Ophthalmic antibiotic use for acute infectious conjunctivitis in children

Abstract

Background

Infectious conjunctivitis is among the most common pediatric infections worldwide; antibiotics are often not indicated. We aimed to determine factors associated with ophthalmic antibiotic prescribing and changes in prescribing prior to and during the COVID-19 pandemic at a single center.

Methods

Encounters for children with infectious conjunctivitis from 2017 to 2020 at Denver Health and Hospital Authority clinics were analyzed retrospectively. Factors associated with prescribing were evaluated using multivariable logistic regression modeling. Encounter numbers and prescribing patterns for telephone versus in-person visits before and during the pandemic were compared and stratified.

Results

Of 5,283 patients encounters for conjunctivitis, 3,841 (72.7%) resulted in an ophthalmic antibiotic prescription. Concurrent diagnosis with acute otitis media (adjusted odds ratio [aOR] 0.20 (95% CI, 0.16–0.25) and later study year (2018-aOR = 0.76 [95% CI, 0.65–0.89]; 2019- aOR = 0.57 [95% CI, 0.48–0.67]) were associated with reduced odds of prescribing. Compared with those evaluated in pediatric clinics, patients evaluated in family medicine (aOR = 0.69 [95% CI, 0.58–0.83]) or optometry/ophthalmology clinics (aOR = 0.06 [95% CI, 0.02–0.14]) were less likely to have antibiotics prescribed, whereas, patients evaluated via telephone had a 5.43 (95% CI, 3.97–7.42) greater odds of being prescribed ophthalmic antibiotics. Antibiotic prescribing increased from 67.8% prior to the COVID-19 pandemic to 81.9% during the pandemic (P < 0.0001).

Conclusions

Discordant with national guideline recommendations, ophthalmic antibiotic use for conjunctivitis was high. Telephone visits were associated with higher rates of prescribing. Rates of prescribing increased significantly during the COVID-19 pandemic.

Acute infectious conjunctivitis is one of the most common pediatric infections worldwide, affecting 1 out of every 8 children annually and resulting in over $857 million in direct and indirect costs per year in the United States.^1,2

Most cases of acute conjunctivitis are self-limited and do not require antibiotics. Although pediatric data are limited, studies suggest that a large proportion or even a majority of infectious conjunctivitis is caused by viruses, including adenovirus.^3–6 Bacterial conjunctivitis tends to be mild and self-resolving, and in cases where ophthalmic antibiotics are prescribed, the benefit is only marginal with a number needed to treat (NNT) of 9–11.⁷

Ophthalmic antibiotics are not without risk. Adverse drug reactions caused by ocular surface toxicity are reported by 8% of patients.^8,9 Additionally, antibiotic resistance is a growing problem in conjunctivitis management. Though ocular-specific breakpoints for antibiotic susceptibility have not been established, clinical trials have demonstrated a gradual decline in the effectiveness of most commonly prescribed ophthalmic antibiotics.^10,11 Acquired resistance for both pathogenic and colonizing bacteria have been demonstrated in the conjunctiva of children prescribed antibiotics.^10,12,13 Furthermore, ophthalmic antibiotics likely disrupt the normal protective microbiota of the eye, which can place children at higher risk for future infections.^14–16

Because of these concerns, the American Academy of Ophthalmology recommends against prescribing ophthalmic antibiotics for viral conjunctivitis and deferring immediate ophthalmic antibiotic therapy in cases where the cause of conjunctivitis is unknown.¹⁷ Nevertheless, it is estimated that 60%–80% of combined adult and pediatric patients with acute conjunctivitis are prescribed broad-spectrum ophthalmic antibiotics.^18–20 This parallels United States systemic antibiotic prescribing trends: it is estimated that at least 30% of antibiotics prescribed in outpatient settings are not needed.²¹ Overall, ophthalmic prescribing rates in children are poorly understood.

Despite widespread ophthalmic antibiotic use for acute conjunctivitis, little is known about patient and system-level factors associated with ophthalmic antibiotic prescribing for children. Furthermore, it is unknown how the dramatic shift from in-person to telehealth care during the COVID-19 pandemic is associated with ophthalmic antibiotic prescribing behavior. In an effort to inform pediatric quality improvement and antimicrobial stewardship efforts, we assessed trends in ophthalmic antibiotic prescribing within a large, integrated safety-net health care system and determined factors associated with ophthalmic antibiotic prescribing.

Subjects and Methods

The project was approved by the Colorado Multiple Institute Review Board. This study analyzed data from children age 6 months to 18 years of age with acute conjunctivitis seen at Denver Health and Hospital Authority (DHHA) in Denver, Colorado, from January 2017 to July 2020. DHHA is a large, urban, safety-net health care system that includes 28 federally qualified health care centers, 3 urgent care centers, and a level one trauma center that includes a pediatric emergency department. In 2018, racial and ethnic demographics for all patients who presented at DHHAwere 58% white, 26% black, and 15% other/unknown race; 47% of patients identified as Hispanic.²² Among all DHHA patients, 16.0% were uninsured, and 60.9% had Medicaid or Children’s Health Insurance Program.²³

First, a primary analysis was completed that evaluated factors associated with ophthalmic antibiotic prescribing use from 2017 to 2019. A post hoc analysis was then completed to compare antibiotic prescribing prior to the emergence of COVID-19 (2017–2019) to antibiotic prescribing during the pandemic (March-July 2020).

Patients were included in the primary analysis whether they had a phone or in-person encounter with an associated International Classification of Diseases (ICD) version 10 code for acute infectious conjunctivitis (B30 [viral conjunctivitis], H10.0 [mucopurulent conjunctivitis], H10.23 [serous conjunctivitis, except viral], H10.3 [unspecified acute conjunctivitis], H10.89 [other conjunctivitis], H10.9 [unspecified conjunctivitis]) from 2017 to 2019. DHHA has a nurse-line protocol that permits nurses to prescribe antibiotics for conjunctivitis over the phone under the supervision of a medical director. Therefore, nurse-line visits were included in the analysis. Patients who had a conjunctivitis diagnosis (using above ICD 10 codes) or were prescribed ophthalmic antibiotics in the 90 days prior to the encounter were excluded. Because our primary aim was to evaluate antibiotic prescribing in uncomplicated conjunctivitis cases, patients with a history of ophthalmic surgeries, history of chronic conjunctivitis (ICD H10.4), or those who were diagnosed in inpatient settings were excluded.

The primary outcome measure was a prescription for an ophthalmic antibiotic on the day of the encounter. First, descriptive statistics evaluating the frequency of antibiotic types prescribed and trends in antibiotic prescribing rates over the study period were computed. Factors associated with antibiotic prescribing were evaluated by means of bivariate comparative tests using χ² or Fisher exact tests. Finally, multivariable logistic regression modeling was used to adjust for covariates. Factors that had a P value of < 0.1 on bivariate analyses were included in the initial model, and backward selection was used to fit the model. The final model included age, race, acute otitis media concurrent diagnosis, clinical location, and study year. Sex, ethnicity, language preference, insurance, smoke exposure, prior conjunctivitis diagnosis, laterality, and season were not included in the final model. Significance was defined as an alpha of 0.05 using two-tailed tests.

Results

A total of 5,283 patient encounters from 2017 to 2019 were included in the primary analysis, 3,841 (72.7%) of which were prescribed ophthalmic antibiotics. Features of included patients are provided in Table 1. The number of cases of conjunctivitis decreased from 1,878 in 2017 to 1,554 in 2019. The rate of ophthalmic antibiotic prescribing steadily decreased from 77.0% in 2017 to 67.8% in 2019 (Figure 1). The most common ophthalmic antibiotic prescribed was polymixin B-trimethoprim (2,822 [73.5%]) followed by erythromycin (677 [17.6%]; Table 2). No patients were prescribed an ophthalmic steroid.

Table 1.

Features of conjunctivitis patients who were and were not prescribed an ophthalmic antibiotic

Feature	No antibiotic, no. (%) n = 1,442	Antibiotic, no. (%) n = 3,841	P valuea
Age
6 mos to 23 mos	342 (23.7)	877 (22.8)	0.07
24 mos to 5 years	558 (38.7)	1375 (35.8)
6–10 years	295 (20.5)	896 (23.3)
11–18 years	247 (26.3)	693 (18.0)
Male	749 (51.9)	1906 (49.6)	0.13
Race
Black	166 (11.5)	497 (12.9)	0.06
White	1152 (79.9)	3078 (80.1)
Other/unknown	124 (8.6)	266 (6.9)
Hispanic	1028 (71.3)	2626 (68.4)	0.04
Language preference
English	973 (67.5)	2719 (70.8)	0.009
Spanish	410 (28.4)	1017 (26.5)
Other	59 (4.1)	105 (2.7)
Insurance
Commercial	221 (15.3)	631 (16.4)	0.33
Public/self pay	1221 (84.7)	3210 (83.6)
Clinical setting
Pediatrics	585 (40.6)	1471 (38.5)	<0.0001
Family medicine	289 (20.0)	521 (13.6)
Emergency medicine	490 (34.0)	1128 (29.4)
Ophthalmology/optometry	30 (2.1)	6 (0.2)
Telephone/nurse line	48 (3.3)	715 (18.6)
Smoke exposure	100 (6.9)	218 (5.7)	0.09
Acute otitis media
In prior year	200 (13.9)	420 (11.0)	0.003
Co-diagnosis	292 (20.3)	170 (4.4)	<0.0001
Conjunctivitis in prior year	215 (27.3)	97 (2.5)	<0.0001
Laterality
Unilateral	216 (15.0)	1031 (26.8)	<0.0001
Bilateral	382 (26.5)	1440 (37.5)
Unknown	844 (58.5)	1370 (35.7)
Season
Fall	244 (16.9)	692 (18.0)	0.56
Spring	521 (36.1)	1406 (36.6)
Summer	228 (15.8)	619 (16.1)
Winter	449 (31.1)	1124 (29.3)
Year
2017	432 (30.0)	1446 (37.7)	<0.0001
2018	509 (35.3)	1342 (34.9)
2019	501 (34.7)	1053 (27.4)

^aCalculated using χ² or Fisher exact tests, as appropriate; significance defined as <0.05 using two-tailed tests.

FIG 1.

Conjunctivitis cases and antibiotic prescribing rates by study year, including COVID-19 encounters from March to July of 2020.

Table 2.

Number and frequency of ophthalmic antibiotics prescribed

Antibiotica	No. (%) N = 3841
Polymixin B-trimethoprim	2822 (73.5)
Erythromycin	677 (17.6)
Ofloxacin	237 (6.2)
Tobramycin	73 (1.9)
Other	32 (0.8)

^aNo patients were prescribed ophthalmic steroids or antibiotic-steroid combination medications.

Factors Associated with Ophthalmic Antibiotic Prescribing

Bivariate analyses of ophthalmic antibiotic prescribing by clinical feature demonstrated differences in prescribing based on ethnicity, language preference, clinical setting, history of acute otitis media (AOM) in the prior year, concurrent diagnosis with AOM, conjunctivitis in the prior year, infection laterality, and study year (Table 1). Of 462 patients with concurrent AOM, 170 (36.8%) were prescribed an ophthalmic antibiotic.

In the adjusted model (Table 3), concurrent diagnosis with AOM (adjusted odds ratio [aOR] = 0.20 [95% CI, 0.16–0.25]) and later study year (2018-aOR = 0.76 [95% CI, 0.65–0.89]; 2019-aOR, 0.57 [95% CI, 0.48–0.67]) were associated with reduced odds of ophthalmic antibiotic prescribing. Patients evaluated in family medicine (aOR = 0.69 [95% CI, 0.58–0.83]) or optometry/ophthalmology clinics (aOR = 0.06 [95% CI, 0.02–0.14]) had reduced odds of having an ophthalmic antibiotic prescribed compared to those evaluated in pediatric clinics. Patients evaluated via telephone encounter or nurse-line encounter had 5.43 (95% CI, 3.97–7.42) greater odds of being prescribed ophthalmic antibiotics than those evaluated in pediatric clinics.

Table 3.

Raw and adjusted odds of being prescribed an ophthalmic antibiotic by clinical feature

Feature	OR (95% CI)a	aOR (95% CI)b
Age
6 mos to 23 mos	0.92 (0.76, 1.11)	1.05 (0.85, 1.29)
24 mos to 5 years	0.88 (0.74, 1.05)	0.89 (0.74, 1.08)
6–10 years	1.08 (0.89, 1.32)	1.08 (0.88, 1.32)
11–18 years	Reference	Reference
Race
Black/African American	1.12 (0.93, 1.35)	1.04 (0.85, 1.27)
Other/unknown	0.80 (0.64, 1.00)	0.77 (0.61, 0.97)
White	Reference	Reference
Acute otitis media co-diagnosis	0.18 (0.15, 0.22)	0.20 (0.16, 0.25)
Location
Emergency department/urgent care	0.92 (0.79, 1.06)	1.02 (0.88, 1.18)
Optometry/ophthalmology	0.08 (0.03, 0.19)	0.06 (0.02, 0.14)
Family medicine	0.72 (0.60, 0.85)	0.69 (0.58, 0.83)
Telephone or nurse line	5.93 (4.36, 8.06)	5.43 (3.97, 7.42)
Pediatrics	Reference	Reference
Year
2019	0.63 (0.54, 0.73)	0.57 (0.48, 0.67)
2018	0.79 (0.68, 0.91)	0.76 (0.65, 0.89)
2017	Reference	Reference

aOR, adjusted OR; CI, confidence interval; OR, odds ratio.

^aUnadjusted OR and 95% CI.

^bAdjusted OR and 95% CI from multivariable regression model. Significance was defined using a two-tailed alpha of 0.05.

Conjunctivitis Prior to and during the COVID-19 Pandemic

A total of 155 patient encounters for conjunctivitis occurred from March to July 2020, representing a 73.2% (155–749) reduction from the same time period in 2019. The proportion of encounters for conjunctivitis that were telephone encounters to the nurse-line or provider increased significantly after the emergence of COVID-19, from 14.4% (762/5283) prior to the COVID-19 pandemic to 60.7% (94/155) during the COVID-19 pandemic (P < 0.0001).

Of patients diagnosed during the COVID-19 pandemic, 81.9% (127/155) were prescribed ophthalmic antibiotics, which was a significant increase from the prior year (67.8% [1053/1554] in 2019; P < 0.0001). All patients evaluated via nurse-line during the COVID-19 pandemic were prescribed ophthalmic antibiotics, which was not statistically different than prior to the COVID-19 pandemic (100% vs 93.8%; P = 0.41). After exclusion of nurse-line visits, providers were more likely to prescribe ophthalmic antibiotics during the COVID-19 pandemic (78.1% [100/128]) than prior to the pandemic (69.1% [3126/5283]; P = 0.04).

Though nearly all nurse-line encounters resulted in an ophthalmic antibiotic prescription, overall ophthalmic antibiotic prescribing (including provider and nurse-line prescribing) was not significantly different between telephonic visits (80.6% [76/94]) and in-person visits (83.6% [51/61]; P = 0.66) during the COVID-19 pandemic. After exclusion of nurse-line visits there was no difference in prescribing rates between in-person and telephone visits completed by providers (84% [51/61] vs 73% [49/67]; P = 0.13), though the analysis was limited by small sample size (128 encounters).

Discussion

We found overall high rates of ophthalmic antibiotic prescribing for acute conjunctivitis that varied significantly by clinical care location and year. Patients who were diagnosed by nurse-line were significantly more likely to have antibiotics prescribed than those with in-person encounters. Ophthalmic antibiotic use had been steadily declining over a 3-year period; however, the start of the COVID-19 pandemic marked a significant reversal in this trend, with an overall decrease in pediatric clinical visits for conjunctivitis, but a marked increase in the proportion of children with conjunctivitis prescribed ophthalmic antibiotics.

Despite national recommendations to limit ophthalmic antibiotic use for conjunctivitis,¹⁷ nearly three-quarters of children at DHHA had an ophthalmic antibiotic prescribed. Local return to daycare and school requirements are a key driver for prescribing for conjunctivitis, because they often require children to be on ophthalmic antibiotics for 24 hours prior to return.^24–27 In contrast, Centers for Disease Control and Prevention (CDC) and American Academy of Pediatrics guidelines recommend exclusion from daycare or school only in cases where patients have systemic symptoms, such as fever, or are unable to avoid close contact with other students.^28,29 It is imperative to recognize that policies that exclude children from school and childcare unnecessarily or require antibiotics that are of limited benefit prior to return are likely to disproportionately affect communities of color and families with lower socio-economic status, and thus contribute to health disparities.^30–32

In this study, 69% of the conjunctivitis encounters were for children who are Hispanic, although Hispanic patients comprise only 47% of the DHHA patient population. They were also less likely to receive ophthalmic antibiotic treatment than patients who were not Hispanic. This could indicate that children who are Hispanic are disproportionately affected by infectious conjunctivitis (or conditions commonly misdiagnosed as infectious conjunctivitis) or are more likely to seek care either from parental preference or need to obtain clearance to return to school or daycare.

Historically, national and institutional antimicrobial stewardship efforts have focused on systemic antibiotic use. However, the widespread use of broad-spectrum ophthalmic antibiotics may contribute to the development of antimicrobial-resistant organisms. Although adverse drug events were not evaluated in this study, given a prior estimate that 8% of ophthalmic antibiotic prescriptions are associated with adverse events, this suggests that over 300 patients over the study period may have experienced an adverse event related to ophthalmic antibiotics.^8,9 In contrast, given the number needed to treat to achieve symptomatic improvement (NNT 9–11), approximately 380 patients would have had some benefit.⁷ Thus, the number harmed from ophthalmic antibiotics use approaches the number benefited. Additionally, although the effect of ophthalmic antibiotics on the microbiome of the eye cannot be easily quantified, it has been conjectured that ophthalmic antibiotic use may predispose patients to future infections.^14–16,33

Diagnostic uncertainty can be an underlying reason for prescription of ophthalmic antibiotics. Although numerous clinical care algorithms have been evaluated to assist with determining whether conjunctivitis is viral or bacterial, no algorithm can reliably distinguish between etiologies.^1,20,34 It is unclear what, if any, role rapid diagnostic tests may have for management of conjunctivitis in children in the future. Providers should recognize that even in cases of bacterial conjunctivitis, antibiotics provide only a small benefit.⁷ Therefore, supportive care, without antibiotics, should be considered.

Surprisingly, in this study, ophthalmic antibiotics were prescribed to nearly 40% of patients with concurrent AOM. Although we did not evaluate oral antibiotic use in this study, at DHHA during the study period, 98% of patients with AOM were prescribed an oral antibiotic; thus, up to 40% of patients with conjunctivitis-otitis syndrome may have unnecessarily received both an oral and ophthalmic antibiotic.³⁷ While national guidelines for AOM describe oral therapy recommendations for conjunctivitis-otitis syndrome, more concrete guidance on ophthalmic antibiotic use would be beneficial.³⁸

Variations in prescribing rates by clinical location were significant. Optometry and ophthalmology providers were most likely to follow current treatment guidelines and prescribed ophthalmic antibiotics the least. While these patients may have had a different clinical presentation, causing them to seek care in ophthalmology compared to those evaluated in other settings, it is likely the severity of presentation would be higher in ophthalmology clinics than other clinical settings. Nearly all patients (94%) who were diagnosed via telephone or nurse-line were prescribed ophthalmic antibiotics. Other studies have also identified higher prescribing rates during telemedicine visits compared with in-person visits.^39,40 Although telephone triage and standing order treatment protocols can be used to rule out pathology that necessitates a higher level of care or evaluation and facilitate return to school, they contributed significantly to ophthalmic antibiotic prescribing. This is a clear target for antimicrobial stewardship, particularly because prescribing patterns in these environments tends to be protocol driven rather than behavior driven.

Although the reasons for the decline in antibiotic use prior to the emergence of COVID-19 are unknown, we speculate that the national and institutional emphasis on reducing unnecessary antibiotic contributed to the decline. No conjunctivitis-specific interventions have been implemented at DHHA. Antibiotic prescribing increased during the COVID-19 pandemic, but this was true of in-person as well as telemedicine visits. These practice patterns will not necessarily persist beyond the pandemic, even if there is a post-pandemic shift toward greater use of telemedicine.

This study has several strengths, including the use of patient-level rather than payer-level data, which allowed us to evaluate and adjust for covariates, such as patient language preference and differences in insurance types that are not available in most payer databases. Because DHHA is an integrated health care system, we were able to evaluate prescribing across a variety of clinical location types (e.g. emergency department, primary care, eye clinic, nurse-line) and provider specialties. We were also able to assess change in prescribing prior to and during the COVID-19 pandemic. This study also has several important limitations. We were not able to assess antibiotic fill rates; thus patients may have not filled prescribed ophthalmic antibiotics. We did not evaluate oral antibiotic utilization for conjunctivitis; thus, it is possible that some conjunctivitis patients may not have been prescribed an ophthalmic antibiotic because they received an oral antibiotic. We were also limited in sample size during the COVID-19 pandemic. Finally, given this is a single center study, the findings may not be generalizable to other organizations. However, prior studies evaluating antibiotic prescribing for other diagnoses suggest that telephone encounters tend to result in higher prescribing rates and the population at DHHA is similar to those at other urban federally qualified healthcare systems, which serve nearly 28 million patients annually.²³

In conclusion, overall ophthalmic antibiotic use for conjunctivitis was high, despite national treatment recommendations against unnecessary ophthalmic antibiotics. Nurse-line visits were associated with higher rates of prescribing because of standing order protocols. Despite a steady decline in ophthalmic antibiotic prescribing for 3 years prior to the emergence of COVID-19, rates of prescribing increased significantly during the COVID-19 pandemic. Finally, providers, health systems, and policy makers should critically evaluate prescribing practices and return to childcare/school recommendations for conjunctivitis, because they likely significantly affect education access, economic burden, and health equity.