Antibiotics for Common Respiratory Tract Infections in Adults

A thorough review of the published information indicates that antibiotics rarely benefit acute bronchitis, exacerbations of asthma and chronic bronchitis, acute pharyngitis, and acute sinusitis, although they are commonly prescribed for these illnesses. Rather than prescribing them for these conditions, practitioners should explain to their patients that antibiotics, which have numerous adverse effects, will not hasten resolution of their symptoms, which will often respond to other medications. Most patients will accept this approach if the clinician addresses their concerns, shows a personal interest in them, discusses the expected course of the ailment, and explains the treatment.

More than 100 years ago, William Osler¹ wrote, "A desire to take medicine is, perhaps, the great feature which distinguishes man from animals." Currently, people have an especial taste for antibiotics, which constitute about 1 of every 7 outpatient prescriptions in the United States.² Among adults nearly one half of prescriptions are for common respiratory tract infections—bronchitis, pharyngitis, sinusitis, and upper respiratory tract infections (otherwise unspecified).³ When seeking medical care for conditions labeled as upper respiratory tract infections, bronchitis, and even the common cold,^3,4 50% to 70% of adults receive an antibiotic prescription, especially if the clinician believes that the patient expects it.⁵

This prescriptive promiscuity, which increasingly involves newer agents with broader antimicrobial activity than the older preparations,² largely explains the escalating antibiotic resistance of common bacterial respiratory pathogens in the community.⁶ In the selective pressure of indiscriminate antimicrobial use, susceptible bacteria succumb; with competition for nutrients and mucosal sites of residence diminished, resistant organisms thrive and proliferate. Consequently, Haemophilus influenzae and Moraxella catarrhalis are now widely resistant to ampicillin. In some areas, many infections with group A streptococci no longer respond to erythromycin and other macrolides, and penicillin is increasingly inactive against Streptococcus pneumoniae, a most worrisome development since few other effective agents exist. These bleak trends have led some infectious disease experts to the gloomy prediction of a "postantibiotic" era; a return to conditions that prevailed before the discovery of antimicrobial agents, with no efficacious therapy for many serious bacterial infections.⁷ Some hope emerges, however, from studies demonstrating that aggressive efforts to diminish the use of antibiotics to which organisms are insensitive may allow susceptible strains to reemerge as the dominant isolates.^8,9 In Finland, for example, a campaign to restrict macrolides led to a decrease in erythromycin resistance among group A streptococci.⁸ A thoughtful approach to reducing the incidence of drug resistance is for clinicians not only to refrain from giving antibiotics for obvious viral diseases, such as the common cold, where they are clearly ineffectual, but also to examine their use in other respiratory tract infections for which they are commonly prescribed. In adults, such ailments include acute bronchitis, exacerbations of asthma and chronic bronchitis, pharyngitis, and acute sinusitis. This article derives data from a MEDLINE search of the English-language literature on topics of "acute bronchitis," "asthma," "chronic bronchitis," "acute pharyngitis," and "acute sinusitis" from 1966-2000, and a review of the pertinent material in the Cochrane Library as of 2001. It focuses primarily on the evidence from randomized, controlled trials, especially, when possible, on recent studies that reflect contemporary conditions.

Clinically, acute bronchitis is a recent, commonly productive cough without evident pneumonia. It lasts approximately 2 weeks in 20% of patients, 3 weeks in 30%, 4 weeks in 30%, and longer in 20%.¹⁰ Typically, 3 to 4 weeks elapse from its onset until patients resume all usual daily activities.¹¹ Diverse viruses presumably cause most acute bronchitis.^12,13Mycoplasma pneumoniae¹⁴ and Chlamydia pneumoniae¹⁵ are each present in about 5% of cases.^16- 20

Some have proposed that respiratory bacteria such as Moraxella (Branhamella) catarrhalis, S pneumoniae, and H influenzae produce acute bronchitis or superinfect an originally viral condition.²¹ Since these organisms commonly inhabit upper airways of healthy adults,^22- 24 their significance as isolates is unclear. No criteria for sputum culture findings distinguish innocent airway colonization from invasiveness, if, indeed, such a process occurs. Thus, the concept of "bacterial bronchitis" from these microbes remains unproven.

Eight placebo-controlled, double-blind trials have evaluated antimicrobial therapy.^10,25- 31 The entry criterion was productive cough, with or without purulence, for less than 2 weeks. The average age was about 35 to 45 years, and all studies enrolled cigarette smokers unless they had chronic bronchitis. Fewer than 10% were febrile (temperatures >38°C).^26,27,31 Outcomes generally included duration of cough, feeling sick, time off work, and sputum production or purulence. In 4 studies, antibiotics conferred no significant advantage.^10,28- 30 In the 4 trials favoring antimicrobials, the differences, though statistically significant, were typically small and clinically unimportant.

This overview is misleading because these studies vary considerably in quality and size. Defects include small numbers of participants,^25,26,28,29 high drop-out rates,²⁵ incomplete data collection,^25,26 clinically meaningless outcome criteria,²⁵ failure to enroll consecutive patients,²⁸ and inappropriate statistical evaluation, such as numerous analyses uncorrected for multiple comparisons.²⁶ Three investigations had numerous participants and reasonable methodological rigor. In one involving 91 patients, 25% with serologic evidence of M pneumoniae infection, erythromycin produced no advantage in duration of cough or chest congestion, improved well-being, and antitussive use, although the antibiotic recipients missed 1 day less of work.²⁷ In another, with 158 patients, doxycycline did not significantly shorten the duration of productive or nocturnal cough, feeling ill, or impaired daily activity.³¹ With doxycycline, daytime cough disappeared 1.5 days faster, although it lasted at least 12 days in each group. In patients older than 55 years, cough lasted 4 days less in doxycycline recipients than in the placebo group, and those with frequent coughing who felt ill at entry returned to normal daily activities 2 days sooner with doxycycline administration. These differences, however, occurred in subgroups defined after the trial ended and emerged from numerous statistical analyses uncorrected for multiple comparisons. The largest trial investigated 212 patients with cough and purulent sputum; doxycycline did not reduce daytime or nocturnal cough, purulent sputum, and time "off-color" or off work.³⁰ Pneumonia was rare (2%), irrespective of the treatment.

In conclusion, although about 70% to 90% of patients seen for acute bronchitis receive antibiotics,^16,32,33 published trials demonstrate no clinically important benefit, even in those with purulent sputum. Four systematic reviews,^34- 37 one³⁷ including results from an unpublished trial, concur in concluding that antibiotics are not warranted for acute bronchitis. Instead of prescribing antibiotics, clinicians should explain that cough persists for 2 to 4 weeks, occasionally longer, and that sputum purulence is unimportant unless other features, such as high fever and chills, suggest pneumonia, which is a rare complication. An antitussive agent, such as dextromethorphan, may relieve symptoms,³⁸ and inhaled bronchodilators help some patients, especially with dyspnea, wheezing, or severe cough.^39,40

Asthma denotes episodic, reversible airway obstruction, usually defined as reduced airflow improved by inhaled bronchodilators. Exacerbations are characterized by increased dyspnea, wheezing, and cough, which is often productive. Microorganisms provoke some attacks: viruses in at least 10% to 55% of patients,^41- 44M pneumoniae in less than 5%,^44,45 and C pneumoniae in 5% to 10%.^46- 48 Investigations have not convincingly incriminated respiratory bacteria such as S pneumoniae.⁴⁴ One study examined putative "infective" asthma—exacerbations with features suggesting a bacterial process, including altered sputum volume or color, radiographically defined sinusitis, nasal congestion or drainage, pharyngitis, and fever.⁴⁹ Twenty-seven patients with asthma and 12 healthy controls underwent transtracheal aspirates to obtain uncontaminated lower respiratory tract secretions. Samples from both groups were sterile or yielded only sparse growth, primarily bacteria of low virulence, such as coagulase-negative staphylococci.

Despite no evidence for bacterial infections in exacerbations, 20% to 50% of patients receive antimicrobials,^41,50- 52 a treatment examined in 1 double-blind trial.⁵³ All 60 hospitalized participants received systemic corticosteroids and bronchodilators for their 71 episodes and randomly received amoxicillin or placebo. This antibiotic improved no outcome measure, including cough severity, dyspnea, wheezing, spirometrically measured airflow, and hospital stay.

In conclusion, without pneumonia, a rare complication, neither obtaining sputum cultures nor prescribing antibiotics is appropriate. Instead, patients should receive bronchodilators and systemic corticosteroids to relieve airway inflammation and obstruction, as authoritative US and British guidelines recommend.^54,55 Neither endorses antibiotic use for uncomplicated exacerbations.

Clinically defined, chronic bronchitis is sputum production on most days for at least 3 months annually for 2 consecutive years.⁵⁶ Patients are usually current or prior cigarette smokers. Most studies characterize exacerbations by 1 or more of these features: purulent phlegm or increased dyspnea, cough, or sputum volume. Investigations implicate viruses in approximately 20% to 50% of attacks,⁵⁷C pneumoniae in about 5%^58,59 (although by serologic examination rather than culture findings), and M pneumoniae in less than 1%.^60,61

Whether respiratory tract bacteria, which chronically colonize many patients' airways, are pathogenic remains controversial. In most investigations, sputum cultures yield S pneumoniae and H influenzae from about 30% to 50% of patients during purulent exacerbations.^61- 67 Two careful prospective studies discovered no significant increase in the number of patients whose sputum grew these organisms during exacerbations when compared with periods of remission,^60,61 and neither the presence nor density of these bacteria consistently correlates with sputum purulence or the development of exacerbations.^60,61,68,69M catarrhalis is present in about 5% to 15% of exacerbations, but usually with S pneumoniae and H influenzae, not as a pure isolate.^65- 67Haemophilus parainfluenzae, common in chronic bronchitis, does not increase during attacks.⁷⁰ Other bacteria identifiable by routine culture findings could explain few, if any, episodes: protected brush samples from bronchoscopy during exacerbations in both outpatient and hospital settings yielded significant growth from the lower respiratory tract in only about 50% of cases, the isolates being primarily those already discussed.^66,67

Studies on the serologic and pathologic changes that occur in exacerbations of chronic bronchitis also fail to provide convincing evidence of a bacterial cause.⁷¹ Another approach to elucidate the role of bacteria is to examine the double-blind, placebo-controlled antibiotic trials in treating exacerbations. Of 11 such investigations,^{62- 64,72- 79} 8 show no statistically significant benefit.^{63,64,72- 74,76,78,79} Unfortunately, these studies vary considerably in quality, size, and outcome criteria, including many subjective ones. A meta-analysis found 6 trials, 2 favoring antimicrobials, that used peak expiratory flow rate as an objective end point and discovered a 10.75 L/min greater improvement in the antibiotic group.⁸⁰ Since the patients' average peak expiratory flow rate was approximately 200 L/min, this result represents about a 5% difference, a finding that is clinically and physiologically inconsequential.⁷¹

The best study supporting antibiotic therapy investigated 173 patients with 362 exacerbations⁷⁷ defined by increased dyspnea, sputum volume, or sputum purulence. Overall, success was significantly greater in those receiving antibiotics (68%) than placebo (55%) and failures were fewer (10% vs 19%). For exacerbations with only 1 feature, antibiotics conferred no clinical advantage, and, with 2 features, it was marginal. With all 3 criteria present, about 40% of exacerbations, antibiotics recipients had greater success (63% vs 43%) and less deterioration (14% vs 30%), although whether these differences were statistically significant is unstated. Another large study, involving 278 patients, however, demonstrated no benefit for amoxicillin vs placebo.⁷⁸ Both trials concur that antimicrobials are not beneficial in mild attacks. Why these 2 studies otherwise disagree is unclear, but potential explanations include differing entry criteria, patient populations, disease severity, and outcome assessments. Another dissimilarity is that the latter study eliminated patients with fever or suspected pneumonia, while 29% of subjects in the study favoring antibiotics reported fever. If some of these patients had pneumonia, a point impossible to ascertain because chest radiographs were not obtained, the benefit for antibiotics is not surprising. The latter trial also excluded patients taking oral corticosteroids, while in the one favoring antibiotics, about 40% of participants received them, but in no systematic assignment, dose, or duration. A disproportionate number of patients receiving these agents in high doses in the various groups might account for some of the differences in outcome.

The issue of corticosteroids illuminates the crux of treating exacerbations: namely, the goal. Patients primarily seek therapy not for altered sputum volume or color but for diminished exercise capacity and increased dyspnea.⁷⁷ Treatment should relieve these symptoms and prevent severe respiratory compromise, the most common reason for hospitalization,⁸¹ by reducing bronchial inflammation and airflow obstruction. Systemic corticosteroids achieve this goal by resolving bronchoconstriction and airway inflammation significantly faster than placebo, both in patients hospitalized with exacerbations^82- 84 and in outpatients.⁸⁵ The outpatient study showed a striking advantage not only in improving oxygenation and peak expiratory flow rate, but also in preventing deterioration and hospital admission.⁸⁵ When patients take systemic corticosteroids for exacerbations, antimicrobials may provide no further benefit. In the large study that favored antibiotics overall,⁷⁷ there was no significant benefit for those concurrently receiving systemic corticosteroids. In another trial, among 71 patients treated with prednisolone, those randomly allocated to antibiotics in addition improved no faster than the placebo group.⁷⁸ This investigation was small, however, and included 10 patients with asthma.

In conclusion, microbiologic studies provide no conclusive evidence that bacteria cause exacerbations, and in most investigations, antibiotics provide no significant benefit. Overall, the preponderance of information indicates that obtaining sputum cultures and prescribing antimicrobials are unnecessary both in mild exacerbations and in the more severe episodes, which should be treated with systemic corticosteroids. This topic warrants further placebo-controlled, randomized trials, however, to clarify the microbiologic characteristics of exacerbations and determine whether any subgroups, particularly with severe disease, benefit from antimicrobial therapy when receiving concurrent systemic corticosteroids.

Infection with Streptococcus pyogenes causes about 5% to 10% of adult pharyngitis in general practice^86- 88 and 35% to 40% in emergency departments.^87,89 Other common causes include other streptococci, viruses, M pneumoniae, and C pneumoniae.⁸⁸ Infection with Neisseria gonorrhoeae, although usually asymptomatic,⁹⁰ constitutes 1% or less of cases.⁹¹

Discussions of pharyngitis have focused on treating S pyogenes. Identifying streptococcal pharyngitis by clinical features^89,92,93 is inaccurate; even when adjusted for disease prevalence,⁹⁴ the criteria are insufficiently reliable for confident management decisions.⁹⁵ Techniques to detect S pyogenes include obtaining pharyngeal culture specimens for everyone or only those with negative rapid test results that demonstrate group A antigen in throat swab findings.⁹⁶ Several strategies for treating acute pharyngitis have emerged. One is to prescribe penicillin for all cases, without obtaining any studies.⁹⁷ Trials examining this approach demonstrate no significant clinical benefit. In one, involving 528 patients, symptoms resolved no more quickly with penicillin than placebo.⁹⁸ Another study randomly assigned 716 patients to 1 of 3 approaches: immediate penicillin, no antibiotics, or delayed therapy—penicillin if symptoms were unimproved after 3 days.⁹⁹ No significant difference emerged for resolution of sore throat, time off school or work, or duration of any symptom, except fever, which abated 1 day sooner with antibiotics.

A second strategy is to treat only those cases of acute pharyngitis whose cultures are positive for S pyogenes. Several studies have shown that in this group antibiotics alter the clinical course little. Sore throat, fever, and cervical lymph node tenderness resolve only about 1 day sooner than with placebo.^100- 102 Furthermore, in the trial reporting these outcomes,¹⁰³ patients receiving antimicrobials did not return to school or work more rapidly and had no reduction in respiratory tract infections over the subsequent 6 months.

A third approach is to treat patients whose clinical features suggest streptococcal infection. One double-blind study¹⁰⁴ included 561 adult patients with sore throat for less than 7 days who met at least 3 of the following 4 criteria⁸⁹: history of fever, absence of cough, pharyngeal exudate, and swollen anterior cervical lymph nodes. Symptoms resolved 1.7 days sooner with 7 days of penicillin therapy than with placebo, and patients resumed usual activities 2 days sooner, but they did not return to school or work more rapidly. Nausea (40%) and abdominal pain (25%) were significantly more frequent in the penicillin recipients than in those receiving placebo (15%), although few discontinued the medications because of these adverse effects. The more rapid resolution of symptoms occurred in those with cultures positive for S pyogenes or when other streptococci were present in at least 3+ growth specimens (into the third inoculation area on the culture plate).

Although penicillin is unhelpful for patients whose cultures specifically fail to grow streptococci,^100,103,105 some pathogens, such as M pneumoniae and C pneumoniae, are susceptible to macrolides but not penicillin. Two trials testing erythromycin when throat cultures grew no S pyogenes, however, showed no important clinical advantage over placebo.^86,106

An argument for detecting and treating streptococcal pharyngitis is to prevent its complications, including the immunologic sequelae of acute rheumatic fever (ARF), especially carditis, acute poststreptococcal glomerulonephritis, and the suppurative disorders, primarily peritonsillar abscesses. The efficacy of treating streptococcal pharyngitis to prevent acute poststreptococcal glomerulonephritis is unknown,¹⁰⁷ although some observations suggest a benefit in epidemics of streptococcal pyoderma.¹⁰⁸ In industrialized countries, acute poststreptococcal glomerulonephritis is quite uncommon following streptococcal pharyngitis: a study among Scottish children in the 1970s estimated that it occurred in about 1 of 17 000 cases of untreated patients.¹⁰⁹ It is even rarer in adults, its frequency and severity are declining,¹¹⁰ and, even if antibiotics were effective, enormous numbers of patients would require treatment to prevent 1 case.

The role of antibiotics in averting peritonsillar abscesses is also uncertain. In a study of 434 untreated patients with sore throat, 1 occurred (0.2%).^111,112 In the trial that used clinical criteria for treatment,¹⁰⁴ 3 (1.7%) of 177 patients given placebo developed this complication. Fortunately, the implications of this infection are currently considerably less alarming than before; nearly all cases satisfactorily respond to outpatient therapy—needle aspiration of the abscess under local anesthesia and antibiotics. Hospitalization and tonsillectomy are rarely necessary.¹¹³

Studies from the mid 20th century, mostly in military personnel, demonstrated that antibiotics prevent ARF,¹¹⁴ providing the rationale for the strategy of detecting and treating streptococcal pharyngitis that has prevailed for decades. Authoritative recommendations from prestigious organizations^95,115 and most experts still endorse this approach, based on the conviction that continued adherence to it remains necessary to control ARF. Some investigators,^114,116 however, have argued that in nations, such as the United States, Great Britain, and Australia, where it is rare, this policy is no longer appropriate, considering its costs (eg, time spent seeing patients with pharyngitis, tests to detect streptococci, antibiotics) and the potential therapeutic complications, ranging from minor adverse effects to fatal anaphylaxis, estimated to occur in about 1 to 2 of 100 000 penicillin recipients.¹¹⁷ Certainly, ARF is very rare in many countries. Outbreaks in the United States during the mid 1980s occasioned concern about a resurgence of the disease⁹⁶; although focal increases occurred; however, no nationwide rise developed. Instead its frequency continued a decline^118,119 that began before the antibiotic era, presumably attributable to 3 factors: better public health, especially reduced household size, which diminishes spread of streptococci; antibiotic administration; and, probably the most important factor in recent decades, decreased prevalence of "rheumatogenic" streptococcal strains likely to cause ARF.¹²⁰

The current risk of adult ARF following untreated streptococcal pharyngitis is unknown, but the estimated incidence among Scottish children during the late 1970s, when its frequency equaled that in US children, was about 1 in 40 000.¹²¹ Since the rate among adults even when the disease was more common was about one fifth of that in children^122,123 and its occurrence is now lower than 2 decades ago, the current probability is considerably less. Whatever the actual risk, enormous numbers of adults with streptococcal pharyngitis would require antibiotics to prevent a single case of ARF, and 3 times as many to prevent carditis, which occurs in about one third of adults with ARF.¹²⁴

In conclusion, controlled trials do not support the policy of prescribing antibiotics to all adults with pharyngitis. Instead, most patients should receive antipyretics, reassurance, and information about the disease's natural history: it lasts about 5 days after consultation, and in almost 40% of patients, symptoms persist even longer.¹¹¹ For those who fulfill at least 3 of 4 clinical criteria, it is unclear whether the modest clinical benefits outweigh the adverse effects for most patients, but 1 reasonable approach is to offer a 7-day course of oral penicillin to those with more severe symptoms. Using antibiotics to prevent acute poststreptococcal glomerulonephritis and peritonsillar abscesses following streptococcal pharyngitis is unjustified because these complications are rare and the benefit of therapy is uncertain.

Most authorities continue to recommend detecting and treating streptococcal pharyngitis to prevent ARF, but in industrialized nations this policy seems outdated. Certainly, however, it remains warranted in those with prior ARF, along with secondary prophylaxis, to prevent recurrent attacks; in epidemics of ARF or in locales where it remains endemic; and in scarlet fever, for which antibiotics provide substantial benefit.¹²⁵

Diagnosing bacterial sinusitis is unreliable without sampling sinus contents by surgery or needle aspiration. Computed tomography demonstrates that nearly 90% of patients with the common cold have radiological evidence of sinus disease that usually resolves or markedly improves in 2 to 3 weeks.¹²⁶ It is thus an acute rhinosinusitis, and many patients with presumed bacterial infection actually have uncomplicated viral disorders. Most cases of bacterial sinusitis, however, probably follow such infections, which obstruct sinus ostia and impair mucociliary clearance. Bacteria colonizing the nasopharynx apparently enter the normally sterile sinuses and, entrapped, provoke inflammation.¹²⁷ The most frequent isolates from needle aspirations, constituting about 60% to 90% of bacteria recovered, are S pneumoniae and H influenzae.^127,128 Viruses are detectable by culture findings in about 15%,¹²⁷ rhinoviruses by reverse transcription polymerase chain reaction in 40%.¹²⁹

Clinicians suspect acute sinusitis when colds or influenza-like illnesses persist for several days, accompanied by nasal congestion, maxillary toothache, sinus discomfort or tenderness, purulent nasal discharge, fever, and headache or facial pain, often worsened on bending forward.^{127,130- 133} In 1 study, clinical findings independently associated with abnormalities, suggesting sinusitis on plain radiographs, included maxillary toothache, visible purulent nasal secretions, history of colored nasal discharge, poor response to decongestants, and altered transillumination.¹³⁰ Predicting the presence or absence of radiographic changes by clinical criteria was excellent at the extremes: sinus radiograph findings were abnormal in 9% of patients with no criterion present, 81% with 4, and 92% with all 5 criteria. With 1, 2, or 3 predictors, however, the probabilities were 21%, 40%, and 63%, respectively, making clinical diagnosis more difficult in these common settings. Moreover, radiographic changes do not reliably identify bacterial sinusitis: only about 60% of patients with abnormal radiograph results have positive culture findings from sinus needle aspiration.¹²⁷

Other studies find no consistently reliable clinical criteria to identify acute sinusitis, including fever, which occurs in 10% to 15% of cases.^130,132,134 Several investigations^131- 134 have also shown that only 45% to 70% of patients with clinically suspected sinusitis have radiographic or needle aspiration evidence of it, and only 30% to 40% have a bacterial infection.

For the 60% to 70% of patients having no bacteria in suspected acute sinusitis, antimicrobials presumably confer no benefit. Moreover, given the pathogenesis, promoting sinus drainage by relieving ostial obstruction may be more important than antibacterial therapy. Six double-blind, placebo-controlled trials evaluated antibiotics in acute sinusitis.^{131,135- 139} Four showed no benefit.^{131,135,138,139} The studies conflict for unclear reasons, but examining the 3 best trials—those with numerous participants and clearly delineated entry and outcome criteria—is illuminating. One supported antibiotics,¹³¹ 2 did not.^135,137 One, using computed tomographic criteria of fluid levels or total sinus opacification, randomly gave 130 patients amoxicillin, penicillin, or placebo.¹³¹ Those receiving antibiotics recovered significantly faster: symptoms lasted a median of 9 days for amoxicillin, 11 days for penicillin, and 17 days for placebo. (In 70 patients with identical clinical features, but with only mucosal thickening on computed tomographic scan, antibiotics were no better than placebo, but the numbers in each group were small.¹³⁹) Another trial randomized 214 patients with abnormal plain radiograph results to amoxicillin or placebo.¹³⁵ No benefit for antibiotic therapy emerged. The third study used clinical criteria: 3 major symptoms (preceding cold or influenza, purulent nasal discharge, maxillary sinus pain on bending forward) or 2 major symptoms plus 1 other complaint (predominant unilateral maxillary pain, toothache, or pain when chewing).¹³⁷ The trial randomized 192 patients to placebo or doxycycline; the antibiotic provided no advantage.

One possible explanation for these discrepant results is that computed tomographic criteria in the first study identified patients more likely to have bacterial infection than the radiographic or clinical criteria of the other 2 trials. Another difference, however, is that the latter studies explicitly prescribed xylometazoline nasal drops (which cause vasoconstriction, reducing mucosal edema) and steam inhalation, while the first neither required nasal decongestants nor specified their type or frequency of use. Since relief of ostial obstruction probably hastens improvement, antibiotics may add nothing to aggressive therapy that decreases mucosal edema.

In all studies, symptoms resolved slowly. On average, they began about 2 weeks before entry into the trials, and, however treated, at least 20% to 30% of patients had significant complaints 7 to 10 days later, some persisting for at least 2 to 3 weeks. In these studies placebo recipients developed no serious complications, such as subdural empyema or chronic sinusitis.

In conclusion, most patients with putative acute sinusitis have no bacterial infection, and antibiotics are unhelpful for most patients with suggestive symptoms, even when suspicious abnormal findings appear on radiographs or computed tomographic scans. Ordinarily, such tests are unnecessary, impractical, and expensive. Instead of antibiotics, most patients should receive nasal drops or sprays (xylometazoline or oxymetazoline), analgesics, and, possibly, steam inhalations 3 times daily. They should understand that symptoms abate slowly; complete resolution may require several weeks, but some improvement usually occurs within a few days. Currently, no clinical criteria reliably define patients who might benefit from antibiotics, but treatment is certainly reasonable for those with high fever, systemic toxicity, immune defects, or features suggesting intracranial or orbital involvement.

In many trials, drug-related complications were much more common with antibiotics, as 3 recent studies of acute sinusitis demonstrate. In 1, doxycycline recipients had a 17% incidence of adverse effects, with 4 (4%) of 98 patients discontinuing the medication, while only 2% reported complications in the placebo group, none stopping the agent for that reason.¹³⁷ In the second trial, adverse effects occurred in 9% in the placebo group vs 28% in the amoxicillin recipients,¹³⁵ and in the third study, they developed in 36% of the placebo group compared with 56% receiving amoxicillin and 59% receiving penicillin.¹³³ Recent investigations of pharyngitis, acute bronchitis, and exacerbations of chronic bronchitis also demonstrated substantially more adverse effects in those receiving antimicrobials.^22,31,78,104 Determining the overall benefits of treating respiratory tract infections, therefore, requires considering symptoms that antibiotics cause, which may be less palatable than those from the original disorder. One study, for example, queried patients about the additional duration of pharyngitis that they would accept compared with the therapeutic risks.¹⁴⁰ They preferred 1.5 to 2.5 days of symptoms to even a 5% chance of a mild penicillin rash (hives).

In a survey, the most common prescribing situation by far that disquieted English practitioners was respiratory tract infections, and the most troubling medicines were antibiotics.¹⁴¹ In these difficult circumstances, patient expectation predominantly dictated whether practitioners provided medications. Indeed, several studies confirm that people expecting prescriptions receive them far more frequently than those who do not,^5,142- 145 even when clinicians consider medications unjustified.¹⁴⁵ The strongest predictor is not the patients' actual attitude, but the practitioners' judgment of their attitude.^5,144,146 Physicians often err,^5,144,146 prescribing medications more often than patients desire^{142,144- 148} or consider appropriate.¹⁴² Moreover, when patients state that they want antibiotics for respiratory tract infections, what they often really want is a medicine to relieve their symptoms.¹⁴⁹ Indeed, they are frequently mistaken about which medications are actually antibiotics.¹⁴⁹

Clinicians may assume that providing antibiotics makes patients happier. Satisfaction, however, usually depends not on whether patients receive antibiotics for respiratory tract infections, even when they expect them,^5,99,150 but on whether practitioners deal with their concerns,⁹⁹ show personal interest in them,¹⁵¹ provide and discuss the diagnosis,^5,151 reassure them that their disease is not serious,¹⁵² and explain the treatment.⁵ Contentment appears therapeutic: patients with pharyngitis improved faster when satisfied, which most closely related to feeling that clinicians addressed their concerns.⁹⁹

Some practitioners may believe that prescribing antimicrobials is quicker than fulfilling these expectations, although no investigations evaluated the time involved with these different strategies or the patients' assessment of such behavior. Two studies of pharyngitis, however, showed that those not receiving antibiotics were less likely to return for subsequent episodes than those who did, although both groups were equally satisfied.^111,153 These trials occurred where patients were unlikely to seek other medical care; rather, they managed their illness by themselves. Physicians who educate patients in this way may actually reduce their workload.

Promiscuous prescribing of antibiotics substantially increases the cost of medical care, but it has a more pernicious effect. Antimicrobials are unique among medicines in that their excessive use, especially of those with a broad spectrum of antibacterial activity, can lead to decreased efficacy, as bacteria become resistant throughout the community. Unsurprisingly, the widespread indiscriminate administration of antibiotics that is common now has diminished the susceptibility of respiratory flora.⁶ To help alleviate this problem, clinicians should avoid prescribing antibiotics for conditions for which they are ineffective, marginal, or unimpressive, reserving them instead for conditions for which they exert substantial clinical impact. Except where noted, the available information indicates that antibiotics provide little or no benefit for disorders reviewed here, which account for nearly half of adult outpatient antibiotic use. Rather than prescribing antimicrobials, practitioners should explain that these ailments are rarely serious; they spontaneously abate, albeit sometimes slowly; antibiotics do not hasten resolution, but often make patients decidedly worse; and treating symptoms by other means frequently helps. For nasal and sinus complaints, vasoconstricting nasal sprays or drops such as oxymetazoline are reasonable, and, in acute bronchitis, antitussives may diminish coughing. Inhaled bronchodilators are often beneficial for dyspnea, wheezing, or severe cough in acute bronchitis and are indicated, along with oral corticosteroids, for exacerbations of asthma and chronic obstructive lung disease.

Using terminology that suggests a viral cause may also help,¹⁵⁴ since many patients understand that antibiotics are ineffective for viral infections: practitioners should use diagnoses such as "viral sore throat" and "chest cold" rather than "acute bronchitis," and "sinus cold" rather than "acute sinusitis." As indicated at the beginning of this article, William Osler¹⁵⁵ acknowledged that humans enjoy taking drugs, but he also said, in a statement remarkably appropriate for antibiotics, "One of the first duties of the physician is to educate the masses not to take medicine."

Accepted for publication July 2, 2001.

Corresponding author and reprints: J. V. Hirschmann, MD, Medical Service (111), Puget Sound VA Medical Center, 1660 S Columbian Way, Seattle, WA 98108 (e-mail: pepsi@u.washington.edu).