Improving parents’ knowledge of early signs of pediatric eye disease: A double-blind RCT

Study design

Using a pre- /post-test design, this double-blind, randomized controlled trial (RCT) compared participants’ responses to two clinical scenarios and statements of belief when faced with observing leukocoria or strabismus in a child. This study was approved by the Royal Women’s Hospital Human Research Ethics Committee (#17-38) and adhered to the tenets of the Declaration of Helsinki. Following study commencement, no significant changes were made to the study protocol.

Participant eligibility criteria and study setting

Participants were recruited from the antenatal outpatient clinic in a large metropolitan hospital in Victoria, Australia. Prima- or multiparous women aged over 18 years in their 2nd or 3rd trimester of pregnancy were invited to participate in the study. The study materials were only available in English.

Intervention and Control Arms

The intervention consisted of a 4-sided, A5 information pamphlet for parents describing normal developmental milestones for infant vision and ocular development; information about strabismus and leukocoria; and recommendations for seeking health advice if observed. The pamphlet as described in Supplementary Materials A, was specifically designed using an evidence-based approach to developing health promotion materials¹⁴ and grounded in a theoretical model for behaviour change.^{13, 14} The control group received an A4 double-sided information pamphlet for parents: “Playing with your baby”. This pamphlet encourages strategies for engagement with a newborn and is described in Supplementary Materials B.

Adherence to the intervention

Following informed consent, completion of the baseline survey and randomization (0-time), participants in both groups were verbally advised by the recruiting researcher SES to read the instructions and pamphlet enclosed in their allocated, sealed research envelope. Two-weeks after baseline (0-time + 2 weeks), participants were sent a secure link by email to complete the follow-up survey using the Research Electronic Data Capture tool (REDCap 7.2.2 © Vanderbilt University). Reminder emails were sent at 0-time + 4 weeks, and 0-time + 5 weeks and a final telephone call was made at 0-time + 6 weeks. Non-responders at the close of the study were recorded as being lost to follow-up. No inducements were offered although at enrolment, participants were advised that those who at the conclusion of the trial were found to have been randomized to the control group, would also receive the intervention.

Measures and Outcomes

At baseline, participants answered socio-demographic questions pertaining to age, previous experience with infant development, primary language spoken at home, educational attainment and professional background. Participant health literacy was measured by embedding the five questions of Domain 9 “Understanding Health Information” of the validated Health Literacy Questionnaire (HLQ).¹⁵

The evaluation survey comprised four- or five-point Likert scale items and single item True/False questions specifically developed for six clinical scenarios and 20 statements of knowledge or belief. To maintain masking of the participants to the specific focus of the study and aiming to minimize any additional search for information they would not otherwise be prompted to undertake after baseline not all questions related to infant vision or ocular health.

To evaluate participants’ knowledge and help-seeking intentions if they observed leukocoria or strabismus in their child, a single clinical scenario and statement of belief each pertaining to leukocoria and strabismus were embedded in the survey. Socio-demographic questions were excluded from the follow-up survey and replaced with questions examining participants’ pamphlet engagement. To account for any other sources of information that may contribute to their responses in the follow-up survey, questions relating to searching for more information were also included. (Supplementary materials C) Either baseline or follow-up surveys took less than 10 minutes to complete.

Outcomes

The primary outcome measure for this study was a change in the level of concern if the participant observed leukocoria or strabismus as measured on a 4-point Likert scale for a clinical scenario and a statement of belief. For example: “How concerned are you?” a) not at all concerned; b) somewhat concerned; c) concerned; d) very concerned; and in response to the statements: “I would be concerned if I saw a white pupil in a photograph of my child” or “I would be concerned if noticed my 2-year-old baby’s eye turning (cross-eyed) some of the time”: a) strongly agree; b) agree; c) disagree; d) strongly disagree. The secondary outcome measure was a change in help-seeking intention if the parent observed either leukocoria or strabismus as measured on a 5-point Likert scale. “How quickly would you seek advice?” a) today/tomorrow; b) within 1 week; c) within 1 month; d) longer than 1 month; e) not at all. We hypothesized that women receiving the intervention would demonstrate an increase in concern if leukocoria or strabismus were observed and demonstrate a change in their help-seeking intention.

Sample size and power calculation

Using STATA (Release 14, 2015),¹⁶ a sample size of 398 participants (199 in each group) was calculated to confer 80% power to demonstrate a 10% effect with a two-sided p=0.05 as we could not be certain of the direction of the intervention effect. Allowing for a 30% attrition rate, the study aimed to recruit 520 participants. All analyses were performed on an “intention-to-treat (ITT)” method, regardless of whether those in the intervention group read the pamphlet or not. Statistical analyses were computed using R, Version 3.5.0¹⁷

Masking, randomisation and implementation

Sealed, opaque research envelopes labelled either Group A or B were prepared by two research assistants from the Centre for Eye Research Australia familiar with the conduct of clinical trials. The control pamphlet was folded in half to ensure the research envelopes in both groups were equal in weight and composition. Group allocation of the intervention was determined by the research assistants preparing the envelopes.

A concealed, randomization allocation sequence generation table for two groups, Group A and B, with a 1:1 allocation [2,4,6, block design] was prepared by the statistician (PGS) using R, Version 3.5.0¹⁷ and uploaded to REDCap. Using the REDCap randomization module, upon completion of their baseline survey each participant was automatically randomized to either Group A or B and given their corresponding research envelope by the recruiting researcher (SES). To SES remained masked to group allocation, participants were instructed to not open their research envelope until after leaving the hospital grounds following their antenatal appointment.

All research participants and researchers directly associated with the trial remained masked to participant group allocation until the conclusion of the study and data were analyzed. The study was described to participants as one of maternal and child health, without disclosing the specific focus of the study or the intervention being tested.

Statistical analysis

Differences among survey responses across treatment groups and follow-up periods were investigated using ordinal logistic mixed model regression analysis. Potentially confounding variables including age, English as a second language at home, previous experience with children, educational attainment, health literacy and past occupational experience in healthcare were adjusted for in the analyses. Specifically, for each outcome measure a ‘Group x Time’ interaction term was assessed and retained in the model if statistically significant, thus indicating a differential response rate across treatment groups at the follow-up compared to baseline visit. Unless otherwise specified, all statistical tests were two-sided, where p ≤0.05 was considered statistically significant.

All participant demographic variables were used to screen for factors that could moderate the outcome of the intervention. Comparison between prima- and multiparous women served as a surrogate for ‘usual care’, providing insights into information parents may have already received or been exposed to in the course of raising their children.

Baseline and follow-up surveys were self-administered using REDCap. This secure, password-protected, web-based data management system is hosted by The University of Melbourne under licence from Vanderbilt University. Each participant’s record was allocated a unique study identifier and data de-identified for statistical analysis.

Participant recruitment and study flow

Figure 1 outlines participant recruitment during the study period of 8 March 2018 to 24 June 2018. Of the 692 invited to participate, 518 consented, completed the baseline survey and were randomized to either the intervention or the control group, representing a 74.9% participation rate. The 20 women who ‘failed to proceed’ initially consented but were called to their antenatal appointment prior to completing the baseline survey and did not wish to continue following their appointment. Six participants formally withdrew following randomization and were excluded from the analysis.

• Download figure
• Open in new tab

Figure 1:

Overview of the study and participant recruitment. Details are displayed as per the CONSORT statement.²⁸

Despite receiving reminders as outlined in the protocol, 382/518 women completed the follow-up survey representing a 73.7% overall response rate. Some women attended their antenatal appointment with their partner. When both parents were present, the pregnant woman provided informed consent and completed the survey; however, the partner was neither encouraged nor discouraged from contributing to the decision-making as this could not be controlled for when the participant completed the follow-up survey at home.

Participant demographics

Generally, participants were tertiary-educated with high health literacy. More than two-thirds of the cohort did not have any other children and few or none had had any previous experience with leukocoria or strabismus. There was no statistically significant difference in demographic characteristics between participants allocated to either the intervention or control group. (Table 1).

Primary outcomes

Primary outcomes for both leukocoria and strabismus are described in Table 2. At follow-up, participants randomized to the intervention group were more likely to report a higher level of concern when faced with the clinical scenario describing leukocoria in a photograph of their child. (OR 1.711, [CI 1.176,2.497] p=0.005) For all participants, those who searched for more information about leukocoria had an almost 3-fold higher odds of being more concerned (OR 2.778, [CI 1.743,4.455] p<0.001). Participants in the intervention group were less likely to disagree with the statement: “I would be concerned if I saw a white pupil in a photograph in my baby” (OR 0.407, [CI 0.267,0.618] p<0.001). Searching for more information about strabismus was associated with a 62% decrease in the odds of disagreeing with this statement (OR 0.379, [CI 0.211,0.672] p=0.001) Moreover, participants referring to their pamphlet when completing the follow-up survey were less likely to disagree. (OR 0.541, [CI 0.309,0.941] p=0.030)

For the clinical scenario describing strabismus in a child, there was no statistically significant difference in the level of concern between the intervention and control groups at follow-up. However, participants who looked up more information about strabismus at follow-up had a 2-fold increase in the odds of being more concerned about this sign. (OR 2.009, [CI 1.128,3.579] p=0.018). Moreover, participants referring to their pamphlet to complete the follow-up survey were less likely to disagree with the statement: “I would be concerned if I saw my 2-year-old baby’s eye turning (cross-eyed) some of the time” (OR 0.400, [CI 0.225,0.707] p=0.002).

Secondary outcomes

As shown in Table 3, responding to the clinical scenario for leukocoria, participants in the intervention group were less likely to delay seeking health advice (OR 0.560, [CI 0.382,0.817] p=0.003), as were participants who searched for more information on strabismus (OR 0.427, [CI 0.248,0.725] p=0.002). Responding to the clinical scenario for strabismus, for all participants there was a 2.7-fold increased odds of seeking advice sooner at follow-up (OR 2.744, [CI 1.384,5.442] p=0.004). Whilst group allocation in isolation did not have a predictive effect on help-seeking intention for the strabismus scenario, an interaction effect between follow-up visit and group allocation was observed, whereby at follow-up, participants who received the intervention were less likely to delay seeking advice (OR 0.318, [CI 0.125,0.806] p=0.016). Moreover, participants who spoke a language other than English at home were more likely to seek advice promptly (OR 0.218 [CI0.098,0.482] p<0.001).

Limitations

The single-centre study design may have unintentionally biased the outcomes. Although participants’ demographic characteristics were equally distributed between the two groups and were adjusted for confounding variables, their homogeneity may have contributed to the positive outcomes observed in this study. Strong literacy skills and command of the English language may have enabled them to engage more fully with, and understand the contents of, the intervention.

In this study, fathers or partners contributing to the RCT could neither be prevented nor controlled. Thus, it is not known whether a difference of opinion may have tempered or intensified their concern or help-seeking intention. Other than for mental health and behaviour²⁵, few studies have explored a father’s (or partner’s) role in child health and help-seeking compared to the mother’s.²⁶ Although no significant differences between mothers’ and fathers’ knowledge and attitudes towards seeking eye-health care for their children have been identified²⁷, the distinction between maternal and paternal help-seeking behaviour is not usually made, referring only to ‘parental’ behaviour.^28–30 Future studies could consider comparing responses between both parents to determine whether there are any differences, as these would need to be addressed to optimize health promotion initiatives.

The time between receipt of their pamphlet and the follow-up survey was short (2 weeks). Thus, participants may have had a better recall or retention of information than would normally occur. Nonetheless, this study did establish that providing information that can be understood and interpreted by parents can improve knowledge and help-seeking behaviour. This study could be extended in the future by repeating the follow-up survey at a later time-point to better understand parent recall of information and effect on help-seeking intention.