Keratitis and other potential complications of acute bacterial conjunctivitis are seen very seldomly in otherwise healthy subjects. Extensive clinical experience has thus proven acute bacterial conjunctivitis to be a disease with a highly favourable prognosis and a high frequency of spontaneous cure.
A meta-analysis published by Sheikh & Hurwitz (2001), based on the few placebo-controlled, randomized, double-blind studies available at that time, showed clinical remission within 5 days in 64% of placebo-treated patients. However, the same analysis also showed that topical antibiotic treatment resulted in a significantly improved early clinical remission, as well as improved early and late microbiological remission. Because this meta-analysis only included patients recruited from specialist care populations, the results are not necessarily representative for primary care patients with suspected acute bacterial conjunctivitis. Therefore, a well-designed and well-controlled study on acute infective conjunctivitis in children aged 6 months to 12 years recruited from 12 general medical practices in the UK was recently carried out (Rose et al. 2005). At baseline, conjunctival swabs yielded one or more of the study-designed pathogenic bacteria (H. influezae, S. pneumoniae and M. catarrhalis) in 78% of the patients, while adenovirus or picornavirus was recovered alone or together with the above-mentioned bacteria in 13% of the cases. Clinical cure was recorded in 83% of the placebo group within 7 days, compared with 86% of those receiving chloramphenicol eye drops. The so-called number needed to treat (NNT) to achieve one more clinical cure than that occurring spontaneously was calculated to be 25. Corresponding frequencies of clinical cure on day 7 in patients growing one or more of the study-designed pathogenic bacteria (H. influenzae, S. pneumoniae and M. catarrhalis) were 80% and 85%, respectively (NNT = 22). When the proportion of children achieving clinical cure each day in the chloramphenicol and placebo groups were compared, a significantly better response was obtained in the chloramphenicol group at day 2 (26.4% versus 15.9%); this difference remained statistically significant up to day 7, although the mean difference in the time to cure in the two groups was only 0.3 days during this period. Bacterial eradication at day 7 was also obtained significantly more frequently in the chloramphenicol group (40.0%) than in the placebo group (23.2%), resulting in NNT = 6. The frequency of new conjunctivitis episodes during the following 6 weeks was almost identical in the two study groups. A weakness of this important study is a possible selection bias: only one third of the children presenting to the participating GPs with acute infective conjunctivitis during the trial period were included in the study. Thus, family doctors may conceivably have recruited the less severe cases for the study, believing that patients with more severe symptoms needed topical antibiotics. In addition, patient recruitment only took place during office hours, while one may speculate if patients presenting out of office hours generally will have more severe symptoms and a more abundant conjunctival flora of pathogens.
Another double-blind, randomized and placebo-controlled study on infectious conjunctivitis in primary care was published by Rietveld et al. (2005). Adult patients presenting at 25 Dutch GP centres with a red eye and either (muco)purulent discharge or ‘sticky eyelids’ were invited to participate. At baseline, only 50 of the 163 patients (30.7%) completing the trial were culture-positive. Based on both the clinical examination by the GPs after 7 days and the daily diary kept by each of the patients, the efficacy of fusidic acid gel compared to placebo gel was evaluated. At this stage, clinical cure was recorded in 62% of the fusidic acid group and 59% of the placebo group. Neither the severity nor the duration of symptoms differed significantly in the two patient groups. However, the treatment effect appeared stronger in culture-positive patients, where bacterial eradication in the fusidic acid and in the placebo group was obtained in 76% and 41%, respectively, although more than 60% of the cultures obtained at baseline showed in vitro resistance to fusidic acid. Minor adverse events, mainly a burning sensation after instillation of the study medication, occurred in 14% of the fusidic acid group and in 3% of the placebo group. The main weaknesses of the study were the limited number of culture-proven cases at baseline and the fact that patient recruitment was performed during office hours only. The authors concluded that although the study lacked the power to conclusively demonstrate equivalence between fusidic acid and placebo, it did not support the current GP prescription practice of using topical antibiotics in the vast majority of patients with suspected infective conjunctivitis.
In another interesting, open, randomized and controlled study from 30 general practices including 307 children and adults with acute, presumably infective conjunctivitis, the results obtained by no treatment, delayed topical antibiotic treatment or immediate topical chloramphenicol treatment were compared (Everitt et al. 2006). The different treatments did not influence the severity of symptoms during the first 3 days, but the duration of moderate symptoms was shortest in the immediate treatment group (3.3 days) and longest in the no treatment group (4.9 days). In the delayed antibiotic treatment group, 53% eventually received topical antibiotics. When asked about their belief in the effectiveness of antibiotics for acute conjunctivitis, as well as their intention to reattend for eye infections, those receiving immediate topical antibiotic therapy had the highest positive score. In accordance with this, the reattendance rate within 2 weeks was lower in the delayed treatment group than in those receiving immediate treatment. The distribution of an information leaflet about acute conjunctivitis did not significantly influence the attitude of the patients regarding acute conjunctivitis. The authors judged a delayed prescription attitude to be the most favourable alternative in many respects, reducing both the use of topical antibiotics and the frequency of early reattendance without increasing the severity of the symptoms.
Despite the clinically experienced and now also well-documented high frequency of spontaneous cure (Rietveld et al. 2005; Rose et al. 2005; Everitt et al. 2006), all questions about the risk of serious adverse events in those not receiving antibiotic treatment have not been fully addressed (Sheikh & Hurwitz 2005). The effect of a general non-prescription attitude on transmission rates of pathogens also remains to be clarified (Rose et al. 2005). So far, it has therefore been the prevailing opinion in both the medical profession and the public that acute bacterial conjunctivitis should preferably be treated with topical antibiotics (Hørven 1994; Ehlers & Bek 2004). In addition to the moderately improved clinical response, topical antibiotic treatment may be socioeconomically favourable, because the risk of spreading the infection is probably reduced and the course of the disease is often shortened. Therefore, absence from work due to illness is reduced, particularly because children with acute conjunctivitis may return to kindergarten after only 1–2 days of treatment (in many countries, children are denied admittance to kindergarten or junior school before they have been treated with antibiotics or the signs of conjunctivitis have disappeared) (Lie 1994; Rose et al. 2005). On the other hand, in countries where the national health system carries the expenses of both consultations and medications, the negative socioeconomic consequences of an active consultation and treatment strategy are easily seen. Other negative sides of an active prescription policy include the patients' medical expenses, increased medical profession workload, medicalization of a minor illness, the risk of adverse reactions to the antibiotics used and the risk of widespread microbiological resistance. The difficulty in making a correct clinical distinction between a bacterial and a viral conjunctivitis without an available slit lamp and time-consuming laboratory tests is also a point in favour of a more expectant treatment attitude. A qualitative study of patients' perceptions of acute infective conjunctivitis (Everitt et al. 2003) showed that patients generally regard acute conjunctivitis as a minor disease, but they still commonly believe that it will not heal spontaneously, and accordingly they seek medical help. However, when the self-limiting nature of the condition is explained, they often express a preference to wait a few days before seeking medical attention or being treated with topical antibiotics.
As mentioned earlier, concurrent otitis media occurs relatively frequently in paediatric patients with acute bacterial conjunctivitis caused by H. influenzae (Block et al. 2000; Buznach et al. 2005). In a study comparing the efficacy of short-term oral cefixime therapy with topical use of polymyxin-bacitracin, the frequency of acute otitis media developing during the first 15 days following topical or systemic antibiotic treatment was almost identical (Wald et al. 2001).
The pros and cons of prescribing topical antibiotics to patients with suspected acute bacterial conjunctivitis may be summarized as in Table 1. A highly personal view on the treatment of this condition is presented in Table 2.
Table 1. The pros and cons of topical antibiotics in suspected acute bacterial conjunctivitis.
|Immediate antibacterial treatment|
|More rapidly reduced bacterial growth|
| Reduced transmission rates?|
| Reduced risk of keratitis and other complications?|
|Increased early clinical remission|
| Reduced time out of work or education|
| Earlier restart of contact lens wear|
| Reduced symptoms and worries|
|Early return to kindergarten/junior school|
| Less parental time out of work|
| Socioeconomically favourable|
|Increased ‘burden’ on the healthcare system|
|Socioeconomically unfavourable (if society pays medication and GP fees)|
|Often unnecessary use of topical antibiotics|
| Patient adverse effects|
| Negative influence on the normal flora of both patient and milieu|
| Increased risk of bacterial resistance|
|Delayed antibacterial treatment|
|Reduced use of topical antibiotics (about 50%?)|
|Reduced medicalization of an ‘innocent’ condition|
|Improved patient ‘education’ and responsibility|
|Little or no reduction of health service attendance|
|Increased time out of work/education/ kindergarten|
|No antibacterial treatment|
|Very high percentage of spontaneous clinical cure ≤ 1 week|
|Less antibiotic ‘load’ on patient and society|
|No adverse events related to topical antibiotics|
|Increased time out of work/education/ kindergartens|
|Increased risk of transmitting the infection (?)|
|At least a theoretically increased risk of complications|
Table 2. Management of suspected acute bacterial conjunctivitis: a personal view.
|Always prescribe topical antibiotics|
|Purulent/mucopurulent secretion and distinct discomfort and ocular redness|
|Patients and staff in nursing homes, neonatal units, intensive care units, etc.|
|Children going to kindergartens|
|Contact lens wearers|
|Patients with dry eyes or corneal epithelial disease|
|Usually prescribe topical antibiotics|
|Purulent/mucopurulent secretion and severe ocular redness|
|Patients with previously known external ocular disease|
|Delayed prescription or no antibiotic treatment|
|Patients who do not want immediate antibiotic treatment|
|Patients with moderate (muco)purulent discharge and little or no discomfort|
|Cooperative and well-informed patients|
Choice of topical antibiotics
In a double-blind study of patients with acute bacterial conjunctivitis, Papa et al. (2002) found netilmicin eye drops to be significantly more effective than gentamycin in both eradicating infection and ameliorating signs and symptoms. On the other hand, numerous clinical studies show no significant differences between the clinical effect of various antibiotic eye drops in patients with suspected acute bacterial conjunctivitis (Leibowitz 1991; Miller et al. 1992a; Miller et al. 1992b; Hørven 1993; Carr 1998; Wall et al. 1998; Jackson et al. 2002; Normann et al. 2002). The choice of treatment will therefore be influenced by the ease of medication, side-effects, general microbiological considerations and price.
With the exception of fusidic acid, most commercially available antibiotic eye drops should initially be applied at least six times daily. Fucithalmic® is an aqueous suspension of microcrystalline fusidic acid formulated as viscous eye drops. A protracted release is thereby obtained, giving an adequate concentration of fusidic acid for more than 12 hrs in tear fluid (Thorn & Johansen 1997). Therefore, Fucithalmic® needs only to be applied twice daily, and in addition gives far less initial blurring of vision than the use of eye ointments. A novel enhanced viscosity ophthalmic formulation of tobramycin given twice daily was also shown by Kernt et al. (2005) to be as effective as ordinary tobramycin eye drops used four times daily for acute bacterial conjunctivitis. Recently, an ordinary ophthalmic formulation of gatifloxacin administered twice daily proved to be as effective as the identical medication administered four times a day (Yee et al. 2005).
Acute bacterial conjunctivitis in young children is most frequently caused by H. influenzae, which in vitro susceptibility tests commonly classify as resistant to fusidic acid. On the other hand, both clinical experience and treatment studies show that even such cases respond well to Fucithalmic® eye drops (Hørven 1993; Jackson et al. 2002). This is probably due to the fact that conventional fucidic acid susceptibility tests generally use 1 µg/ml as the breakpoint for susceptibility (Hørven 1993), while concentrations in conjunctival fluid of about 15 µg/ml after 1 hr, 10 µg/ml after 6 hrs and 6 µg/ml after 12 hrs are obtained after one drop of Fucithalmic® (Thorn & Johansen 1997). In vitro studies have shown H. influenzae to be susceptible to fusidic acid in concentrations of approximately 8 µg/ml (Hørven 1993). A general observation in medical treatment is that the fewer the prescribed daily doses, the better the patient compliance. In glaucoma treatment it has been shown that compliance is significantly improved if the treatment schedule is reduced from three or four times a day to twice a day (Norell 1981; MacKean & Elkington 1983; Kass et al. 1987) and that the midday dose is particularly likely to be omitted (Rotchford & Murphy 1998). Correspondingly, several studies have shown a slight, but statistically significant, improvement of treatment compliance by using Fucithalmic® rather than chloramphenicol eye drops in acute bacterial conjunctivitis (Carr 1998; Jackson et al. 2002; Normann et al. 2002). Fucithalmic® therefore appears to be an attractive therapeutic alternative in most cases of acute bacterial conjunctivitis, particularly in children and in the elderly, who need assistance to administer the eye drops. In addition to the antibiotic action, the lubricating effect of the viscous formulation probably contributes to the subjectively improved condition of the patients.
Systemic use of fusidic acid is an important therapeutic option in patients with severe S. aureus infections, particularly osteomyelitis. However, it has been known for several decades that because of the high spontaneous mutation rate if fusidic acid is given as systemic monotherapy, the drug should always be given in combination with other antistaphylococcal drugs (Brown & Thomas 2002; Dobie & Gray 2004; Howden & Grayson 2006). During the last decade, a markedly increased incidence of fusidic-acid-resistant S. aureus has been reported in many countries, while only low rates of fusidic acid resistance have been found in the USA, where fusidic acid is not generally available as a therapeutic option (Ravenscroft et al. 2000; Rørtveit & Rørtveit 2003; Shah & Mohanraj 2003; Dobie & Gray 2004; El-Zimaity et al. 2004; Zinn et al. 2004). It is commonly believed that long-term topical fusidic acid monotherapy is the main cause of the increasing resistance seen in many countries, particularly the widespread use of fusidic acid in chronic skin infections such as impetigo, folliculitis and crural ulcers (Ravenscroft et al. 2000; Brown & Thomas 2002; Mason & Howard 2004; Howden & Grayson 2006). The use of fusidic acid in patients with acute conjunctivitis has so far not been implicated as an important source of increased staphylococcal resistance, most probably because Fucithalmic® delivers a high topical drug concentration and is usually only prescribed for short-term use.
Although the simple treatment schedule (morning and evening) is an important argument in favour of Fucithalmic® in the treatment of acute bacterial conjunctivitis, it should be emphasized that chloramphenicol eye drops are also an excellent choice in the treatment of this disease. The effect of chloramphenicol is as good as that obtained with both Fucithalmic® and more recent, broad-spectrum antibiotics such as ciprofloxacin, norfloxacin and tobramycin (Miller et al. 1992b; Hørven 1993; Carr 1998; Orden Martinez et al. 2004). The greatest disadvantage of chloramphenicol eye drops is the need for frequent application, which may increase the risk of suboptimal patient compliance. However, Lærum et al. (1994) reported a satisfactory clinical effect even with a simplified regimen (four times daily), although the success rate was slightly lower than that obtained by the ordinary chloramphenicol treatment regime.
Chloramphenicol eye drops are the medication of choice in acute bacterial conjunctivitis in many countries, having more than a 90% market share in Australia and almost a 70% market share in the UK (Sheikh & Hurwitz 2001). Chloramphenicol also represents more than 50% of the topical ocular antibiotics prescribed in Ireland, and is the topical medication used in about 55% of patients treated for an acute red eye in England (Doona & Walsh 1995). On the other hand, as a result of the fear of serious systemic side-effects, particularly aplastic anaemia and ‘grey baby syndrome’, topical treatment with chloramphenicol is rarely used in the USA and some other countries (Rayner & Buckley 1996; Doona & Walsh 1998). The relationship between oral use of chloramphenicol and severe, often fatal, aplastic anaemia has been known for more than 50 years (Ritch et al. 1950). Later, the incidence of aplastic anaemia following oral chloramphenicol therapy has been estimated to be about one in 36 000 patients (Wallerstein et al. 1969), which is approximately 13 times greater than the risk of idiopathic aplastic anaemia in the whole population (Fraunfelder et al. 1982; Doona & Walsh 1995). A reversible bone marrow disease mainly suppressing the red blood cell line is caused by a dose-related response, and has never been suspected to be caused by topical chloramphenicol use. On the other hand, a progressive and often fatal bone marrow aplasia affecting all three haematopoetic lines is an idiosyncratic reaction, which may conceivably be precipitated even by topical use (Stern & Killingsworth 1989). It has been proposed that aplastic anaemia triggered by topical chloramphenicol medication may be the result of an individual metabolic predisposition and that the idiosyncratic reaction therefore only occurs in genetically predisposed subjects (Yunis 1973; Fraunfelder et al. 1993). Several reports have suggested a relationship between aplastic anaemia and the use of eye drops or ointment containing chloramphenicol (Rosenthal & Blackman 1965; Carpenter 1975; Abrams et al. 1980; Fraunfelder et al. 1982). However, the aetiological role of chloramphenicol in these published cases has been questioned, particularly because many of the patients had also used several systemic medications, or had extraocular diseases or a family history of blood dyscrasias (Besamusca & Bastiaensen 1986; Buckley et al. 1995; Rayner & Buckley 1996; Lancaster et al. 1998; Wiholm et al. 1998). Two population studies published by Wiholm et al. (1998) did not support the claim that chloramphenicol eye drops increase the risk of aplastic anaemia. Furthermore, in a survey including about 400 general practices in the UK between 1988 and 1995, a total of 442 543 patients received 674 148 prescriptions for chloramphenicol eye drops (Lancaster et al. 1998). Among these, three patients with serious haematological suppression were detected. The authors conclude that even in the unlikely event that all these three cases were caused by chloramphenicol eye drops, the risk of serious haematological toxicity is small. A Dutch regional retrospective case–control study identified 12 patients with aplastic anaemia and 190 patients with other cytopenic dyscrasias during a 4 year study period (Besamusca & Bastiaensen 1986). None of these cases were unequivocally related to the topical use of chloramphenicol, which on a yearly basis was used by one out of 29 patients included in the survey; the frequency of topical chloramphenicol use was approximately the same in patients with and without blood dyscrasias. Therefore, although the occurrence of non-dose-related cases of idiosyncratic aplastic anaemia can not be totally excluded, it must evidently be extremely rare (Walker et al. 1998). In conclusion, topical ophthalmic use of chloramphenicol appears to be a safe, effective and cheap treatment in patients with external ocular infections, but the medication should not be given to individuals with a personal or family history indicating haematological disease (Buckley et al. 1995).
In the USA and many other countries, eye drops containing ciprofloxacin, norfloxacin, gentamycin, tobramycin and other broad-spectrum antibacterial agents are frequently used in acute bacterial conjunctivitis. Because these medications are often used in topical or systemic treatment of more serious infections, the unnecessary development of bacterial resistance is highly undesirable. Increasing occurrence of ciprofloxacin-resistant P. aeruginosa has recently been reported by several authors. Chaudhry et al. (1999) found that while only one of 227 (0.44%) ocular isolates was resistant to ciprofloxacin from 1991 to 1994, eight of 196 (4.1%) ocular isolates showed in vitro resistance from 1995 to 1998. Resistance to ciprofloxacin implied resistance also to most other (at that time) readily available fluoroquinolones. In the same study, all isolates were sensitive to gentamycin.
A 5 year review of the in vitro efficacy of fluoroquinolones on 1053 ocular isolates obtained in 1993–1997 from 825 corneal ulcers (Goldstein et al. 1999) showed a significantly increased resistance of S. aureus to both ciprofloxacin and ofloxacin (from about 5% in 1993 to 35% in 1997). In contrast to the findings of Chaudhry et al. (1999), none of the P. aeruginosa isolates were classified as resistant to fluoroquinolones, but resistance among other pseudomonas species increased from zero to 28.6% during the same period. Considerable gaps in the coverage against coagulase-negative staphylococci, streptococcus species, enterococci species and anaerobs were also found, but these gaps did not appear to change over the years included in the study. Later, levofloxacin eye drops have been reported to give a higher rate of microbial eradication than ofloxacin eye drops in patients with bacterial conjunctivitis (Schwab et al. 2003), mainly because of improved eradication of S. pneumoniae and H. influenzae. However, clinical cure rates did not differ in the two patient groups, illustrating that clinical cure and microbiological eradication are not synonyms. Contrasting the report by Goldstein et al. (1999), both drugs were highly effective against S. aureus, eradicating all strains present at baseline. Recent in vitro studies of bacterial isolates from patients with acute conjunctivitis showed that the fourth-generation fluoroquinolones (gatifloxacin and moxifloxacin) had somewhat better in vitro efficacy than earlier fluoroquinolones against G+ isolates, but not against H. influenzae (Kowalski et al. 2005).
Several other studies have also shown alarmingly high and increasing frequencies of antibiotic resistance among bacteria isolated from patients with presumed bacterial conjunctivitis. Block et al. (2000) reported that the occurrence of beta-lactamase-producing H. influenzae isolated from children with acute conjunctivitis increased from 44% in the late 1980s to 69% at the end of their study, and that the frequency of penicillin-non-susceptible S. pneumoniae (PNSP) then was three times higher than that found 5 years earlier (Doern et al. 1996). Their study also showed diminished antibacterial activity of gentamicin, tobramycin and polymyxin B. In conjunctival cultures from 428 children aged 2–36 months with suspected acute bacterial conjunctivitis, Buznach et al. (2005) found beta-lactamase production in 29% of H. influenzae isolates, while penicillin non-susceptibility was observed in 60% of S. pneumoniae isolates. As much as 96% of the H. influenzae isolates and 97% of the S. pneumoniae isolates showed in vitro susceptibility to chloramphenicol.
Every antibacterial agent will to some extent influence the ecological balance both in the patient's normal bacterial flora and in the surrounding milieu. Broad-spectrum medications with a large ‘ecological shadow’, such as fluoroquinolons and tetracyclines, have a greater tendency than agents will a smaller ecoshadow to cause bacterial resistance as well as altering the normal bacterial flora (Midtvedt 2004). Among others, O'Brien & Hahn (2005) have suggested that fluoroquinolones should not be used routinely for bacterial conjunctivitis, but should be reserved for particularly severe cases. In my opinion, neither fluoroquinolones nor aminoglycosides should be used in the treatment of uncomplicated acute bacterial conjunctivitis, and topical administration of these medications should be restricted to more serious eye infections and to cases of conjunctivitis in patients not tolerating or improving on fusidic acid and chloramphenicol treatment.