Settings and patients
Between October 1999 and December 2002, 41 GPs working in 25 care centres in the Amsterdam and Alkmaar region recruited patients for the trial. All eligible patients were referred for inclusion to nine designated ‘study’ GPs who worked in nine of the 25 centres. These GPs were instructed to include patients with a red eye and either (muco)purulent discharge or sticking of the eyelids. The exclusion criteria were age younger than 18 years, pre-existing symptoms longer than 7 days, acute loss of vision, wearing of contact lenses, systemic or local antibiotic use within the previous 2 weeks, ciliary redness, eye trauma, and a history of eye operation. All eligible patients were referred to one of the nine study GPs for enrolment in the study. Patients were recruited during office hours only. Before inclusion, a written informed consent was obtained. The patients were informed about the goal and duration of the study, the additional investigations (cultures), and about the choice of fusidic acid and the comparison to placebo. The patients were informed that, in case of infectious conjunctivitis, most patients receive a treatment with fusidic acid, and that there is no evidence that this treatment is beneficial. The patients were informed that possible harms due to withholding treatment were not expected. Subsequently, a standardised questionnaire and a standardised physical examination were completed by the study GP. The questionnaire contained questions about previous medical history (self-reported), duration of symptoms (days), self-medication and self-therapy, itching, burning sensation, foreign body sensation, and about the number of glued eyes in the morning (zero, one or two). The physical examination contained investigation of the degree of redness (peripheral, whole conjunctiva, or whole conjunctiva and pericorneal), the presence of peri-orbital oedema, the kind of discharge (watery, mucous, or purulent), and bilateral involvement (yes/no). Next, in a standardised way, one conjunctival sample of each eye was taken for a bacterial culture. For each patient, one eye was designated as the ‘study eye’. In case of two affected eyes, the one with worse signs and/or symptoms was the ‘study eye’. In case of two equally affected eyes, the first affected eye was the ‘study eye’. Next, the ‘study eye’ (that is, the patient) was allocated to either treatment with fusidic acid or placebo. In case of two affected eyes, the bilateral eye received the same treatment as the ‘study eye’.
How this fits in
Although ocular antibiotics, in particular fusidic acid, are usually prescribed, there is no publicised evidence on their effect in primary care patients with an acute infectious conjunctivitis. This study demonstrates that the effect of fusidic acid gel seems similar to placebo in primary care patients with an acute infectious conjunctivitis. GPs should be restrictive in blindly prescribing fusidic acid.
During a week, the patients kept a daily diary, containing six questions about the presence of symptoms in the study eye. These signs and symptoms were: discharge (yes/no); redness (yes/no); itching (yes/no); foreign body sensation (yes/no); glued eyelids in the morning (yes/no); and photophobia (yes/no).
The patients were instructed to apply one drop of the study medication four times daily to the affected eye(s), starting on the day of inclusion. The GP demonstrated the first application. The patients received a written instruction for use. The study medication was either fusidic acid gel 10 mg/g (Fucithalmic®; Leo Pharmaceutical Products) or placebo gel (Vidisic® 2 mg/g: Tramedico). The patients were advised to use the study medication until 1 day after the signs and symptoms were recovered.
Seven days after inclusion the patients were asked to visit their GP for evaluation. Data about recovery, adverse effects, and ocular signs and symptoms were recorded using a standardised questionnaire. The questionnaire contained questions about itching, burning sensation, foreign body sensation, and about the number of glued eyes in the morning (zero, one or two). The physical examination contained investigation of the degree of redness (peripheral, whole conjunctiva, or whole conjunctiva and pericorneal), the presence of peri-orbital oedema, the kind of discharge (watery, mucous, or purulent), adverse effect, and the overall judgement of the GP if the study eye had recovered completely (yes/no). Again, two conjunctival samples were taken for a bacterial culture, one from each eye.
Any remaining study medication was weighed to get an impression of treatment adherence.
Objectives and outcomes
The objective of this study was to assess the effectiveness of fusidic acid gel compared to placebo for recovery of acute infectious conjunctivitis in primary care patients.
The primary outcome measure was the difference in the proportions of patients recovered after 7 days of treatment. Recovery was defined as absence of any signs and symptoms, objectified by the GP, indicating conjunctivitis. Secondary outcome measures were the difference in bacterial eradication rates after 7 days, adverse effects, and a survival time analysis of the duration of symptoms. Finally, the extent to which the 7-day recovery rate in culture-positive patients differed from that in culture-negatives was studied.
Randomisation and blinding
The study medication was repacked into identical tubes by a local pharmacist under aseptic circumstances, according to the consecutive numbers on nine computer-generated randomisation lists, one for each GP. The tubes were then labelled with a code number. The GPs received the coded tubes, and handed out the tube with the first available code number to their consecutive patients. In this way, the patients were assigned in double blind fashion to either fusidic acid gel or placebo. The pharmacist was the only person in possession of the randomisation lists and was not involved in outcome measurement or data analysis. The code was broken only after the follow up had been completed and all data was entered into a database.
Conjunctiva samples were taken by rolling a cotton swab (Laboratory Service Provider) over the conjunctiva of the lower fornix. The swabs were put in a transport-medium and sent to the investigating laboratory. On arrival, the swabs were inoculated onto blood agar enriched with 5% sheep blood, MacConkey agar, and chocolate agar. All media were house made using standard ingredients. After standard inoculation, the blood agar and MacConkey agar plates were incubated for a period of 48 hours at 35°C, whereas the chocolate agar plates were inoculated during the same period and temperature, but in 7% CO2
atmosphere. Cultures were further analysed daily according to the guidelines of the American Society for Microbiology.20
All pathogens were identified using routine standard biochemical procedures. The susceptibility of fusidic acid was assessed by determination of the minimal inhibitory concentration of this antibiotic using an Etest® (AB Biodisk) assay on Mueller–Hinton agar. Minimal inhibitory concentrations were read after incubation at 35°C for 24 hours according to the manufacturers’ instructions. Critical minimal inhibitory concentrations were used according to the Dutch National Standards.21
The difference in recovery rates after 7 days between the study groups was analysed using a χ2
test (two-sided, unpaired). Logistic regression was used to adjust the (crude) treatment effect estimate for potential confounding effects due to post-randomisation imbalances (), and to study the extent to which the recovery rates between culture-positive and culture-negative patients differed. For easier interpretation, the odds ratios obtained from the regression analysis were converted into relative risks.22
These relative risks were converted into risk differences and numbers needed to treat using the recovery rate in the placebo group as the background risk. The 95% CIs of the numbers needed to treat (benefit) and numbers needed to treat (harm) are presented as recommended by Altman.23
An additional analysis, in which we assumed that all patients (18 in total) who were lost to follow up had not recovered at 7 days, was performed.
Characteristics at baseline (n = 181).
The median duration of the combined symptoms (diary) in both groups was analysed using a Kaplan–Meier survival analysis. Of each patient, all recorded symptoms in the diary were combined in a new variable that was coded ‘1’ for patients who had become symptom free (all six symptoms in the diary absent) at day 1–7, and coded ‘0’ otherwise. In this way we were able to analyse the median duration of the combined symptoms (diary) in both groups.
The difference in bacterial eradication rates after 7 days was analysed using a χ2 test (two-sided). The distributions of the milligrams of study medication were compared between the study arms using the Mann-Whitney U test. A P-value of ≤0.05 indicated a statistically significant difference between the distributions of study medication use.
With a postulated recovery rate after 7 days of 95% in the intervention group and 80% in the placebo group, a difference in recovery of 15% was considered clinically relevant. With the type I and type II error rates at 0.05, and 0.20, respectively, the required sample size was 88 patients per group.
The statistical analyses were performed using SPSS (version 11.5.2).
Forty-one GPs referred 184 patients to the GPs, of which 181 were randomised (). With regard to baseline characteristics, the groups appeared comparable with possible exception of age, sex, history of infectious conjunctivitis, a foreign body sensation in the eye, and bilateral involvement (). In the fusidic acid and the placebo group 8 and 10 patients, respectively, were lost to follow up (). Thus, 163 patients were analysed.
The median consumption of study medication was 1.51 g (interquartile range ([IQR] = 0.75–2.24) in the fusidic acid group, and 1.21 g (IQR = 0.87–1.69) in the placebo group (P = 0.303).
After 7 days, the proportion of recovered patients was 45/73 (62%) in the fusidic acid gel group and 53/90 (59%) in the placebo group (). Consequently, the probability of recovery was 2.8% greater in the fusidic acid group with a risk difference of 2.8% (95% CI = −13.5 to 18.6), the number needed to treat (NNT) (benefit) was 36.3 (95% CI = NNT [harm] 7.4 to ∞ to NNT [benefit] 5.4). Age was the only confounding factor and after adjustment the risk difference was 5.3% (95% CI = −11.0 to 18.0). The treatment effect seemed stronger in culture-positive patients (adjusted risk difference = 22.9% [95% CI = −6.0 to 42.0]) (). The additional analysis showed only a small effect on our results where the risk difference decreased from 5.3% (95% CI = −11.0 to 18.0) to 3.8% (95% CI = −11.0 to 18.0) ().
Numbers recovered at 1 week (n = 163).
Logistic regression analysis of effect of treatment by culture result.a
Additional analysis (patients who were lost to follow up (18 in total) were assumed not recovered at 7 days)
The Kaplan–Meier survival curves of symptoms (diary) did not differ significantly between the two groups (; P = 0.422, logrank test). No patients became symptom-free within 2 days.
Kaplan–Meier curves showing times to disappearance of symptoms for the two groups.
Within the group of patients whose study eye had recovered at 1 week, 3.1% (3/98) of the non-study eyes showed signs and symptoms of conjunctivitis; 2.2% (1/45) in the fusidic acid group, and 3.8% (2/53) in the placebo group, respectively.
In both trial arms there were no clinically serious adverse outcomes.
At baseline, 58/181 (32%) patients were culture positive. The most prevalent cultured species was Streptococcus pneumoniae, accounting for 27/58 (47%) of the positive cultures. Overall, 38/58 (66%) cultures proved to be resistant to fusidic acid (). After 7 days, the bacterial eradication rates were 16/21 (76%) in the treatment group and 12/29 (41%) in the placebo group with a risk difference of 34.8% (95% CI = 9.3 to 60.4) and NNT (benefit) of 2.9 (95% CI = 1.7 to 10.8) ().
Cultured species at baseline and their susceptibility to fusidic acid (n = 181).
Bacterial eradication after 1 week in culture-positive group (n = 50).a
The proportion of patients that recorded adverse effects was 10/73 (14%) in the treatment group and 3/90 (3%) in the placebo group with a risk difference of 10.4% (95% CI = 1.6 to 19.1) and a NNT to treat of 9.7 (95%CI = 5.2 to 60.6). The most common adverse effect was a burning sensation after instillation of the study medication, with a prevalence of eight out of 10 in the treatment group and one in three in placebo group.