Prevalence of Severe Visual Disability Among Preterm Children With Retinopathy of Prematurity and Association With Adherence to Best Practice Guidelines

IMPORTANCE Retinopathy of prematurity (ROP) can cause severe visual disability even in highresource settings. A better understanding of the prevalence and processes leading to ROP-induced severe visual impairment may help health care professionals design preventive measures. OBJECTIVES To determine the prevalence of severe visual disability among children born preterm in Sweden, evaluate adherence to best practice, and determine the health system’s structural capacity. DESIGN, SETTING, AND PARTICIPANTS Population-based, nationwide cohort study of 1 310 227 children born between January 1, 2004, and December 31, 2015, in Sweden, of whom 17 588 (1.3%) were born very preterm (<32 weeks of gestation). Children born preterm with a verified diagnosis of severe visual disability had their medical records reviewed for evaluation of ROP screening, diagnosis, and treatment. In addition, a questionnaire on structural capacity was sent to all ophthalmology departments. EXPOSURES Stages 4 and 5 ROP. MAIN OUTCOMES AND MEASURES The primary outcome was prevalence of severe visual disability (visual acuity 20/200 for both eyes) associated with ROP stages 4 and 5. Secondary outcomes included adherence to national ROP guidelines using a predefined protocol with 15 key performance indicators for screening, diagnosis, and treatment; assessment of whether visual disability was deemed avoidable; and examination of structural capacity, including information on equipment and facilities, staffing, and patients. RESULTS Seventeen children (10 boys; mean [range] birth weight, 756 [454-1900] g; mean [range] gestational age, 25 [22-33] weeks) became severely visually disabled because of ROP, corresponding to a prevalence of 1 in 1000 very preterm infants (<32 weeks of gestational age) and 1 in 77 000 for all live births. Severe visual impairment was considered potentially avoidable in 11 of 17 affected children (65%) owing to untimely or no screening, missed diagnosis, or untimely and suboptimal treatment. Large variations in infrastructure (facilities, guidelines, staffing, and annual patient numbers) were also identified as potential contributors to these findings. CONCLUSIONS AND RELEVANCE Retinopathy of prematurity still causes severe visual disability in Sweden, resulting in 1 affected infant per 1000 very preterm births. In most of these infants, (continued) Key Points Question What is the prevalence of visually disabling retinopathy of prematurity in Sweden and how is it associated with quality of care? Findings In this population-based cohort study of 1 310 227 infants born alive in Sweden in 2004 to 2015, 1 in 77 000 or 1 in 1000 very preterm infants (<32 weeks of gestational age) became severely visually disabled by retinopathy of prematurity. In 11 of 17 infants (65%), disability was considered avoidable because of nonadherence to best practice, as well as flaws in structural capacity. Meaning Severe visual impairment associated with retinopathy of prematurity is preventable in most cases. Author affiliations and article information are listed at the end of this article. Open Access. This is an open access article distributed under the terms of the CC-BY License. JAMA Network Open. 2019;2(1):e186801. doi:10.1001/jamanetworkopen.2018.6801 (Reprinted) January 4, 2019 1/12 Downloaded From: https://jamanetwork.com/ by a Uppsala University User on 05/23/2019 Abstract (continued)continued) noncompliance with best practice was identified, indicating that a significant proportion could have been avoided. JAMA Network Open. 2019;2(1):e186801. doi:10.1001/jamanetworkopen.2018.6801


Introduction
Retinopathy of prematurity (ROP) is a major cause of lifelong visual disability among children born preterm, particularly in resource-poor settings. Globally, an estimated 20 000 infants become severely visually disabled from ROP each year, and ROP has become an increasing cause of blindness in China, Southeast and South Asia, Latin America, and parts of Eastern Europe. 1 Given effective tools for screening, diagnosis, and treatment, severe disability due to ROP should be possible to avoid in most infants. However, disabling ROP still occurs even in high-resource settings. 2,3 Therefore, a better understanding of the structures and processes leading to disabling ROP is needed.
In regions defined as developed by the World Health Organization Millennium Development Goals, an estimated 1 200 000 babies are born preterm each year. 4 Among the 150 000 believed to be at risk for developing ROP, 1700 are estimated to become severely visually disabled, giving a prevalence of disabling ROP in high-resource settings of 1 in 90 to 1 in 100 very preterm (<32 weeks of gestation) births. 1 However, there is only limited information on disabling ROP from populationbased studies. Although the prevalence of severe ROP may vary in relation to case mix and to varying approaches to oxygen saturation targets, monitoring, and nutrition policies, [5][6][7] less is known about the progression to disabling ROP. In a previous Swedish study of extremely preterm infants, 20% of those with an indication for ROP treatment were treated in an untimely manner, indicating that nonadherence to best practice may play a role. 8 We know of no previous studies on severely disabling ROP combining population-based epidemiology to establish the true prevalence with analysis of adherence to best practice and underlying structural capacity. In 2008, national guidelines for screening, diagnosis, and treatment of ROP were issued based on population studies, [9][10][11] including the Swedish National Register for Retinopathy of Prematurity (SWEDROP). 12 Using these and other sources of registry and database information, the aim of this study was to determine the prevalence of severely disabling ROP in a population of more than 1 000 000 children, determine adherence to best practice in infancy, and decide to what extent severe visual disability could be considered avoidable. We also aimed to capture information from all ophthalmology departments in Sweden regarding their present structural capacity.
Methods criterion (n = 0 in our search output). Children with visual disability residing in Sweden but born outside the country were also excluded (n = 1).

Data Sources
Open-access statistics provided by the National Board of Health and Welfare were used to assess the total number of preterm births in the study period. We used SWEDROP 12

Medical Record Review
Personal identification numbers retrieved from the registers and databases were used to access medical records at hospital databases and county archives. Records from both the neonatal units and the ophthalmology departments were scrutinized according a predefined protocol by 5 of us (M.N., A.H., B.H., A.W., and G.H.). First, misclassifications of ROP, gestational age, or date of birth were checked for. Second, we reviewed the medical records for present visual status and included only those with severe visual disability (n = 17). Third, we applied the exclusion criteria (n = 1 born outside Sweden). The flowchart for inclusion can be seen in the Figure. Among children with visual disability, we could not get access to medical records in 3 cases. These participants were evaluated using available registry and database information.
Using national 17 and international 18-21 guidelines, we identified 15 key performance indicators: 5 related to documentation, 5 related to ROP screening, and 5 related to ROP diagnosis and treatment ( Table 1). Based on these indicators, we established a predefined protocol of best practice in management of ROP and applied this protocol to our medical record review. Although not outlined in our guideline, we considered fewer than 1500 laser photocoagulation effects per eye as an indication of suboptimal density to prevent disease progression at first treatment. 19 The World Health Organization defines patient safety as "the absence of preventable harm to a patient during the process of health care and reduction of risk of unnecessary harm associated with health care to an acceptable minimum. An acceptable minimum refers to the collective notions of given current knowledge, resources available, and the context in which care was delivered." 22 Based on compliance with best practice and using the classification of the National Board of Health and Welfare in Sweden, 23 we characterized cases of disabling ROP as unavoidable, possibly unavoidable, possibly avoidable (estimated probability Ն51%) or avoidable (estimated probability Ն75%). The

Questionnaire to Ophthalmology Departments
To evaluate current health care systems' capacity, a questionnaire was sent to all ophthalmology departments (n = 35) in Sweden in 2017, including 2 reminders. The questionnaire contained 26 questions relating to infrastructure, facilities, equipment, staffing, and education, as well as number of patients and ROP examinations and treatments.

Statistical Analysis
Data are presented as number and proportions (percentages), mean (SD), or median (range) values.
Calculations were performed using Microsoft Excel 2013 software.

Perinatal Characteristics
Perinatal characteristics and selected information on ROP stage and treatment in children with severe disability due to ROP are presented in Table 2. Most perinatal characteristics were similar to those reported in a previous population-based cohort of infants born before 27 weeks of gestation in Sweden, 15 ie, the infants who became visually disabled did not differ from the general extremely preterm population except that almost all (94%) had bronchopulmonary dysplasia. One child with visual disability was born at 33 weeks' gestation with severe lung hypoplasia secondary to a large diaphragmatic hernia, illustrating that ROP requiring treatment may also occur in more mature infants with exposure to high fraction of inspired oxygen for a prolonged period (in this case >3 months).

Adherence to Best Practice
Auditing the medical records of the 17 children who developed severe visual disability, we found several flaws in documentation, screening, and treatment. Poor documentation in neonatal records, postponing screening examinations, and delayed screening after transfer between hospitals were of particular concern. Untimely and suboptimal treatment also occurred, and follow-up after first treatment was delayed in 1 of 3 infants (Table 1).

Avoidable Visual Disability
We identified 6 possibly avoidable and 5 avoidable injuries of severe visual impairment due to ROP.
In 1 infant, both screening and treatment failed, meaning that severe visual disability in 11 of 17 children (65%) most likely could have been avoided had best practice care been provided. Examples of avoidable visual disability included situations in which screening was never performed or had been interrupted for up to 6 weeks, retinal detachment was not recognized in spite of weekly examinations for a month after detachment had occurred, and treatment was delayed or clearly insufficient ( Table 3). Laser coagulation was used as first-line treatment in 16 of 17 infants (all but 1 infant who was 33 weeks gestational age and was never screened, in whom retinal detachment was discovered at 4.5 months of postnatal age). Additional treatment included encircling band (n = 6), vitrectomy (n = 3), and anti-vascular endothelial growth factor (n = 2).

Financial Compensation
During the study period, 3 of 11 families (27%) affected by possibly avoidable or avoidable visual disability due to ROP applied for financial compensation from the Swedish National Patient Insurance. Two families who had applied for financial compensation were subsequently granted it.

Capacity of the Health Care System
In total, 25 of 35 ophthalmology departments responded to our questionnaire, including all 6 university centers in Sweden. Several of the nonresponders were smaller hospitals. One responder from a small hospital reported that they had no very preterm patients in their hospital, and responses from this hospital were therefore excluded from further analysis.    Fourth, our survey disclosed large variations in national infrastructure regarding screening, diagnosis, and treatment for ROP, which may have contributed to substandard care and adverse outcomes.
The prevalence of severe visual disability secondary to ROP was one-tenth of that recently estimated from meta-analyses pooling information from high-income countries. 1 This discrepancy was surprising given the comparatively high survival rate among extremely preterm infants, the high incidence of ROP, and high prevalence of later visual problems found in this group of Swedish children. 2,7 Other European population-based studies of children surviving extremely preterm birth have reported similar results as in our study, 24,25 suggesting that the prevalence of disabling ROP in high-resource settings may be overestimated in global surveys. 1 Despite national guidelines that have been continuously improved and modified national guidelines since 1993, [9][10][11]17 there were 17 children with end-stage ROP causing severe visual impairment, with no decline in prevalence over time. With few exceptions, the children with visual disability exhibited well-known risk factors for severe ROP, such as birth after extremely short gestation (22-26 weeks) and neonatal exposure to supplemental oxygen for prolonged periods, underlining a need for meticulous screening and proactive treatment. 8 Only 2 children had neonatal brain injury (intraventricular hemorrhage grade 3 or cystic periventricular leukomalacia), and in both cases we considered the severe visual disability to be mainly associated with ROP, as these infants had retinal detachment (ROP stage 4-5). As shown in EXPRESS, 2,3 the total prevalence of severe visual disability among extremely preterm survivors would be approximately twice that reported herein, if adding severe visual disability secondary to comorbidities of prematurity (such as severe neonatal brain injury) to ROP.
The structured medical record review performed by 3 experienced pediatric ophthalmologists and 2 neonatologists revealed deviations from national recommendations regarding both ROP screening and treatment procedures. Successful ROP management requires adherence to best practice, including timely start and frequency of the screening examinations. This requires a close collaboration between the neonatologist and the ophthalmologist and is particularly important to consider when transferring the infant to other hospitals. In children with severe disability due to ROP, we found that ROP screenings and treatments had been poorly documented in neonatal records, and that only 1 in 3 neonatal records contained specific information on ROP when the patients had been transferred between hospitals. Delayed screening and lack of screening were associated with severe visual disability in 3 children.
The most common deviation from best practice that was associated with potentially avoidable visual disability in our study was untimely or suboptimal ROP treatment. Once criteria for treatment have been fulfilled, treatment must be undertaken in a timely manner, ie, within 72 hours. This requires not only excellent collaboration between ophthalmologists and neonatologists, but also with anesthetists, as laser treatment is generally performed under general anesthesia. In addition, only 1 of 4 departments reported that ROP treatment could be provided in the facility, a challenging structural emergency for transport organizations, hospitals, and administrations, as well as for families. Furthermore, and very importantly, treatment has to be adequate, with an appropriately placed and sufficiently large number of laser applications, covering the avascular, peripheral retinal area. Finally, follow-up after treatment is crucial to identify infants with so-called skip areas in need of The strengths of this study include the large population base and the use of personal identification numbers to link information from several registers and databases and validate this information against medical records. We included all preterm infants born in Sweden over more than a decade and collected information on several neonatal diagnoses, enabling us to present truly population-based prevalence of severe visual disability due to ROP as distinct from severe visual disability caused by other comorbidities. Including not only ROP stage but also final visual ability, we critically evaluated practice using a predefined protocol based on national guidelines in all children with the most unfavorable outcome, ie, severe visual impairment. Combining registry data and medical record reviews with a contemporary department survey, we were able to identify several potential areas of improvement in clinical practice as well as in the structure of the health system.

Limitations
Our study also has some limitations. It is retrospective in design and therefore the registers as well as medical records may have lacked some information. We had no access to photographs to verify adequacy of laser treatment, and some centers may have used lasers with a larger spot size, requiring fewer spots. In addition, the number of delivered spots may have been higher than those that were actually focused and became spots. Therefore, our limit of fewer than 1500 spots as an indication of suboptimal density to prevent disease progression at first treatment could be insufficient, contributing to our more cautious conclusion that flaws in treatment were possibly avoidable. Our structured medical record reviews were limited to worst-case scenarios and say little if anything about clinical performance regarding ROP in general. The survey on infrastructure and capacity was self-reported, which could introduce bias. In Sweden, there is no regular ophthalmological follow-up of all children born very preterm, only for those who have been treated for ROP or had major neonatal brain injury.

Conclusions
Despite national guidelines for ROP screening and treatment, children born preterm still become severely visually impaired in Sweden. Although the prevalence is low in an international comparison, severe visual disability could have been avoided in more than half of the children had there been better adherence to best practice, a better structural capacity, and higher competence among ophthalmologists and neonatologists. We request that screening and treatment for ROP be prioritized by departments and hospital administrations, including continuous education and training of the ophthalmologists screening and treating these infants. To avoid lifelong visual disability in preterm infants, equal care within different parts of a country should be sought, and centralization of the most difficult cases and treatment may be a way forward.