Improving Red Reflex Screening in a Level III NICU Through a Quality Improvement-based Approach

This quality improvement project was conducted between March 2020 and March 2022 in a 72-bed level-III academic NICU where pediatric residents, neonatology fellows, and NNPs are the clinical providers and are supervised by neonatology attending physicians. The hospital uses an electronic medical records system for documentation and results. Per unit policy, all patients <32 weeks gestational age and <1500 g birth weight receive ROP screen by a retina specialist at 31 weeks corrected gestational age or at 4 weeks of chronological age (whichever is later). Additionally, every infant 32 to 36 weeks of gestation who receives supplemental oxygen for ≥6 hours also receives an ROP screen at 4 weeks chronological age. We aimed at improving red reflex documentation in neonates discharged from the hospital who were not examined by a retinologist.

We identified the various barriers to the documentation of red reflex before discharge after obtaining feedback from the clinical providers and the nursing staff as mentioned above. The feedback was obtained via informal discussions by the neonatology fellow. We created a key driver diagram summarizing the primary drivers and developing the possible interventions to improve the documentation rate (Fig 1).

Our quality improvement team included a neonatology fellow, a neonatology attending physician, an NNP, and a NICU charge nurse. We implemented the various interventions formulated through a series of PDSA cycles.

We used informal education tools to educate the providers and the nursing staff. At the beginning of the intervention period in September 2020, we sent emails to the nurses, fellows, and the NNPs explaining the problem, the importance of improving documentation rates, and the goals and the strategies to achieve these goals. The neonatology attending physician educated the pediatric residents in the beginning of every month starting September 2020 during their NICU orientation. In October 2020, the neonatology fellow part of this QI team further educated the nurses, residents, and NNPs through informal conversations.

We placed visual reminders to remind both the clinical providers as well as the nursing staff in October 2020. We placed “Have you checked red reflex on your baby?” stickers on all physician computers and “Discharging a baby? Pls check the H&P for red reflex documentation. If not documented, please inform resident, NNP, or Fellow on service” stickers on all the nursing computers.

During the first week after initiating this project, nurses were reminded by the charge nurses during daily huddle to check for red reflex documentation on their patients being discharged. Monthly emails were sent to the nurses updating them regarding the progress of the project and addressing concern areas.

To remind the providers to perform an RRE, we prepared a discharge checklist (Supplemental Fig 4). Starting February 2021, a copy of this checklist was displayed at the unit clerk’s desk, and another was displayed in the resident work room. Multiple copies were kept in the unit for the residents to use when discharging the patients. Although it was expected of residents to go through the checklist for every patient before discharge to avoid missing any tasks or tests that needed to be done, they were not expected to fill out the checklist for every patient unless their patient was planned to be discharged when they were not available.

A single handheld ophthalmoscope is available in our unit for red reflex examination. This handheld ophthalmoscope is kept in a drawer at the provider desk in the unit. Some residents shared that they could not find the ophthalmoscope or they forgot to perform RRE since the ophthalmoscope was not in sight. To remedy this, we requested a mobile ophthalmoscope unit that would be housed in the residents’ work room. Although this request was made in December 2020, we did not receive the mobile unit until May 2021 because of budgetary issues.

The above interventions were implemented over 4 PDSA cycles as shown in Fig 2.

The neonatology fellow initially performed a retrospective chart review for all the infants discharged from the NICU 6 months before the intervention period. From September 2020 to September 2021, the neonatology fellow and the NNP reviewed charts of all infants discharged from the NICU at the end of each month, after which the chart review was performed quarterly. Infants who did not receive a formal exam by an ophthalmologist before discharge were included in the evaluation. Chart review included a review of admission and discharge notes as well as progress notes 2 days after admission and 2 days before discharge. Process measures included the percent of admissions and discharges by residents or fellows and NNPs respectively, with or without red reflex documentation. Outcome measure was the total percentage of qualified infants who had red reflex documented before discharge.

We used run charts to show trends of red reflex documentation during the study period. The median was shifted when there were 6 or more consecutive values below or above the median. We also used the Minitab Statistical Software 2022 (State College, PA) to calculate the means and control limits for the P control chart for total proportion of infants with red reflex documentation. Mean was shifted when there were 6 or more consecutive values above or below the mean.

This was project was reviewed by the institutional review board and deemed exempt.

One thousand one hundred sixty-eight qualified infants were discharged during this project from our NICU; 314 infants were discharged during the baseline period from March 1, 2020 to August 31, 2020 and 854 infants during the 19-month intervention period from September 1, 2020 to March 31, 2022.

During the baseline period, 60 (19.1%) infants had red reflex documented before discharge. The percentage of infants with RRE documentation during these 6 months ranged from 12.7% to 29%. Documentation on admission each month was present in 11% to 25% infants, whereas that on discharge (when not documented on admission) was present in 2% to 16% infants. Residents consistently had a lower rate of documentation compared with NNPs on both admission and discharge.

During the intervention period, the documentation rate before discharge improved to 89.5% (765) infants. Only a modest improvement was seen after the first PDSA cycle (43%), whereas during the second PDSA cycle, the rate of documentation improved to an average of 79% over 4 months (range: 68% to 98%; median:75.5%); the highest rate during the second PDSA cycle was observed in October 2020, when the neonatology fellow part of this QI project was in service (Fig 3A). This was noted to be a common cause variation (Supplemental Fig 5). During the third and fourth PDSA cycles, the rates of red reflex documentation improved to and remained >90% (range 91% to 100%; median: 94.8%). (Fig 3A)

There was a significant improvement in the rate of red reflex documented at the time of both admission (Fig 3B and C) and discharge (Fig 3D and E) by residents as well as NNPs during the intervention period compared with the baseline period. Fellows in our program perform very few admissions (n = 134; 11.5%) and almost no discharges (n = 11; 0.9%). No improvement was seen in the documentation rate by fellows on admission.

Three infants had equivocal red reflex; Repeat examination of these infants by the neonatology fellow revealed a normal bilateral red reflex. One infant had an abnormal (white) red reflex, however, a formal examination by an ophthalmologist did not show any ocular abnormalities (false positive rate: 0.09%)

In our study, we demonstrated that by educating the providers and the nurses on the importance of RRE and implementing a discharge checklist including RRE, an increased rate of red reflex documentation was attained and sustained. Our PDSA cycles 1 and 2 were focused on education and visual reminders that improved red reflex documentation rates from an average 19.7% before intervention to 71.7% during the 5 months. The false positive rate in our study (0.09%) during the 25 months was lower than that reported by Cagini et al (2%) but higher than that reported by Eventov-Friedman et al (0.0006%).^4,6 

Discharge checklists can help improve discharge readiness and help with safe discharge of infants from the NICU.^7,8  However, not all NICU discharge checklists include red reflex or eye exam. By creating a discharge checklist for the residents and NNPs that included RRE, we were able to achieve our SMART aim of >80% documentation rates as well as improve and maintain our rates >90%.

Despite an overall low sensitivity for detecting ocular abnormalities, the newborn RRE is highly sensitive for the detection of anterior segment abnormalities such as congenital cataracts and retinoblastoma and has stood the test of time for screening of ocular abnormalities.^2,9,10  In a study by Haargaard et al, congenital cataract was detected much earlier in Sweden, where RRE was routinely performed before neonatal discharge, compared with in Denmark, where a pencil light was used for eye examination at 5 weeks of age.¹¹  Although we did not identify any cases of congenital cataracts or retinoblastoma with an increase in red reflex screening rate after the intervention, this is consistent with the reported incidence of these anterior segment abnormalities. Congenital cataract, 1 of the leading causes of treatable blindness worldwide has an estimated incidence of 2 in 10 000 births in the United States and a prevalence of 0.63 to 9.74 per 10 000 children worldwide.^12–14  Retinoblastoma, although rarer compared with congenital cataract, has an incidence of 1 per 15 000 to 20 000 live births, and is a potentially vision and life-threatening disease.¹⁵ 

There were many limitations to this project. A major limitation was the lack of balancing measures. We did not assess the examiner’s technique in our study. Ulanovsky et al noted increased rates of neonatal conjunctivitis with RRE.¹⁶  We did not measure the rates of conjunctivitis in this study. Additionally, as majority of our patients received RRE at the time of discharge, difference in the rates of conjunctivitis before and after the intervention period could not be compared. The study also had limited process measures. Although both NNPs and residents noted reminders by nurses being helpful, we did not track the number of RREs performed after nursing prompts. Also, the discharge checklist was not a part of the patient medical records, and the completion of discharge checklist was not mandatory, thus making it difficult to track the compliance with checklist use among the providers. Whereas we would have preferred to incorporate the discharge checklist into the electronic medical records, we were unable to do so because of system challenges. Similarly, we were unable to ascertain the reason for the difference in the outcomes with various stakeholders. Another limitation was the absence of a pediatric resident in the QI team. However, ours is a large size pediatric residency program, where all the residents rotate through the NICU for only a month every year. Hence, inclusion of a pediatric resident was not possible and would have not been very beneficial to the study.

There is an overall paucity of literature on red reflex examination rates in the United States and worldwide with essentially no data regarding red reflex documentation rates in the NICU. Unlike the well-baby nursery, many neonates admitted to the NICU are acutely ill on admission because RRE on admission might either not be possible or might not be considered a priority. Additionally, many premature infants receive formal ophthalmology exam in the NICU before discharge to screen for retinopathy of prematurity. It is possible that because of these reasons, providers in the NICU might forget to perform RRE for the other neonates before discharge. Education of the providers on the importance of RRE as well as incorporation of RRE as a part of discharge checklists can prove to be beneficial in improving RRE in infants discharged from the NICU. Although the American Academy of Pediatrics recommends routine red reflex examination for all infants before discharge from the newborn nursery, unlike the newborn nursery, infants in the NICU may remain hospitalized for a longer duration, sometimes months. Failure to perform RRE on admission in these infants may potentially delay the diagnosis of ocular abnormalities. Unfortunately, in our study, we were only able to improve RRE on admission from a mean of 14% before intervention to 34% postintervention.

RRE is an important part of routine newborn examination and can be easily performed on undilated eyes by nurses, nurse practitioners, residents, and physicians.¹⁷  Our QI project emphasizes the importance of RRE for infants discharged from the NICU and describes a simple, sustainable, and reproducible QI-based methodology for improving not only RRE documentation, but also other important components of newborn screening. Implementation of discharge checklists in the NICU should be considered to improve the care of patients discharged from the NICU and RRE must be included in the discharge checklist. Although no ocular abnormalities were noted during this project, the implementation of this project has led to a culture change that will help prevent us from missing the diagnosis of serious vision-threatening conditions in the future.