This cross sectional study covered the whole country.
Sri Lanka is an island nation in the Indian Ocean, southeast of India, with a total land area of 65,610 km². The population of Sri Lanka is approximately 21 million. Approximately 18% of the population is resident in urban areas . Administratively, Sri Lanka is divided into 9 provinces, and the 9 provinces are further divided into 25 districts. The Medical Officer of Health (MOH) is responsible for preventive health services in a defined area. The MOH area is further divided into Public Health Inspector (PHI) and Public Health Midwife (PHM) areas .
The study population consisted of members of all households in all districts of Sri Lanka. A household was defined as ‘one or more persons living together and who have a common arrangement for provision of food living in a housing unit’ . Public places like homes for elders, orphanages, and religious homes were excluded.
Conceptual setting: Susceptibility as adomain of the social vulnerability framework
Social vulnerability is a developing concept on communication of natural hazards and disasters and is partially the result of social inequalities. Social factors influence susceptibility of various groups and their ability to respond to harm . Social vulnerability is not only due to exposure to hazards alone, but also depends on the sensitivity and resilience of the system to prepare, cope and recover from such hazards . Wisner al. define vulnerability as “the characteristics of a person or group in terms of their capacity to anticipate, cope with, resist and recover from the impact of a natural hazard” . Methods for the Improvement of Vulnerability Assessment in Europe (MOVE) project have elaborated a framework in the context of natural hazards and climate change .
Figure 2 Conceptual framework for social vulnerability adapted from Kienberger, S., & Hagenlocher, M. 
Susceptibility domains were adapted from the conceptual framework for social vulnerability to vector borne diseases by Kienberger & Hagenlocher (Figure 2) . Kienberger and Hagenlocher define social vulnerability to malaria as the predisposition of the population to acquire malaria . It is characterized by different interrelated domains, generic and biological susceptibility, and lack of resilience. The generic and biological susceptibility factors predispose a community to the malaria burden. The resilience factors determine the community’s ability to anticipate, respond to, cope with, or recover from the malaria burden.
Adapted framework for susceptibility to malaria during the prevention of re-establishment phase
Based on the review of literature and extensive consultations with malaria experts, social vulnerability indicators of malaria during the prevention of re-establishment phase in Sri Lanka were identified under two main domains; susceptibility and resilience. The susceptibility domain, assessed in this study, is shown in Figure 3.
Figure 3: Adapted framework for susceptibility to malaria during the prevention of re-establishment phase from Kienberger and Hagenlocher 
Age, gender, pregnancy status, prevalence of parasitaemia, and history of malaria (immunity status) during the past 3 years were considered as biological susceptibility indicators. Migration, socio-economic status and residency of the population (urban/rural/estate) were considered as generic susceptibility indicators. Urban areas were defined as all areas administered by municipal and urban councils. ‘Pradeshiya Sabhas’ were included under the rural sector. Estate sector included the plantation areas . Table 1 compares susceptibility indicators in the framework of Kienberger and Hegenlocher  and those used in this study with a justification for the selection of the indicators.
Table 1: Comparison of susceptibility indicators between framework of Kienberger and Hegenlocher  and the framework used in this study with justification
Kienberger and Hagenlocher, during control phase 
Adapted for PoR* phase
Justification for the selection of indicators
- Number of women of childbearing age
- Number of children under 5
- Prevalence of stunting in children under 5
- HIV prevalence among 15-49 year olds
- Altitude (proxy for immunity)
- Immunity (past history of malaria within last 3 years)
Age <5 &>65 and females are susceptible groups for malaria.
Pregnant mothers and non-immune persons are more prone to get malaria.
Parasitaemia among the population may lead to local transmission. Due to the presence of foreign labour in large numbers, the return of refugees from India re-settled in the Northern Province and the receptivity in the country, this indicator was included.
- Number of women
- Population change
- Travel time to closest urban center
- Number of people living on less than USD 2 per day(Poverty)
- Place of residency(urban, rural and estate sectors)
- Socio-economic status (wealth index)
Malaria was a disease of rural populationswhere the receptivity was higher.
History of travel to malaria endemic countries is an important factor for re-establishment of malaria
There are differences in access to health care, immunity and migration patterns in different socio-economic groups
*PoR refers to prevention of re-establishment
Development of the data collection tool
Based on the conceptual framework, a household survey questionnaire was developed to assess susceptibility to malaria during the prevention of re-establishment phase. The following details were collected: identification of households including socio demographic characteristics of household members, and household characteristics to assess socio-economic status (SES). The assessment of socio economic status (SES) was based on the Demographic Health Survey (DHS) format used worldwide . Wealth index was used to classify the socio-economic status of participants. The source of water, toilet facilities, type of fuel used, material used for the floor of the house, mode of transport, access to mass media, electricity in the household, possession of some household items, and ownership of agricultural land and farm animals were included in the questionnaire. The migration history of household members within the last 3 years and history of fever prior to two weeks of survey was included.
Validation of tool
Questionnaires along with the adapted conceptual framework were distributed among experts to assess judgmental validity (face and content) of the questionnaire. Face and content validities were assessed by malaria experts. The questionnaires were translated using standard methods and pre-tested in the field; necessary changes were made accordingly.
Sample size and sampling
For sample size calculation to assess susceptibility to malaria, there are no guidelines or literature that could be used in the prevention of re-establishment phase. Hence, we considered the number of Sri Lankans traveling overseas as the most important variable to consider in sample size calculation. In 2014, 1,311,258 Sri Lankans departed from the Bandaranaike International Airport which was about 6% of the estimated population . Therefore, to estimate the proportion of Sri Lankans traveling overseas in a year as 6%, with a margin of error ranging from 4.5%-7.5% for a 95% confidence interval, assuming a design effect of 3.2 for using cluster sampling (cluster size of 12), and a non-response rate of 10%, a sample of 3424 households had to be surveyed . Households were selected equally from among the 25 districts proportionate to population in urban, rural and estate sectors. Therefore, from each district 137 households had to be surveyed from 11.4 clusters per district. Hence, twelve clusters of 12 households from each cluster were randomly selected from each district (25districts × 12 PHM clusters ×12 households) to give a total sample size of 3600 households.
A stratified multistage cluster sampling method was used with the primary sampling unit being MOH areas. 6-8 MOH areas were randomly selected from each district. 12 Public Health Midwife (PHM) area clusters were randomly selected from the selected MOH areas (on average 2 PHM clusters per MOH area). From each PHM area cluster, the starting point of the household survey was randomly selected by dropping a headed pin on the PHM area map and the house closest to the pointed edge was selected. After the first house was identified, every tenth house to the left of the selected house was chosen until 12 households for that PHM area were surveyed.
The validated tool was used in the household survey in all districts of the country. The data collectors were trained on sampling technique and the importance of sampling during the training sessions; in the field they were supervised by malaria officers or the principal investigator. The manual for interviewers and Global Positioning System (GPS) data collection were prepared based on the guidelines for conducting Malaria Indicator Surveys as given by the Roll Back Malaria Initiative and given to all data collectors for easy reference. Fieldwork was carried out from July 2016 to March 2017.
The PHM of the area informed the heads of selected households about the survey and the date of interview. On the assigned date, data collectors visited the households and interviewed the head of the household. If the head of the household could not be interviewed (e.g.: household closed), the next closest household was selected. If the head of the household had responded and was not available at the time of survey, the contact number was obtained; the household head was approached a second time, after confirmation of his/her presence, mainly during weekends. If the second attempt failed, the house was visited for a third time; if the head of the household could not be contacted a third time, it was considered a non-responder.
Screening for malaria was done according to the standard operating procedure for parasitological screening for malaria of the Anti Malaria Campaign (AMC), Sri Lanka on the same day the house was visited. Both microscopy and RDTs were used for screening. RDTs were used in areas that did not have microscopy facilities. Microscopy was done by a trained Public Health Laboratory Technologist (PHLT) from each region. Data collection was supervised by regional malaria officers, medical officers attached to Anti Malaria Campaign/Headquarters and the principal investigator.
Data analysis was done using IBM SPSS statistics version 20 software package. Descriptive analyses were used to describe socio-demographic characteristics. Socio-economic status of the population was assessed by the wealth index using exploratory factor analysis. Principal components analysis (PCA) was used to generate a weighted score based on household assets. The wealth index for a household is the linear combination defined as the principal component variable across households or individuals with a mean of zero and a variance of one, corresponding to the ‘Eigenvalue’ of the correlation matrix. The estimated wealth index was based on a population of 13,365 resident in 3,454 households located in all districts in Sri Lanka.
The variables selected for derivation of the wealth index were based on the methodology used by the Department of Census and Statistics in their routine surveys (DHS)  and in the guidelines to conduct a Malaria Indicator Survey.
Descriptive analysis was used to describe variables included in ascertaining biological susceptibility. The association between wealth index, sector of residency and history of migration was analyzed using the chi-square test.