Development of a Prediction Model for the Management of Noncommunicable Diseases Among Older Syrian Refugees Amidst the COVID-19 Pandemic in Lebanon

Key Points Question What are the predictors and barriers to managing noncommunicable diseases (NCDs) for older Syrian refugees in Lebanon? Findings This prognostic study including 1893 refugees with at least 1 NCD (chronic respiratory disease, diabetes, history of cardiovascular disease or hypertension) developed a predictive model for the inability to manage any NCD with a moderate discriminative ability. Predictors of inability to manage any NCD included age, no cash assistance, household water and food insecurity, and having multiple chronic diseases. Meaning These findings suggest context-appropriate assistance is required to overcome financial barriers and enable equitable access to health care and medication required to manage NCDs among refugees.

Prior to consent, individuals aged 65 years or older were assessed for their capacity to consent, using five modified items from the University of California, San Diego Brief Assessment of Capacity to Consent 1 . Notably, these five questions were chosen and modified as relevant to the study. The questions used were: (1) What is the purpose of the study that was just described to you?; (2) Do you think you are participating in a study or in an evaluation of services/humanitarian aid?; (3) If you withdraw from this study, will you still be able to receive regular benefits?; (4) If you participate in this study, what are some of the things that you will be asked to do? ; (5) Is it possible that being in this study will not have any direct benefit to you? Each item was rated on a Likert scale from 0 (little to no understanding) to 2 (clear understanding). Therefore, participants who had a score of 7 out 10 were considered as able to participate.

Development of the survey tool
The survey tool was drafted in English, then back translated into Arabic. The instrument included many modules, such as sociodemographic characteristics, health, COVID-19, shelter, household water insecurity, safety and security, social support, violence and trauma, decision making, communication, assets, expenditure, assistance, income, debt, food security, and regularization. Notably, modules varied between waves. The full questionnaires and modules are available online [link:https://scholarworks.aub.edu.lb/handle/10938/22852] 2 .
Following extensive literature review, the survey tool was first drafted by a group of academics who have experience in survey methods and have expertise in humanitarian settings, and migrant and refugee health. It included many validated tools such as Coping Strategies Index, Food Insecurity Experience Scale (FIES) 3 , Household Water Insecurity Experiences Scale (HWISE-4) 4 , 20-Item Short Form Survey Instrument (SF-20) 5 , Mental Health Inventory Scale (MHI-5) 6 , and the Washington Group Disability Scale 7 . The tool was then internally reviewed by academics and humanitarian actors (different departments and sector leads) through multiple consultations to ensure the reliability and usefulness of the tool. The program specialists (water, protection, legal, shelter etc.) within NRC provided questions that would be useful for informing humanitarian programming and they reviewed the survey tool. Once consensus was reached on the different sections of the tool, community consultations with beneficiaries, community focal points and other stakeholders (e.g., Municipality members, Mukhtars, community leaders) took place in three regions in Lebanon (North, Bekaa and South). In these consultations we received in-depth community feedback on the overall study and specific modules/questions within the survey tool. The community consultations allowed the team to better understand the acceptability and validity of the modules included, and whether they met the needs of the beneficiaries. Beneficiaries and other community stakeholders had valuable contributions regarding what factors, needs and vulnerabilities were important to explore. Based on the feedback received, the survey tool was adjusted to capture contextually-relevant perceptions and experiences that were not previously included.
Prior to data collection, a three-day training was delivered to data collectors on the survey tool, humanitarian and research ethics, communication techniques, use of tablets, as well as the referral mechanisms for participants requiring any humanitarian assistance. A one-day refresher training was also given prior to each wave. The survey tool was piloted to assess face validity amongst the data collectors and to capture any errors in the survey tool.
The Arabic version was pilot tested among a random sample of 6 Syrian refugees aged 50 years and older. Several updates were made to the survey instrument following community feedback, data collector training, and the pilot test to ensure that the Arabic version was culturally and linguistically appropriate.
We used computer-assisted telephone interviewing (CATI) where the survey tool was programmed on KoBoToolbox and the interviewer entered the respondent's answers into an electronic form. This data collection platform included logical skip patterns and validation minimizing data entry errors. It also had quick basic data analysis and visualization features, which allowed for real-time monitoring of collected data. During data collection, recordings of the initial interviews for each data collector were listened to and feedback was given to the data collectors. In addition, two research assistants were regularly exporting the data, and ran data monitoring checks to ensure data quality. A weekly meeting was also held with the data collection company to follow up on the data collection process and raise any concerns. The data collection company completed call back checks with a subset of participants each week and randomly checked values entered into the questionnaire. If there were impossible values or missing data on key variables these participants were recalled by the data collection company to check these values and ascertain if the missing items could be completed.

Representativeness of sample
This study used the entire listing of households with an identified older adult from a large NGO in Lebanon (Norwegian Refugee Council (NRC)) as a sampling frame, as a result, this study will be representative of the NRC's beneficiaries. NRC provides humanitarian assistance in the form of shelter, protection, legal assistance, water, hygiene and sanitation (WASH) and education. This study cannot be used to estimate the national prevalence of NCDs for older Syrian refugees in Lebanon. Each year the UNHCR, UNICEF and WFP conduct a nationally representative survey of Syrian refugees households in Lebanon and this sample includes Syrian refugees of all ages so it cannot be directly compared with this study's sample. 8 In addition, a sample of older Syrian refugees are likely to be more vulnerable than a sample of Syrian refugees of all ages. Similar to VASrY, the largest proportions of refugees in the study population where from the Bekaa and the North of Lebanon. In addition, there were comparable proportion of households with family debts (95% in study population vs. 92% in VASyR, 2021). Receipt of cash or voucher assistance was broadly similar; however, they were collected at different time points and cash assistance increased from 47% in August 2020 to 73% in June 2021 (71% in study population (September 2020-January 2021); 73% in VASrY June-July, 2021 (ATM cards and E-cards)). 8