This protocol is structured in accordance with the Standard Protocol Items: Recommendations for Intervention Trials (SPIRIT) 2013 Checklist.[43] (see Supl 1) .
Trial Design and overview
This trial is a single masked, parallel group, restricted cluster-randomised controlled trial. Thirty-six community units with their health facilities (dispensary or health centres) will be randomly selected to receive either the intervention (Community screening using Peek Screening system) or the current standard of care (periodic health centre-based outreach clinics). The health workers involved in the study will be trained to ensure standardised screening. Participants who provide consent will be enrolled to the arm that their cluster is randomised to.
In the Peek arm, all households in the cluster will be visited in turn. Consenting individuals will have their visual acuity tested using the Peek visual acuity screening application on a smartphone. All participants with reduced visual acuity or reporting another eye problem will be referred to the linked Primary Health Centre (PHC) for assessment and management. Those requiring treatment not available from the PHC facility will be referred onwards to Kitale Eye Unit (KEU). In the control arm, communities will be notified about the periodic eye health outreach clinic that will be held in the local health centre. People attending this service will be assessed and if necessary referred onwards to Kitale eye unit.
The participants will be followed up for 8 weeks after referral from the community. The Primary Outcome will be the number of people per 10,000 population (rate) attending Triage at a local health facilities (PHC) with any confirmed eye conditions (true positive cases determined at triage by hospital outreach team) following a referral or by self-referral, within four weeks from the time of sensitization. The Secondary Outcome will be the proportion of people referred from the PHC triage attending their referrals at KEU within four weeks of being referred. A participant (standard or Peek) who attends the hospital appointment within four weeks will be considered to be an “attender” while anyone who is referred but does not attend within the same time is a “non-attender”.
Participant timeline and study flowchart
The study flowchart and participant timeline are presented in Figure 1 and table 1, respectively.
Participants, interventions and outcomes
Study Setting
The trial will be conducted in community units that are served by government-run dispensaries and health centres in Trans Nzoia County in northern Kenya. Trans Nzoia County has a population of 818,757 people (2009 census) of which 407,172 (49.7%) were male.[44] It is organised into five sub-counties. There were 173,719 households, with an average of five people per household. The large majority have no internet access (669,347 [81.8%]).[45] There are 61 government facilities (6 hospitals, 12 health centres, 43 dispensaries) and 76 facilities owned privately or by faith based organisations.[46] Eye services are offered at KEU and through outreach services, provided by eye care staff from KEU to other health facilities. Screening and treatment of eye conditions (Triage) is offered during outreach. The trial will be coordinated from Kitale Hospital by a team consisting of a programme manager, administrator, and ophthalmic nurses, field workers and an ophthalmologist.
Cluster definition
The unit of randomisation for this trial will be Community Units (CU). These are defined as a dispensary or health centre together with the community they serve (Figure 2). A typical CU comprises a population of 5,000 to 10,000 people. It has a dispensary or health centre, staffed by one or two Community Health Extension Workers (CHEWs). Associated with each CU there are usually 20 to 50 Community Volunteers (CVs).[47] The CHEWs based at the health centre or dispensary train, support and supervise the CVs. To date 85 CUs have been established and personnel trained in this county.[46] CUs were chosen because it represents the future shape of health care in Kenya, they are distributed throughout the county and have a good referral network that provides linkages between community and health system. The CUs with untrained personnel provide a buffer zone that will minimize contamination.
Cluster eligibility criteria
A list of all health facilities with their geo-coordinates, and with corresponding community units and catchment population will be obtained from the Trans Nzoia County Department of Health. The location of each hospital will be determined using Google Maps. Health facilities without community units, those with existing screening programs and the communities directly served by KEU will be excluded. We will also exclude all the non-government health facility associated CUs. From the remaining 66 community units, a total of 36 units will be randomly selected for the study. A restricted cluster random sampling technique (described below) will be applied to allocate the selected CUs to the Peek intervention (18 CUs) or the standard care group (18 CUs). The restriction will be based on the location of the CU’s heath facility relative to the Kitale Eye Unit.
Participant eligibility criteria
All people irrespective of age, present in the community unit area during the study period who consent to participate will be included. People who are unwilling to give consent or who have had an eye condition treated at hospital within two weeks prior to the study beginning will be excluded.
Interventions
A comparison of the two arms is shown in Table 2 and Figure 1. Prior to the commencing of the trial, each of the clusters in both arms will be visited by the field team to explain the study, consent and enumerate the residents. At the beginning of the trial, in both there will be posters and verbal notices (churches and schools) advertising the forthcoming outreach clinic for eye checks, encouraging people with eye problems to self-report to the clinic on a specific date when the team will visit.
Peek CEH intervention arm
In each cluster a small mobile team of a “Peek User” (Community Volunteers (CVs) trained specifically on how to use Peek Community Screening App and who travel to multiple communities to perform their duties) and local CV will visit each household. The CV, a person from that same community, will guide the Peek User around the village. After reconfirming consent, people who are resident in the household at the time of the visit will have a vision assessment. The visual acuity of each eye will be measured separately using the Peek Acuity app.[38] This smartphone application presents a series of E-optotypes in one of four orientations, selected at random. The test algorithm prompts the following screening questions to the parents or guardian with a child (“Does the child have any problem with their eyes today?”) or directly to participant themselves, (“Do you have any discomfort or pain in your eyes today?” and “Do you have a problem with your sight when seeing far or near objects?”). If the participant is six years or older, the app prompts for distant visuals acuity assessment using Peek Acuity app and assessment of near visual acuity for all people aged 40 years and older. Near vision will be assessed at 40 cm using the RADNER reading chart. [48] If the visual acuity is less than 6/12 in either eye; or presence of any self-reported eye pain or discomfort; difficulty seeing distant or near objects; or not able see N8 on near vision assessment, they will be referred to primary health centre for subsequent assessment by the visiting team. Household members absent during the initial visit will be asked to join the examination team at the next household or next day.
Those who have reduced visual acuity on screening or report an eye problem will be referred to a health posts for triage on a specific date when the KEU team visit. The system will generate several SMS text messages: (1) to the patient and family associate asking them to present to the health facility on a specific day (set to be within four weeks); (2) CV will receive an SMS list of patients from their community that have been referred; (3) CHEW responsible for that CU will similarly receive the same list of referred patients. A weekly reminder SMS will be sent to the patient for them to attend their referral appointment with the last reminder being one day to the appointment.
On the pre-advertised date a team from Kitale Eye Unit (KEU) will be based at the CU’s dispensary. The participants referred from the household screening because of reduced vision or a specific eye problem will be reminded to attend. They will assess the presenting patients using the current standard procedure (Snellen chart visual acuity, magnifying loop, refraction and direct ophthalmoscopy when indicated). They will provide simple treatments or refer patients to KEU for further assessment as indicated. A pre-numbered paper referral letter will be given to the patient to present at KEU. The referral slip has their study number, name and triage centre, telephone number, and indicates that assessment and treatment will be provided at no cost. It is expected that they will report to Kitale within 4 weeks from being referred
Immediately after referral from the PHC, a SMS will be sent to the patient and the family associate asking them to present to KEU. A weekly reminder SMS will be sent for those who have not attended their referral to KEU. SMS with a list of patients who have not attended their referral will be sent to the CHEW responsible for the PHC.
Standard of Care (Control) arm
In the control arm there will be no active peek screening in the community, however potential participants with eye problems at the community will be notified through community sensitization (posters and local announcements) that if they have an eye problem, to present themselves to health facility for the triage clinic on a specified date. On that advertised date the team from KEU will conduct an outreach clinic within the CU, which will be identical to the ones in the Peek arm described above. If an individual needs to be referred to KEU they will be given an identical referral letter to the ones used in the Peek arm. Each letter will have a unique code number to link the patient referral record to their KEU attendance.
Outcomes
Primary Outcome: The number of people per 10,000 population (rate) attending Triage at a local health facility (PHC) with any confirmed eye conditions (true positive) following a CV referral or by self-referral, within four weeks from the time of sensitization. The rate will be based on baseline enumeration census for each CU. The true positives will be determined at triage by hospital outreach team.
Secondary Outcome: The proportion of participants referred from the PHC who attend the referral at KEU within four weeks of being referred from a PHC.
Other outcomes: (1) The time taken by a participant referred from PHC to attend KEU. (2) The number of people per 10,000 who attend triage post without any eye condition (false positives) as determined by eye team.
Sample size
The sample size of 36 clusters was determined using the Hayes formula for rates in unmatched cluster-randomised trials.[49] In Trans Nzoia County, a typical health facility has a catchment population of 5,000 people,[46] and about 15,000 new patients are seen per year in Kitale at eye unit. With a total catchment population of one million people, this translates to an annual rate of 15 per 1000 population. Assuming a cluster coefficient of variation of 0.25, desired power of 90% and significance level of 5%, a sample of 36 community units (18 in each arm) would be sufficient to detect a difference of 30 % (1.5 % vs. 2.0 %) in overall attendance rates.
Assignment of interventions
Allocation
There are 66 potentially eligible community units in the county (see above). We will select 36 for inclusion in the trial. In order to ensure balance between the arms, restricted randomisation will be used. A list of the 66 community units with their sub-county, distance from Kitale and direction from Kitale (categorised into four quadrants, North, South, East and West) will be compiled and used during randomisation. A statistician, who will not participate in recruitment, will generate a random allocation sequence. Randomisation will consider the direction, cluster size and distance from the hospital. The following restrictions will be used in the randomisation:
- Each arm must include at least two community units from each sub-county
- Each arm must include at least two community units from each direction of North, South, East and West
- The ratio between number of community units in each arm from each direction must be in range 0.67-1.5.
- The difference in mean health centre distance from Kitale in each of the arms should not be greater than 4km.
- Not more than 1 community unit per link health facility
A list of 10,000 valid permutations will be generated (and checked that there are no clear deviations in randomness (e.g. pairs of health centres that occur within the same arm considerably more/less often than would be expected by chance). One of these 10,000 permutations will be computer-selected at random. A list of community units allocated to the control, intervention and those not involved will be prepared.
In health facilities where there are larger catchment populations and served by more than one community unit, one of the community units will be randomly selected along with its population unit, so that the size of the clusters studied is around 5000.
Masking
It will not be possible to mask the participants or the health workers from the intervention they are allocated to, however the study statistician, hospital registration clerk and clinician assessing outcomes will be masked. The data clerk will be masked to the intervention arm because all the patients will present with paper referral. The clinician assessing secondary outcomes will not participate in patient recruitment or assessing attendance and all patients will be given similar assessment questionnaires. The statistician will not participate in patient recruitment.
Data collection, management and analysis
Data Collection
In both arms, we will use electronic data capture and management using dedicated Peek software with built-in consistency checks. In both arms this will include the enumeration data, the triage data in the health centre/dispensary and the outcome data collected in the KEU. In addition, the household screening data will also be captured electronically for the Peek arm during the study period and in the control arm following the study when the team will screen all the control clusters. Field workers will be provided with tablets for data entry. Information will be backed up regularly.
During triage assessment at the health centre / dispensary, trained field workers will verify that the participant comes from the catchment population. From each eligible participant; date of attendance, name, age, gender, and own or parents’ mobile phone number, whether referred using the peek system or self-referral, the diagnosis and treatment plan (treated or referred). At Kitale eye unit all referred patient will be marked as attended upon presentation and record the date of visit, diagnosis and outcome of the visit.
Data management
Data will be entered directly onto smartphones by trained field workers and uploaded to a secure server once connected to the Internet before being exported into Stata for analysis. The database will be encrypted and password protected. At the end of the study, the data will be archived at LSHTM.
Data Analyses
The trial will be reported using the 2010 CONSORT guidelines, with the cluster RCT extension.[50] Analysis will be by intention to treat. Socio-demographic characteristics of participants at baseline will be tabulated by arm; age, sex, residence and distances from hospitals (categorised distances). The distributions of these variables by intervention arm will be compared, to assess whether there is imbalance at baseline in these potential confounding factors.
Analysis of the Primary outcome:
Attendance rates will be calculated for the intervention, sex, and age group. The population size will be determined by the baseline enumeration census in both arms. A t-test will be performed on the cluster-level rates providing an estimate of the rate difference (with a confidence interval) between the two arms. [51] We expect the two study arms to be balanced, however if not, an appropriately adjusted linear regression will be performed, with attendance rate as outcome and study arm as the exposure.
Analysis of Secondary outcomes:
The proportion of patients referred from the PHC to the KEU who attend their referral within four weeks, by arm, will be tested by logistic regression. It is expected that the important baseline characteristics of the people who present to the PHC will be balanced between the two arms by stratified randomisation. If they are balanced, an unadjusted logistic regression model with 95% confidence interval will be used. If the arms are found to be imbalanced, an appropriately adjusted logistic regression model will be developed.
We will assess effect modification of the intervention on referral attendance by sex, age and distance from health facility to Kitale, by including an interaction term between the intervention and each of these variables. In addition to this, these three factors will be used as exposures (alongside intervention) in a multivariable logistic regression model to identify if any are potential explanatory factors for attendance.
The Impact of the intervention on time-to-attendance will be investigated by estimating the mean time-to-attendance of the referral. The mean and the 95% confidence interval will be calculated in each arm for comparison. Kaplan-Meier analysis will be used to plot the survival curves for both arms to compare attendance of referral. The hazard ratio will be estimated, Cox regression will be used to assess the impact of intervention on referral.
The number of patients attending with no eye problems (false positives) at the primary health facilities during the study period will be analysed using standardised attendance rates. We will perform a t-test to estimate the mean difference in attendance by study arm.
Monitoring
Data monitoring
The study presents minimal risk and we do not anticipate significant adverse events. Therefore, a data and safety monitoring committee was not considered to be required, however an audit will be done by London School of Hygiene and Tropical Medicine (LSHTM), the Trial Sponsor if deemed necessary. No interim analysis is planned due to the relatively short duration of the study.
Harm
The tests being done are in routine clinical use in Kenya and internationally. There are no anticipated harms from this non-invasive assessment process in either arm. Assessment in the community will take 5 minutes per person. Experienced Certified Ophthalmic clinical officers will provide treatment for all participants with eye problems, under the supervision of an ophthalmologist.
Protocol amendments
There have been no protocol amendments since the initial application. Amendments to the protocol are not currently anticipated, however, if they are required they will be submitted to the two committees mentioned above.
Consent
Trained field workers will obtain written informed consent from all participants. Where an individual is unable to read the information will be read to them and their consent documented by thumbprint, in the presence of an independent witness. Consent for children will be obtained from parents or guardians accompanying them. A copy of the information sheet will be given to each participant. Verbal assent will also be obtained from children before being examined.
Confidentiality
Data will be anonymized prior to analysis and long-term storage by the removal of personal identifying information. Peek database will be encrypted and password-protected with access only granted to staff involved in the study. Data with identifiable information will be secured within a locked project office at KEU, with limited access to only authorised staff.
Access to data
Investigators at LSTHM and Kitale hospital will have access to the final trial dataset. An agreement exists on data sharing and intellectual property. All the data will be archived at LSHTM after the study is completed.
Post-trial care
Given that the trial is being conducted by Kitale Eye Unit, it is integrated into existing health systems and the patients will be managed through these. The control arm clusters will have the same screening service as the intervention arm after the end of the trial.
Dissemination
Summary of the findings will be provided for local stakeholders, Ministry of Health and participating institutions. Publications will be submitted to peer reviewed journals (open access), and presentations made at regional and international conferences and meetings in Kenya and United Kingdon.