A Controlled Trial of Povidone–Iodine to Treat
Infectious Conjunctivitis in Children
SHERWIN J. ISENBERG, MD, LEONARD APT, MD, MARIO VALENTON, MD,
MADELINE DEL SIGNORE, RN, LEO CUBILLAN, MD, MARIA A. LABRADOR, MD,
PAULINE CHAN, RN, AND NANCY G. BERMAN, PHD
● PURPOSE: To report the efficacy of povidone–iodine as
a treatment for conjunctivitis in pediatric patients.
● DESIGN: Double-masked, controlled, prospective clin-
ical trial.
● METHODS: In an ophthalmology clinic in a general
hospital in Manila, Philippines, 459 children (mean [SD]
age 6.6 [6.6] years; range, 7 months–21 years) with
acute conjunctivitis were studied. Infected eyes were
cultured for bacteria and underwent immunofluorescent
testing for Chlamydia trachomatis. Viral conjunctivitis
was diagnosed if bacterial cultures were negative and
diagnostic criteria were met. Subjects were alternated to
receive povidone–iodine 1.25% or neomycin–polymyxin-
B-gramicidin ophthalmic solution, one drop 4 times daily
in the affected eye. Ocular inflammation was evaluated
daily by the family or patient and weekly by an ophthal-
mologist. The main outcome measures were days until
cured and proportion cured after 1 and 2 weeks of
treatment.
● RESULTS: Despite adequate statistical power (power
>80% for a 1-day difference and P < .05), there was no
significant difference between treatment groups regarding
the number of days to cure or proportion cured at 1 or 2
weeks whether caused by bacteria or virus (P ⴝ .133–
.824 for the four comparisons). After 1 week of treat-
ment, povidone–iodine cured marginally more chlamydial
infections than the antibiotic (P ⴝ .057). By 2 weeks,
fewer chlamydial infections were cured than those of
viral or bacterial etiology (P ⴝ .0001). The younger the
Accepted for publication June 28, 2002.
From the Jules Stein Eye Institute (S.J.I., L.A., M.D.S., N.G.B.),
Departments of Ophthalmology and Pediatrics, Harbor/UCLA Medical
Center, UCLA School of Medicine, Los Angeles, California, and
Torrance, California; and the Department of Ophthalmology (M.V.,
L.C., M.A.L., P.C.), Philippine General Hospital, University of the
Philippines, Manila, Philippines.
This study was supported by the Thrasher Research Fund of Salt Lake
City and Research to Prevent Blindness Senior Scientific Investigator
Award (S.J.I.), New York, New York.
Inquiries to Sherwin J. Isenberg, MD, Department of Ophthalmology,
Harbor/UCLA Medical Center, 1000 W. Carson St, Torrance, CA
90509; fax: (310) 837-4078; e-mail: isenberg@ucla.edu
0002-9394/02/$22.00 © 2002 BY ELSEVIER SCIENCE INC. ALL
PII S0002-9394(02)01701-4
patient, the faster their conjunctivitis resolved (R ⴝ
0.13, P ⴝ .013).
● CONCLUSIONS: Povidone–iodine 1.25% ophthalmic
solution was as effective as neomycin–polymyxin B-
gramicidin for treating bacterial conjunctivitis, somewhat
more effective against chlamydia, and as ineffective
against viral conjunctivitis. Povidone–iodine ophthalmic
solution should be strongly considered as treatment for
bacterial and chlamydial conjunctivitis, especially in de-
veloping countries where topical antibiotics are often
unavailable or costly. (Am J Ophthalmol 2002;134:
681– 688. © 2002 by Elsevier Science Inc. All rights
reserved.)
A
CUTE CONJUNCTIVITIS IS ONE OF THE MOST COM-
mon eye conditions seen by medical and pediatric
practitioners. Without treatment, the infection
may improve spontaneously or, infrequently, worsen to
cause keratitis, corneal ulceration, scarring, and possible
blindness. These serious complications occur more fre-
quently in the setting of vitamin A deficiency, malnutri-
tion, and shortage of appropriate topical antibiotics—a
scenario all too common in many developing countries,
resulting in thousands of children becoming blind annual-
ly.1 Of the 1.5 million blind children in the world most are
in developing countries, where corneal opacification is the
leading preventable cause of blindness.2
Low cost, effectiveness, and lack of microbial resistance
make povidone–iodine an appealing drug to treat ocular
infections, especially in developing countries. Even in
developed countries, increasing bacterial resistance to
conjunctivitis treatment, as reported in the United States
may also make povidone–iodine attractive.3 Povidone–
iodine interacts strongly with the double bonds of satu-
rated fatty acids in the bacterial cell wall and cell organelle
membranes and also oxidizes amino acid and nucleotides.
It causes pore formation and solid–liquid interfaces at the
lipid membrane level of cell walls to lose cytosol material.4
Povidone–iodine ophthalmic solution has been previ-
ously shown by our group to be an effective preventative
germicidal agent. It was effective and safe when used
RIGHTS RESERVED. 681
before ophthalmic surgery, whether applied intermittently
for 3 days before surgery or immediately before surgery
when the patient was on the operating table.5–7 It was also
found to be safe and effective when used after ophthalmic
surgery.8,9 In Kenya, we found povidone–iodine safe, inex-
pensive, and better than silver nitrate or erythromycin
when applied to the eyes of newborns to prevent ophthal-
mia neonatorum.10 Povidone–iodine has an extremely
broad spectrum of antimicrobial activity in vitro, including
essentially all bacteria, viruses, chlamydia, and fungi, given
enough contact time. Povidone–iodine has been found to
be effective against chlamydia and viruses when tested in
vitro in concentrations even more dilute than that used in
this study (1.25%).11 In one Japanese investigation, a
number of viruses, including adenovirus and herpes, were
found to be very susceptible to povidone–iodine in vitro,
even in concentrations at or below 1.0%.12 Microbial
resistance and allergic reactions to this drug are rare. In
Manila where this study was conducted, a 10 ml bottle of
the ophthalmic antibiotic solution consisting of neomy-
cin-polymyxin-B-gramicidin is much more expensive than
povidone-iodine solution. In our Nairobi study, a 5-ml
bottle of povidone-iodine ophthalmic solution prepared by
the local pharmacist was also inexpensive.10 A 10 ml bottle
of Neosporin Ophthalmic Solution (neomycin-polymyxin-
B-gramicidin; Burroughs Wellcome, Research Triangle
Park, North Carolina, USA) costs about $41 in a Los
Angeles pharmacy. Because povidone–iodine ophthalmic
solution can be prepared from powder or stock solutions
meant for other antiseptic purposes, it is not only inex-
pensive but also widely available in underdeveloped coun-
tries.
The use of povidone–iodine for treatment, instead of
prophylaxis, has never been studied in a prospective,
controlled, and masked clinical trial. We therefore con-
ducted this trial of topical povidone-iodine to treat infec-
tious conjunctivitis compared with the locally used
antibiotic for superiority or at least equivalency. The
etiologies of the infections were subdivided into three
groups by causative organism: bacteria, chlamydia, and
viruses. Although chlamydiae are a type of bacteria, we
chose to segregate their data because of their different
nature, such as intracellular replication and frequent ve-
nereal transmission, and their antibiotic sensitivity pat-
tern, which is often different from most bacteria. As this
treatment would potentially have its greatest benefit in
developing countries, we decided to conduct this investi-
gation in the Philippine General Hospital of the Univer-
sity of the Philippines in Manila, where external eye
infections are more prevalent than in developed areas.13
METHODS
THE PROTOCOL WAS REVIEWED AND APPROVED BY THE
institutional review boards of the Harbor–UCLA Medical
682 AMERICAN JOURNAL
Center and the Philippine General Hospital. Signed,
informed consent was obtained from each family.
The study was designed as a double-masked, controlled,
prospective clinical trial. It was set in the ophthalmology
clinic in a general hospital in Manila, Philippines.
The enrollment period was from February 1997 through
June 1999. Patients between the ages of 1 month and 21
years with a history of an untreated red inflamed eye with
discharge that began within 14 days of the examination
date were candidates for the study. Children were excluded
if (1) the eye had no discharge, (2) oral or topical ocular
antibiotics had been used within 14 days, (3) pruritis of the
eyes or eyelids associated with giant papillae of the tarsal
conjunctiva was present, (4) a history of allergy to povi-
done–iodine, neomycin, polymyxin, or bacitracin was
elicited, (5) the cornea or sclera had been perforated, or
(6) a hypopyon was present.
At the initial ophthalmic examination, both eyes were
cultured. Conjunctival and eyelid margin specimens were
obtained after topical application of 0.5% proparacaine
hydrochloride. The conjunctival culture was obtained by
everting the lower eyelid and scraping with a Kimura
spatula within the lower cul-de-sac. Cytologic examina-
tion of the conjunctival scrapings of stained smears was
performed.14 The swab was directly plated onto three
culture plates: 10% sheep blood agar, chocolate agar, and
Sabouraud medium. The eyelid margins (upper and lower)
were then wiped with an applicator and streaked on the
same plate. The cotton applicator was cultured in brain–
heart infusion broth. Chocolate agar plates were incubated
with 50% carbon dioxide at 35 C for at least 7 days. After
incubation, colony-forming units were differentiated
and enumerated by standard bacteriologic techniques.
Chlamydiae were identified by direct immuno-
fluorescence (Microtrak System; Syva Corporation, Palo
Alto, California, USA). Viral conjunctivitis was diagnosed
if the bacterial cultures were negative and the clinical
examination showed two or more typical findings, such as
a follicular conjunctival reaction, discharge that was not
primarily purulent, conjunctival membranes, or corneal
infiltrates.15
All subjects were alternately assigned to one of two
treatment arms. Our previous experience in studies in
developing countries showed the method of alternate
assignment to be satisfactory, as it was technically easier to
accomplish and allowed for less confusion. There was little
confounding with this method of assignment, since the
demographic variables and type of infection were similar
for each treatment group. In bilateral cases, both eyes
received the same medication but only the right eye was
used for analysis. Subjects received either povidone–iodine
1.25% solution prepared in the hospital pharmacy using
aseptic techniques or an antibiotic solution consisting of
neomycin–polymyxin B-gramicidin in each infected eye
four times daily, with the last dose at bedtime. This
antibiotic combination ophthalmic solution was chosen
OF OPHTHALMOLOGY NOVEMBER 2002
since it is currently the most frequently used medication to
treat conjunctivitis in the Manila area. The medication
was used until the ophthalmologist, at the weekly exami-
nation, directed discontinuation after determining that the
patient was cured. This might have occurred after 1, 2, or
3 weeks of treatment. No patient required treatment for
more than 3 weeks. The patient or family was given a local
telephone number to call if any problem arose.
Two special forms of conjunctivitis were treated differ-
ently, according to the current standard of care. If the
clinical examination and culture results indicated a gono-
coccal infection, ceftriaxone was given intramuscularly. If
a chlamydial infection was confirmed, the patient received
oral doxycycline or erythromycin. In either case, the
subject still received the topical eye medication as assigned
by the alternation.
If, despite treatment, the infection worsened to the
point of developing a corneal ulcer, the treatment regimen
was discontinued. The eye was recultured and treated as
deemed appropriate by the ophthalmologist.
The course of treatment was graded both by the oph-
thalmologist and the patient or family. All ophthalmologic
examinations were conducted directly or under the direct
supervision of a senior author (M.V.). The ophthalmolo-
gist graded the infection weekly, or as close to weekly as
possible, in four categories: redness of the conjunctiva,
swelling of the conjunctiva, discharge of the conjunctiva,
and swelling of the eyelid. Each category was scored from
0 to 5, corresponding to none (0), mild (1), moderate (3),
or severe (5). A maximum score was 20 for each infected
eye. A “cure” was defined by the ophthalmologist as a score
of 0.
To assess recovery more finely than achieved on the
weekly ophthalmologic examination, the patient or family
member was asked to grade the infection daily. Each family
was given a color brochure printed in either English or
Tagalog that contained four pictures of eyes illustrating no
inflammation, mild inflammation, moderate inflammation,
and severe inflammation. The patient or family recorded
the evaluation of the degree of inflammation, the date and
time of instillation of the eyedrop, and any adverse
reaction on the back of the brochure daily. To encourage
compliance, each family was given a small gift and travel
reimbursement at each weekly visit to the eye clinic.
The study was masked on different levels. The medica-
tion bottles were identical and not labeled. Thus, the
subjects did not know the name of their medication. Only
the study nurse who dispensed the bottles knew the
medication used by each subject. She instructed the
patients and family not to disclose the appearance of the
medication to the ophthalmologist. The examining oph-
thalmologist and the microbiologist were also unaware of
the identity of the medication used. Although povidone–
iodine stains the eye brown, the color is gone within 5
minutes and would not provide any information to the
examining ophthalmologist.
VOL. 134, NO. 5 POVIDONE–IODINE FOR INFECTIOUS
As a standard clinical trial, power analysis was based on
the differences between treatment groups on time to cure
based on nonparametric tests. Assuming a standard devi-
ation of 3 days to have 80% power to detect a difference of
1 day with a significance level of at least 0.05, we would
need at least 143 subjects in each arm. The treatment
groups were compared for numbers of days until resolution
of the infection as judged separately by the patient and the
ophthalmologist; degree of inflammation, as noted weekly
by the ophthalmologist, number of treatment failures, and
any treatment complications. The analyses were run sep-
arately by type of infecting organism. Simple between-
group comparisons were done using the Wilcoxon Mann-
Whitney or ␹2 tests. In addition, the treatment groups
were compared on the number of subjects who completed
treatment by weeks 1, 2, or 3 using ␹2 tests. The confound-
ing effect of a patient’s age, gender, or duration of infection
before treatment on the primary outcomes was tested using
Wilcoxon Mann–Whitney tests. If any of these tests were
significant, the interaction between the confounder and
treatment was tested using polychotomous logistic regres-
sion. The outcome for the logistic regression models was
week to cure (1, 2, or 3). A parallel lines regression model
based on the cumulative probabilities of the time to cure
was also used. The independent variables were treatment,
age or age group, and onset. The statistical software used
was SAS (SAS Institute Inc, Cary, NC). A two-sided P
value of .05 or less was considered significant.
RESULTS
IN ALL, 459 PATIENTS (MEAN [SD] AGE, 6.6 [6.6] YEARS;
range, 7 months to 21 years; 258 male and 201 female)
were enrolled in the study. Because 288 (62.7%) of the
cases were bilateral, the total number of eyes treated was
747. Ninety-one percent of the subjects were younger than
18 years of age. There were no statistically significant age
or gender differences between the two groups (Table 1).
Povidone–iodine was used by 230 patients and the antibi-
otic by 229. Eighty-eight children did not complete the
protocol— 46 in the povidone–iodine group and 42 in the
antibiotic group (P ⫽ .74). These dropout subjects, who
were largely from rural villages and did not return to the
treatment center, were considered to be missing at random.
Because the number of patients not completing the proto-
col was almost identical in each group, even if they were
considered as treatment failures, the study conclusions
would not have been impacted. The proportion of type of
infection was similar in each group. For bacterial infec-
tions, the causative organisms are presented in Table 2.
Haemophilus species was the organism most frequently
cultured.
There were no significant effects of gender or duration of
infection on the time to cure as evaluated by the physician
or patient. There was an overall effect of age in the group
CONJUNCTIVITIS IN CHILDREN 683
 There was no significant difference between the two
TABLE 1. Characteristics and Outcomes of the Treatment treatment groups for the overall number of days until cured
Groups or for any specific class of microorganism (P ⫽ .150 to
0.826). Chlamydial cases took longer to cure than either
Povidone–Iodine Antibiotic Total
viral or bacterial infections regardless of treatment group
Demographic (P ⫽ .0001).
Age, mean (SD) 7.0 (7.0) 6.0 (6.1) By 7 or 14 days of treatment, the proportion of eyes
Male 129 129 258 considered cured by either the ophthalmologist or patient
Female 101 100 201 was the same for each treatment group, except for chla-
Total 230 229 459 mydial infections (Tables 5 and 6). According to the
Etiology of infection* patient’s evaluation, povidone–iodine-treated chlamydial
Bacteria 68 56 124
infections had marginally more cured by day 7 (P ⫽ .057)
Chlamydia 17 9 26
than the antibiotic. Whether evaluated at 7 or 14 days,
Virus 98 123 221
Outcome
fewer chlamydial infections were cured than those of
Cured* 179 186 365 bacterial or viral etiology (P ⫽ .001). No significant
Drop-out 43 41 84 difference was found between the proportions of viral and
Misdiagnosis 3† 1† 4 bacteria infections cured by 7 or 14 days for either
Worsened* 5† 1† 6 treatment group (P ⫽ .133–.824 for the four comparisons).
By the end of the study, excluding dropouts and three
SD ⫽ standard deviation. misdiagnosed patients (nasolacrimal duct obstructions and
*Subjects who completed the protocol.
†
allergic conjunctivitis), the proportion cured was 97.2%
The numbers of worsened and misdiagnosed cases were too
for the povidone–iodine group and 99.4% for the antibi-
small for valid statistical comparison.
otic group (Table 1). The resolution of swelling, redness,
and discharge as graded by the ophthalmologist, was not
statistically different between treatment groups regardless
TABLE 2. Bacterial Species Cultured of the type of infecting organism (P ⫽ .157–.981). The age
of the patient correlated with the number of days until
Positive Proportion of
Species Cultures* Total (%)
cured. The younger patients overall resolved their con-
junctivitis faster (R ⫽ 0.13, P ⫽ .013).
Haemophilus sp. 35 27.1 Only one patient returned with a new infection after
Moraxella sp. 21 16.3 having been declared cured. Because he returned more
Staphylococcus aureus 18 14.0
than 2 months after treatment, it was believed that this
Streptococcus pneumoniae 15 11.6
was a new infection and not a recurrence.
Veridans streptococcus 9 7.0
Of the six patients who worsened on therapy, one used
Staphylococcus epidermidis 8 6.2
Acinetobacter sp. 4 3.1
the antibiotic and five used povidone–iodine. The infect-
Beta streptococcus 4 3.1 ing organism in the antibiotic case was Staphylococcus
Neisseria gonorrhoeae 4 3.1 epidermidis, while those in the povidone–iodines cases were
Others (ⱕ2 cultures per species) 11 8.5 one each of chlamydia, virus, Staphylococcus aureus, Staph-
Total 129 100 ylococcus epidermidis, and Haemophilus species. The num-
bers of worsened cases were too small for valid statistical
*Some patients had more than one positive culture. comparison. No case of allergic reaction was reported.

diagnosed with bacterial infection, but there was no

interaction of age with medication. Therefore, the results
reported are from the two-group tests without covariates.
The period for treatment and follow-up was 7 to 21 days.
The number of days until cure for all microorganisms, as
evaluated by the ophthalmologist, is presented in Table 3
and as evaluated by the patient in Table 4. As evaluated by
the ophthalmologist, the mean (SD) number of days until
cure was 9.4 (3.6) for all povidone–iodine treated eyes and
9.1 (3.2) for all antibiotic treated eyes. As evaluated by the
patient, the corresponding figures were 6.2 (3.5) and 5.9
(3.3). Whether the conjunctivitis was caused by bacteria,
chlamydia, or viruses, the patients thought that they were
cured earlier than did the ophthalmologist.
DISCUSSION
MOST CASES OF PEDIATRIC CONJUNCTIVITIS IN DEVELOPED
countries are bacterial in origin, as demonstrated in reports
from New York and Seattle (80% and 75%, respectively),
but rarely lead to vision loss since they usually respond to
treatment.16,17 In 1984, Gigliotti and coworkers18 found
that topical polymyxin– bacitracin hastened cure in 102
children compared with no treatment. Leibowitz and
associates19 reported neomycin sulfate–polymyxin B sulfate
to be more effective than placebo. Leibowitz20 later re-
ported tobramycin and ciprofloxacin to each be better than
placebo. Polymyxin– bacitracin was also found to be supe-

684 AMERICAN JOURNAL OF OPHTHALMOLOGY NOVEMBER 2002

 TABLE 3. Number of Days Until Cured as Evaluated by the Ophthalmologist

Second Quartile
Organism Medication Mean (SD) Minimum (Median) Maximum

Bacteria Povidone–I 9.0 (3.7) 7 7 22

Antibiotic 8.6 (2.7) 7 7 15
Chlamydia Povidone–I 14.2 (3.8) 7 14 21
Antibiotic 14.5 (4.6) 7 14 21
Virus Povidone–I 8.8 (2.8) 7 7 18
Antibiotic 9.0 (3.0) 7 7 19

SD ⫽ standard deviation.

TABLE 4. Number of Days Until Cured as Evaluated by the Patient

Second Quartile
Organism Medication Mean (SD) Minimum (Median) Maximum

Bacteria Povidone–I 5.8 (3.3) 1 5 19

Antibiotic 5.4 (3.0) 1 5 12
Chlamydia Povidone–I 9.8 (4.7) 1 10 20
Antibiotic 12.9 (4.2) 6 12 18
Virus Povidone–I 5.8 (3.0) 0 5 14
Antibiotic 5.7 (2.8) 1 5 13

SD ⫽ standard deviation.

TABLE 5. Percent Cured by Day 7

As Evaluated by Medication Bacteria Chlamydia Virus Overall

Ophthalmologist Povidone–Iodine 60.3 12.5 56.3 53.7

Antibiotic 63.6 12.5 52.0 53.8
Total 61.9 12.5 53.9 53.8
Patient Povidone–Iodine 83.6 30.8* 77.2 75.5
Antibiotic 77.6 14.3* 78.6 75.6
Total 80.8 25.0 78.0 75.6

*The difference for treatment of chlamydia was marginally significant (P ⫽ 0.057).

rior to no treatment.18 With this evidence, we believed
that the inclusion of a placebo treatment arm in our study
was not justified and might have been unethical.21
Lohr and associates22 found trimethoprim–polymyxin,
gentamycin, and sodium sulfacetamide to be equally effec-
tive in treating bacterial conjunctivitis. A Norwegian
study of 340 children and adults with acute conjunctivitis
found that cure was achieved in approximately 6.5 days
whether fusidic acid or chloramphenicol eyedrops was
given.23 In 1997, Gross and associates24 found that approx-
imately 90% of 147 patients, whether treated with cipro-
floxin or tobramycin eyedrops, were cured by day 7.
Despite the success of antibacterial treatment, Block and
associates3 recently raised concern over increasing resis-
tance to antibiotic treatment of pediatric conjunctivitis in
the United States.
In our study, neomycin-polymyxin-B-gramicidin oph-
thalmic solution was chosen as the control drug because it
is customarily used in the Philippines for infectious con-
junctivitis. The drug also is frequently used to treat
conjunctivitis in industrialized as well as other underde-
veloped countries. In a recent Bangladeshi study, neomy-
cin alone was found effective against 68% of all bacterial
isolates.25 Gibson26 reported the combination of neomy-
cin-polymyxin-B-gramicidin more effective than chloram-
phenicol in a double-masked randomized trial. The
combination was as effective as fusidic acid in another
trial.27 This eyedrop is considered effective against organ-

VOL. 134, NO. 5 POVIDONE–IODINE FOR INFECTIOUS CONJUNCTIVITIS IN CHILDREN 685

 TABLE 6. Percent Cured by Day 14
As Evaluated by Medication
Ophthalmologist Povidone–Iodine
Antibiotic
Total
Patient Povidone–Iodine
Antibiotic
Total
isms responsible for at least 92% of the bacterial infections
in our series (Table 2).28,29 It is, however, considered
ineffective against chlamydia or viruses.
Among bacteria, we found the most common causative
bacterial species in Manila was Haemophilus species fol-
lowed by Moraxella species (Table 2). Immunization
against Haemophilus in Manila is not required by the
government and is therefore only utilized in the minority
of children whose parents are willing to pay directly. The
next two were Staphylococcus aureus and Streptococcus
pneumonia. Cultural differences in hygiene may explain
this order of frequency of bacterial species, as well the
prominence of viral infections in this study. In the United
States, bacteria cause approximately 80% of acute con-
junctivitis cases in children.17 In our study, however,
almost 60% of the cases were attributed to viruses. In the
Philippines, it is common to have multiple family members
present simultaneously or sequentially with viral conjunc-
tivitis, presumably due to easy spread of the virus among
them. We believe this is less common in the United States.
Valenton and associates13 suggested that the high preva-
lence of viral conjunctivitis in their 1993 Manila study
might have occurred because patients were less disturbed
by bacterial conjunctivitis and frequently did not seek
medical attention.
All the previously mentioned investigations of povi-
done–iodine studied prophylaxis of infection. We are not
aware of previous controlled studies dealing with the use of
povidone–iodine to treat ongoing infections. In 1985,
Schuhman and Vidic30 treated 40 patients with conjunc-
tivitis or keratoconjunctivitis with a 3-mg/100 ml solution
of povidone–iodine instilled 3 to 12 times a day either
alone or in conjunction with betamethasone eyedrops.
There were no controls, masking, or randomization. A
“marked improvement” was found in 28 patients by day 3
and 35 patients were cured by day 6. Two patients
complained of ocular discomfort after instillation of the
eyedrop. In 1995, Duffey31 successfully treated one patient
who developed a Serratia marcescens keratitis after radial
keratotomy with 5% povidone–iodine ophthalmic solution
along with tobramycin and ciprofloxin. Aras and associ-
ates32 successfully treated a single culture-negative eye
infection that followed a laser in situ keratomileusis
686 AMERICAN JOURNAL
Bacteria Chlamydia Virus Overall
89.1 68.8 96.9 91.5
96.4 75.0 95.1 94.6
92.4 70.8 95.9 93.1
98.4 93.8 100 98.9
100 75.0 100 98.9
99.2 87.5 100 98.9
procedure with 5% povidone–iodine solution, as well as
cefazolin, vancomycin, and ciprofloxin.
In 1998, Abel and Abel33 reported the treatment of 36
children and adults with presumed adenoviral keratocon-
junctivitis. Twenty-one patients had received previous
treatment with topical corticosteroids with or without
antibiotics. Thirty patients were then treated only with
10% povidone–iodine eyelid scrubs. Six patients received
5% povidone–iodine solution 3 times a day for 2 days.
Follow-up communication was made by telephone unless
they had not improved. There were no controls, random-
ization, or masking. Thirty-four patients had improved by
the second day of treatment.
Our project is the first controlled and masked study to
evaluate the efficacy of povidone–iodine to treat an ocular
infection. As most external eye infections begin with acute
conjunctivitis, we began our treatment investigation with
this disorder. The 1.25% concentration of povidone–
iodine was chosen for this study based on our experience in
the postoperative use of povidone–iodine.9 In that study,
we began with the use of a 2.5% concentration, but
decreased it to 1.25% when a few subjects complained of
mild stinging when the eyedrop was used several times a
day. The drug is still highly germicidal when used at the
lower concentration. Subsequent patients in the postoper-
ative study and in this investigation tolerated the 1.25%
concentration well.
We found the number of days needed for cure, whether
evaluated by the ophthalmologist or the patient, to be the
same for either medication for bacterial and viral infec-
tions. Thus, povidone–iodine was as effective as the
antibiotic against bacterial conjunctivitis. Bacteria gener-
ally adhere to the surfaces of infected tissue in biofilms
within a glycocalyx matrix.34 Povidone–iodine can reach
the biofilm and has been shown to cause the outer layers of
the biomass to slough and be killed.35 It not only destroys
a wide range of bacteria, but also inactivates and inhibits
the release of bacterial exotoxins.36 The superficial loca-
tion of bacteria may provide susceptibility to the effects of
povidone–iodine.
In this study, povidone–iodine was found as ineffective
as the antibiotic against viral conjunctivitis. Viruses begin
replication by attaching to and penetrating a host cell.
OF OPHTHALMOLOGY NOVEMBER 2002
Their position within host cells may protect them from
contact from povidone–iodine better than chlamydia. An
antiviral agent must traverse the membrane of the host cell
in sufficient quantity to be effective. Perhaps if contact
time were increased, povidone–iodine might be able to
penetrate the cell and affect viruses. Another consider-
ation is the timing of therapy. Viral replication usually
peaks during the incubation period and may already be in
decline when symptoms become apparent.37 The povi-
done–iodine solution may not reach the conjunctiva early
enough to shorten the duration of the disease.
Whereas the patients evaluated their ocular inflamma-
tion daily while the ophthalmologist examined the pa-
tients weekly, the time to cure was generally shorter
according to the patients. This effect was particularly
noticeable in the case of chlamydial infections, where by
the 7-day examination, marginally more patients using
povidone–iodine judged themselves cured than those
treated by the antibiotic (P ⫽ .57). The ophthalmologist,
whose first examination since treatment began was gener-
ally conducted on day 7, found a similar proportion in both
treatment groups. By the time they reached the ophthal-
mologist for the 14-day examination, the same proportion
was cured in both treatment groups. The elementary body
form of Chlamydia trachomatis is internalized by the host
cell, where it remains within the phagosome, undergoes
fission, and becomes the inclusion body.38 Despite this
intracellular location, chlamydia were found to be some-
what vulnerable to povidone–iodine. In our previous study
in Kenya, povidone–iodine was found very effective
against chlamydia present on the ocular surface directly
after birth.10
For bacterial infections in this study, the number of days
until cure by the patient’s evaluation, regardless of treat-
ment group, was approximately 5.5 days. This figure is
close to the 6- to 7-day figure found in the literature.18,23
Even by the ophthalmologist’s examination, more than
60% of the patients were cured of bacterial conjunctivitis
by one week. By the patient’s evaluation, povidone–iodine
was also somewhat more effective against chlamydial
infections. Thus, for bacterial and chlamydial conjuncti-
vitis, povidone–iodine 1.25% solution is an attractive
medication to use.
For viral conjunctival infections, povidone–iodine was
not effective. Clinicians should still treat viral infections
only symptomatically. If, however, the clinician is unsure
of the clinical diagnosis from the examination, povidone-
iodine ophthalmic solution could be considered, at least
until the culture or immunofluorescent test results are
received.
Our results should be considered in light of some
limitations of the study. While we recorded no allergic
responses to either treatment arm, patients who may have
developed eyelid edema or other allergic reactions possibly
dropped out of the study by deciding not to return for
follow-up. Such patients would have been listed as “drop-
VOL. 134, NO. 5 POVIDONE–IODINE FOR INFECTIOUS
outs” in our analysis. Our population size may have been
too small to determine adverse events. The alternation
method was chosen to allocate patients to treatment arms.
Since the groups were similar regarding demographic
variables and type of infection, there would be little
confounding with this method of assignment. A number of
children were lost to follow-up, but the proportion was the
same in each treatment group.
In conclusion, povidone–iodine 1.25% ophthalmic so-
lution was as effective as neomycin-polymyxin B-gramici-
din for treating bacterial conjunctivitis, somewhat more
effective against chlamydia, and as ineffective against viral
conjunctivitis. Povidone-iodine ophthalmic solution
should be strongly considered as treatment for bacterial
and chlamydial conjunctivitis, especially in developing
countries where topical antibiotics are often unavailable or
costly.
REFERENCES
1. Foster A. How can blind children be helped? Community
Eye Health 1998;11:33–34.
2. Steinkulller PG, Du L, Gilbert C, Foster A, Collins ML,
Coats DK. Childhood blindness. J AAPOS 1999;3:26 –32.
3. Block SL, Hedrick J, Tyler R, et al. Increasing bacterial
resistance in pediatric acute conjunctivitis (1997–1998).
Antimicrob Agents Chemother 2000;44:1650 –1654.
4. Schreier H, Erdos G, Reimer K, Konig B, Konig W, Fleischer
W. Molecular effects of povidone–iodine on relevant micro-
organisms: an electron-microscopic and biochemical study.
Dermatology 1997;195:111–116.
5. Apt L, Isenberg SJ, Yoshimori R, Spierer A. Outpatient
topical use of povidone–iodine in preparing the eye for
surgery. Ophthalmology 1989;96:289 –292.
6. Isenberg SJ, Apt L, Yoshimori R, Khwarg S. Chemical
preparation of the eye in ophthalmic surgery. IV. The
antibacterial effect on the conjunctiva of povidone–iodine
compared with a prophylactic antibiotic. Arch Ophthalmol
1985;103:1340 –1342.
7. Apt L, Isenberg SJ, Paez JH, Yoshimori R. Chemical prepa-
ration of the eye in ophthalmic surgery: III. Effect of
povidone–iodine on the conjunctiva. Arch Ophthalmol
1984;102:728 –729.
8. Apt L, Isenberg SJ, Yoshimori R, et al. The effect of
povidone–iodine solution applied at the conclusion of oph-
thalmic surgery. Am J Ophthalmol 1995;119:701–705.
9. Isenberg SJ, Apt L, Yoshimori R, Pham CB, Lam NK.
Efficacy of topical povidone–iodine during the first week after
ophthalmic surgery. Am J Ophthalmol 1997;124:31–35.
10. Isenberg SJ, Apt L, Wood M. A controlled trial of povidone–
iodine as prophylaxis against ophthalmia neonatorum.
N Engl J Med 1995;332:562–566.
11. Benevento WJ, Murray P, Reed, CA, Pepose JS. The
sensitivity of Neisseria gonorrhoeae, Chlamydia trachomatis,
and herpes simplex type II to disinfection with povidone–
iodine. Am J Ophthalmol 1990;109:329 –333.
12. Kawana R, Kitamura T, Nagakomi O, et al. Inactivation of
human viruses by povidone–iodine in comparison with other
antiseptics. Dermatology 1997;195:29 –35.
13. Valenton MJ, Dy-Liacco JU, Espiritu CG. Cornea and
CONJUNCTIVITIS IN CHILDREN 687
 external eye disease problems in the Philippines. A twenty
year survey (1971–1991). Phil J Ophthalmol 1993;22:1– 69.
14. Brisner JH, Weiss A. Laboratory diagnosis in ocular disease.
In: Duane TD, editor. Clinical ophthalmology. New York:
Lippincott-Raven, 1997;4:5.
15. Shiuey S, Ambati BK, Adamis AP. A randomized double-
masked trial of topical ketorolac versus artificial tears for
treatment of viral conjunctivitis. Ophthalmology 2000;107:
1512–1517.
16. Gigliotti F, Williams WT, Hayden FG, et al. Etiology of
acute conjunctivitis in children. J Pediatr 1981;98:531–536.
17. Weiss A, Brinser JH, Nazar-Stewart V. Acute conjunctivitis
in childhood. J Pediatr 1993;122:10 –14.
18. Gigliotti F, Hendley JO, Morgan J, Michaels R, Dickens M,
Lohr J. Efficacy of topical antibiotic therapy in acute con-
junctivitis in children. J Pediatr 1984;104:623– 626.
19. Leibowitz HM, Pratt MV, Flagstad IJ, Berrospi AR, Kundsin
R. Human conjunctivitis. II. Treatment. Arch Ophthalmol
1976;94:1752–1756.
20. Leibowitz HM. Antibacterial effectiveness of ciprofloxacin
0.3 per cent ophthalmic solution in the treatment of bacte-
rial conjunctivitis. Am J Ophthalmol 1991;112:29s–33s.
21. Vastag B. Helsinki discord? A controversial declaration.
JAMA 2000;284:2983–2985.
22. Lohr J, Austin BA, Grossman M, Hayden GF, Knowlton
GM, Dudley SM. Comparison of three topical antimicrobials
for acute bacterial conjunctivitis. Pediatr Infect Dis J 1988;
7:626 – 629.
23. Horven I. Acute conjunctivitis. A comparison of fusidic acid
viscous eyedrops and chloramphenicol. Acta Ophthalmo-
logica 1993;71:165–168.
24. Gross RD, Hoffman RO, Lindsay RN. A comparison of
ciprofloxacin and tobramycin in bacterial conjunctivitis in
children. Clin Pediatr 1997;36:435– 444.
25. Iqbal J, Rahman M, Kabir MS, Sharif AJ. Comparative
in-vitro activities of topical antibiotics against conjunctival
isolates. J Antimicrob Chemother 1999;43:611– 612.
26. Gibson JR. Trimethoprim-polymyxin B ophthalmic solution
in the treatment of presumptive bacterial conjunctivitis—a
688 AMERICAN JOURNAL
multicentre trial of its efficacy versus neomycin-polymyxin
B-gramicidin and chloramphenicol ophthalmic solutions. J
Antimicrob Chemother 1983;11:217–221.
27. Dy-Liacco JU, Cruz-Nievera LF, Thorn P. A comparison of
fusidic acid 1 per cent viscous eye drops (Fucithalmic, Leo)
and Neosporin eye ointment (Wellcome) in patients with
external eye infections. Int J Clin Pract 1991;7:81– 83.
28. Weiss A. Acute conjunctivitis in childhood. Curr Probl
Pediatr 1994;24:4 –11.
29. Ophthalmic drug facts 2000. St. Louis: Wolters Kluwer,
2000:106.
30. Schuhman G, Vidic B. Clinical experience with povidone-
iodine eyedrops in patients with conjunctivitis and kerato-
conjunctivitis. J Hosp Infect 1985;6:173–175.
31. Duffey RJ. Bilateral Serratia marcescens keratitis after simul-
taneous bilateral radial keratotomy. Am J Ophthal 1995;119:
233–234.
32. Aras C, Ozdamar A, Bahcecioglu H, Sener B. Corneal
interface abscess after excimer laser in situ keratomileusis. J
Refract Surg 1998;14:156 –157.
33. Abel R, Abel AD. Use of povidone–iodine in the treatment
of presumptive adenoviral conjunctivitis. Ann Ophthalmol
1998;30:341–343.
34. Akiyama H, Torigoe R, Aratwa J. Interaction of Staphyloc-
cocus aureus cells and silk threads in vitro and in mouse skin.
J Dermatol Sci 1993;6:247–257.
35. Brown ML, Aldrich HC, Gauthier JJ. Relationship between
glycocalyx and povidone–iodine resistance in Pseudomonas
aeuginosa (ATCC 27853) biofilms. Appl Environ Microbiol
1995;61:187–193.
36. Konig B, Reimer K, Fleischer W. Konig W. Effects of
Betaisodona on parameters of host defense. Dermatology
1997;195:42– 48.
37. Galasso GJ. Antiviral drugs. In: Notkins AL, Oldstone
MBA, editors. Concepts in viral pathogenesis. New York:
Springer-Verlag, 1984:382–388.
38. Hodinka RL, Davis CH, Choong J, Wyrick PB. Ultrastruc-
tural study of endocytosis of Chlamydia trachomatis by
McCoy cells. Infect Immun 1988;56:1456 –1463.
OF OPHTHALMOLOGY NOVEMBER 2002