Professional Documents
Culture Documents
Summary objective To compare a topical quinolone antibiotic (ciprofloxacin) with a cheaper topical antiseptic
(boric acid) for treating chronic suppurative otitis media in children.
design Randomized controlled trial.
setting and participants A total of 427 children with chronic suppurative otitis media enrolled from
141 schools following screening of 39 841 schoolchildren in Kenya.
intervention Topical ciprofloxacin (n ¼ 216) or boric acid in alcohol (n ¼ 211); child-to-child
treatment twice daily for 2 weeks.
main outcome measures Resolution of discharge (at 2 weeks for primary outcome), healing of the
tympanic membrane, and change in hearing threshold from baseline, all at 2 and 4 weeks.
results At 2 weeks, discharge was resolved in 123 of 207 (59%) children given ciprofloxacin, and in
65 of 204 (32%) given boric acid (relative risk 1.86; 95% CI 1.48–2.35; P < 0.0001). This effect was
also significant at 4 weeks, and ciprofloxacin was associated with better hearing at both visits. No
difference with respect to tympanic membrane healing was detected. There were significantly fewer
adverse events of ear pain, irritation, and bleeding on mopping with ciprofloxacin than boric acid.
conclusions Ciprofloxacin performed better than boric acid and alcohol for treating chronic
suppurative otitis media in children in Kenya.
C. Macfadyen et al. Topical quinolone vs. antiseptic for chronic suppurative otitis media
Cochrane Review found quinolones superior to other experienced in ENT] screened children by class, inviting
topical antibiotics for CSOM; we conducted a randomized children with CSOM and their guardians to an induction
trial to evaluate this expensive but apparently effective visit, usually two school days later. At induction, a child’s
drug against topical antiseptics in Kenyan schoolchildren. legal guardian(s) provided signed informed consent before
assessments. After consent, but before randomization,
teams performed baseline and eligibility assessments.
Methods Demographic data and a medical history were taken.
Nurses established pure tone hearing threshold for air
Study site
conduction, in decibels at 500 Hz, 1 kHz, 2 kHz, 4 kHz
Rural primary schools in Kisumu District, West Kenya: we and 8 kHz, in a quiet location and by a standard technique
visited 165 of the 186 total. The District has the highest using portable Kamplex screening audiometers (AKM KS8
infant (116.7/1000) and under 5-year (194.7/1000) MP, battery operated); they recorded ambient noise levels
mortality in Kenya, and ear infection is common (Ministry using sound level meters. After audiometry, registrars and
of Planning and National Development 2004). clinical officers swabbed infected ears (for bacteriology and
sensitivity analysis in Kisumu), then examined both ears for
presence and degree of discharge, tympanic membrane
Participants
perforation, and any other otoscopic findings using Earlite
School children aged 5 years and older, with (a) purulent, Kite and Heine mini otoscopes.
aural discharge for 14 days or longer; (b) pus in the After completing all induction assessments, eligible
external canal on otoscopy; and (c) perforation of the children were allocated their sequential treatment pack;
tympanic membrane. We excluded children who had been clinical officers trained children and teachers and super-
treated for ear infection or received antibiotics for any vised the first dose. Every child was given a treatment
other disorder in the previous 2 weeks, or who had other record to complete to monitor treatment administration.
ear problems (pre-existing disease, complicated otitis
media, anatomical abnormalities) or allergy to study drugs.
Follow-up and outcomes
Children were seen at 2 and 4 weeks to collect data. Where
Interventions
children were absent, we revisited the school the next day.
Topical eardrops were given twice daily, after dry mop- All teams rotated schools for follow-up visits, to avoid
ping, for 10 consecutive school days (no treatment at them assessing children they had previously seen. Out-
weekends) with either ciprofloxacin eardrops (Ciloxan comes were (a) resolution of aural discharge; (b) healing of
0.3%; Alcon), or antiseptic eardrops (2% boric acid in tympanic membrane; and (c) change in hearing threshold.
45% alcohol). Older children were appointed as ‘ear The primary outcome was resolution of aural discharge at
monitors’ and trained to clean and treat the infected ears, 2 weeks. We also recorded adverse events.
under the supervision of trained teachers, as described in Children with persistent discharge at week 2 were
Smith et al. (1996). instructed to dry mop the ear(s) until week 4; those
discharging at week 4 were also instructed to dry mop, and
given a new bottle of eardrops and referred, with children
Randomization and masking
with persistent perforation or other ear pathology, to the
Children were randomized in a 1:1 ratio using computer- ENT surgeon in Kisumu. Children with other illnesses were
generated block randomization, stratified by school. Each referred to the nearby clinic or hospital. Any children
treatment pack contained two bottles of randomized referred during the study, remained in the study, and details
treatment and remained sealed until allocated to a child; of their referral and subsequent treatment were collected.
packs and the bottles were identical in appearance and
both treatments identical in colour and smell. Participants,
Sample size
carers, and outcome assessors remained blind to the
treatment allocated throughout the study. Rates of CSOM may be higher in Kisumu than the 1.1%
found other parts of Kenya, with its higher infant mortality
rates (http://www.cbs.go.ke) and higher rates of ear diseases
Field procedures
estimated by local medical staff and local surveys conducted
Four trained teams [each consisting of an Ear Nose and in 2000 (Educational Assessment and Resource Centre
Throat (ENT) registrar, clinical officer, and a nurse Kisumu, and New Nyanza Provincial General Hospital
C. Macfadyen et al. Topical quinolone vs. antiseptic for chronic suppurative otitis media
(NNPGH) ENT Clinic, Kisumu, unpublished data; and ficance using the Pearson’s chi-square statistic. We used
Dr David Odeny, personal communication) – we therefore logistic regression for resolution at 2 weeks, to assess
used a higher estimate than that of Hatcher et al. 1995, but whether the age of the child, bilateral disease, length of
more conservative than the local Kisumu estimates, 15.1%, current CSOM episode and degree of perforation at
which were judged to be overestimated as they were based induction affected the relative treatment effect. Results are
on 900 students screened in three schools selected following expressed as odds ratio (OR).
high referral numbers to the education assessment and For audiometry, analysis of covariance (ancova) was
rehabilitation centre in Kisumu and rates may not be undertaken for hearing threshold, according to the study
limited to CSOM. Assuming a prevalence of 2%, and a total protocol. The WHO classifications were also used to
of 46,116 children in 186 schools in February 2002 (Office describe hearing impairment levels and Fisher’s exact test
of the District Education Officer 2002), we estimated that was used consider changes between levels.
920 children would be potentially eligible for the study. We report adverse events, notably local symptoms such
Assuming 750 agreed to participate and an estimated 50% as pain, irritation or bleeding (for example on mopping), as
resolution rate at 2 weeks in the antiseptic group (Acuin a secondary outcome.
et al. 2004), we calculated the study would have a power of
79% to detect the defined minimum clinically worthwhile
Ethical approval
absolute difference in resolution rates of 10%, at a two-
sided significance level of 5%. We obtained ethical approval from the Ministries of
An interim analysis was carried out for the first 200 Education and Health, the Provincial Ethical Review
study participants to check the estimated resolution rate for Committee, the Kenyatta National Hospital Institutional
the control group. The results indicated a resolution rate at Review Board in Kenya, and the Liverpool School of
2 weeks of 0.31 (95% CI 0.22–0.40) in the boric acid Tropical Medicine Research Ethics Committee in the UK.
group. Although the original estimate of 0.5 was excluded We received consent for the study from local education and
from the confidence interval, further sample size calcula- medical staff, from head teachers of the schools, who
tions assuming, in turn, the limits of this interval to be the informed students, and parents. We obtained informed
true proportion (i.e. 0.22 and 0.40 respectively), showed parental consent from eligible children’s legally acceptable
our original numbers would still be sufficient to provide representatives.
approximately 80% power to detect an absolute difference
of 10% in either of these scenarios. Thus our target sample
Results
size seemed compatible with the plausible estimates for the
control group resolution rate from our interim analysis. As We conducted the study in 165 of the 186 Kisumu rural
this calculation did not involve any comparison between primary schools in the May to August 2002 school term.
the two groups, the overall type 1 error rate is unaffected Twenty-one schools were not reached in the time
(Wittes et al. 1999). available. We reviewed 39 266 children at the screening
visit, plus an additional 575 children seen at the
induction visit. We found eligible children in 141
Statistical methods (analyses)
schools, and randomized a total of 427 children (Fig-
All analyses followed the intention-to-treat principle and ure 1). At the week 4 visit, three subjects in the boric
were conducted using SAS v8.2 2001. For bilateral disease, acid group and one in the ciprofloxacin group were seen
resolution and healing were considered to occur when both 1 week late. Seven children received some or all of the
ears were resolved or healed. We also carried out sensitivity wrong treatment because of switching with other par-
analyses defining bilateral resolution and healing to have ticipants’ bottles at allocation or during treatment
occurred when either or both ears resolved or healed. (Figure 1). All participants were analysed in the group
For audiometry readings, a single reading was taken they were initially randomized to. The groups were
from the diseased ear in unilateral cases, whilst the average comparable with respect to baseline data (Table 1).
across the two ears was calculated for bilateral disease.
Hearing threshold was recorded in dBHL and was
Outcomes
averaged across the 500 Hz, 1 kHz, 2 kHz and 4 kHz
frequency range. Resolution of aural discharge is higher with ciprofloxacin
For resolution and healing outcomes, ciprofloxacin was at 2 and 4 weeks (Table 2). Sensitivity analyses allowing
compared with boric acid using the relative risk (RR) and for varying outcome definition in bilateral disease did not
95% confidence interval. We assessed the statistical signi- significantly affect the results. The estimated treatment
C. Macfadyen et al. Topical quinolone vs. antiseptic for chronic suppurative otitis media
Randomised (n = 427)
effect for resolution of discharge did not differ substantially expressed as OR, 95% CI were 3.13, 2.09–4.69 (unad-
after adjusting for age of the child, bilateral disease, length justed); 3.16, 2.04–4.93 (adjusted); and at 4 weeks were
of current CSOM episode and degree of perforation at 2.86, 1.91–4.27 (unadjusted); and 2.91, 1.88–4.50
induction, using logistic regression: at 2 weeks, the results, (adjusted).
C. Macfadyen et al. Topical quinolone vs. antiseptic for chronic suppurative otitis media
Characteristic N* N*
Resolution at 2 weeks
1. Both ears resolve 123/207 (59.4) 65/204 (31.9) 1.86 (1.48–2.35) <0.0001
2. Either or both ears resolve 132/207 (63.8) 77/202 (38.1) 1.67 (1.36–2.05) <0.0001
Resolution at 4 weeks
1. Both ears resolve 130/196 (66.3) 90/198 (45.5) 1.46 (1.22–1.75) <0.0001
2. Either or both ears resolve 143/197 (72.6) 105/198 (53.0) 1.37 (1.17–1.60) <0.0001
C. Macfadyen et al. Topical quinolone vs. antiseptic for chronic suppurative otitis media
Healing at 2 weeks
1. Both ears healed 15/207 (7.2) 14/204 (6.9) 1.06 (0.52–2.13) 0.879
2. Either or both ears 19/207 (9.2) 18/202 (8.9) 1.03 (0.56–1.90) 0.925
healed
Healing at 4 weeks
1. Both ears healed 31/200 (15.5) 20/199 (10.1) 1.54 (0.91–2.61) 0.103
2. Either or both ears 38/200 (19.0) 28/199 (14.1) 1.35 (0.86–2.11) 0.185
healed
C. Macfadyen et al. Topical quinolone vs. antiseptic for chronic suppurative otitis media
Table 5 Levels of hearing impairment at each visit (dB averaged over 500, 1000, 2000 and 4000 Hz)
* An improvement in the level of hearing impairment means there was a decrease in the
average hearing threshold. WHO grades are as presented in Table 5.
Hearing in the diseased ear is usually significantly impaired et al. 2003; Acuin et al. 2004; Acuin 2004). No formal
compared with the uninfected ear, and children with analyses were performed on any other safety findings in our
unilateral disease, as for bilateral impairment, are likely to trial, which was not designed or powered to assess more
have significant educational and social problems (Ballan- specific safety questions, as numbers were small and
tyne & Martin 1984; Brookhouser et al. 1991). Longer- specific adverse event information may not have been
term follow-up would be needed to assess the impact of systematically detected because of the open-ended nature
treatment on hearing in the long-term and on behaviour of the questioning.
and learning. Although ciprofloxacin remains more expensive than
Smith et al. (1996) found that healing of the tympanic antiseptics and other non-quinolone antibiotics (approxi-
membrane was an important factor for improving hearing. mately 12 times the price of boric acid in Kenya), the
While we found few cases of complete healing with either difference varies between countries. It is likely that the cost
treatment, there was a trend towards a small benefit of of quinolone eardrops will fall with time, particularly if
ciprofloxacin by week 4; the wide confidence intervals imply generic preparations are available. In addition to direct
a potentially clinically important benefit cannot be ruled out purchase costs, other treatment-related costs and practi-
for tympanic membrane healing with ciprofloxacin. calities for treatment preparation, transport and storage
We followed up students for just 4 weeks, because of must also be considered when comparing treatments.
logistic and financial constraints. However, longer follow-
up would be needed to assess whether the differences found
Conclusions
here (for resolution, healing and hearing) remain over time.
The higher number of children reporting adverse events In children with chronic suppurative otitis media, we have
related to ear pain, irritation, and bleeding on mopping found topical ciprofloxacin ear drops to be more effective
with boric acid in alcohol is probably because the alcohol than boric acid at 2 and 4 weeks, for both resolution of
may sting. This finding is in line with other studies to date, discharge and improvement in hearing. We also found
which have supported a good safety profile for quinolone significantly fewer adverse events of ear pain, irritation, and
eardrops (Brownlee et al. 1992; Morpeth et al. 2001; Jaya bleeding on mopping, with ciprofloxacin than boric acid.
C. Macfadyen et al. Topical quinolone vs. antiseptic for chronic suppurative otitis media
Authors
Carolyn Macfadyen (corresponding author), Paul Garner, Ian Mackenzie, Kennedy Otwombe, International Health Research Group,
Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, UK. Fax: +44 (0) 151 705 3364; Tel: +44 (0) 151 708
9393, E-mail: cmacfadyenuk@yahoo.co.uk, pgarner@liverpool.ac.uk, macken34@liverpool.ac.uk, notwombe@yahoo.com
Carrol Gamble, Stephen Taylor, Paula Williamson, Centre for Medical Statistics and Health Evaluation, School of Health Sciences,
Shelley’s Cottage, Brownlow Street, University of Liverpool, Liverpool L69 3GS. Tel.: 44 (0) 151 794 5121; Fax: +44 (0) 151 794 5130;
E-mail: c.gamble@liverpool.ac.uk, s.taylor01@liverpool.ac.uk, p.r.williamson@liverpool.ac.uk
Isaac Macharia, Peter Mugwe, Herbert Oburra, Section of Ear, Nose and Throat Diseases, Department of Surgery, University of
Nairobi, Kenyatta National Hospital, Nairobi, Kenya. E-mail: machariaim@wananchi.com, mugwe@wananchi.com, kansel@
insightkenya.com