocardiograms were not routinely collected. The committee concluded that there was no evidence of increased mortality or detectable treatment-associated ventricular tachyarrhythmias in that trial. In a study of 253 patients treated with sparfloxacin conducted by Garrison et al.¹² which did collect electrocardiograms on all patients, there was a mean increase in the QT_c interval from baseline to day 4 of 0.010±0.024 sec., but no cardiovascular adverse events related to this increase.

Some women in each antibiotic class experienced vaginal moniliasis. Comparison of the rates of moniliasis between classes is problematic as the incidence reported by some studies may be under-estimated. Some reports are unclear as to whether they excluded men from the denominator in their calculations. Other studies did not specifically report the incidence of vaginal moniliasis but the incidence of vaginitis or urogenital complaints.

Tables 8 to 11 present the specific adverse events reported in each antibiotic class.

Table 8

Adverse events for placebo and penicillins, N (%).

Table 9

Adverse events for cephalosporins and faropenem, N (%).

Table 10

Adverse events for macrolides, azalides and ketolide, N (%).

Table 11

Adverse events for quinolones, N (%).

3. How does the introduction of the pneumococcal vaccine affect the resistance patterns of pneumococcus and the treatment decisions in acute bacterial rhinosinusitis?

We did not identify any article in our literature search that directly addressed this question. Streptococcus pneumoniae is one of the most common pathogens identified in acute bacterial sinusitis. In the early 2000, a 7-valent pneumococcal conjugate vaccine (PCV7) was approved for routine administration to infants and children in United States. In the same year, the American Academy of Pediatrics (AAP) recommended routine administration of PCV7 to all children 23 months and younger.¹³ It is therefore, important to monitor the changes in serotypes and antibiotic susceptibility of S. pneumoniae as a result of the introduction of PCV7 to the population. Furthermore, recommendations for treatment of acute bacterial sinusitis may well have to be modified depending on the results of the surveillance. Surveillance data are already appearing demonstrating the changing epidemiology of S. pneumoniae serogroups after the introduction of PCV7. Data collected from the US Pediatric Multicenter Pneumococcal Surveillance Group reported that before the licensure, nonvaccine serogroups accounted for 6% of the isolates recovered from children ≤ 24 months old; in 2002, nonvaccine-serogroup isolates were 37.6% of the total isolates in this age group. Also, among the isolates of S. Pneumoniae belonging to the serogroups contained in PCV7, the proportion that were nonsusceptible to penicillin decreased from 54% in 2001 to 43% in 2002.¹⁴ Another report reported that rate of invasive pneumococcal disease decreased from an average of 24.3 cases per 100,000 persons in 1998 and 1999 to 17.3 per 100,000 in 2001. The largest decline was in children under two years of age. Disease rates also fell for adults. This observation suggests that the use of pneumococcal vaccine in children may be reducing the rate of disease in adults as well.¹⁵ A randomized controlled trial between 1995 and 1999 involving 1,662 infants reported that the heptavalent pneumococcal vaccine reduced the number of episodes of acute otitis media by 6% (95% CI -4 to 16%) and also reduced the number of episodes due to the serotypes contained in the vaccine by 57% (95% CI 44 to 67%), whereas the number of episodes due to all other serotypes increased by 33%.¹⁶ This suggests that the impact of pneumococcal immunization on acute sinusitis in adults will also depend on the virulence and resistance patterns of the serotypes that replace those contained in the vaccine.