Index | Subindex | Domain | ID | Indicator and definition | Definition | Maximum–minimum | Year/source | Resolution |
SEVI | SVI | Socioeconomic deprivation | 1 | Informal employment | Percent of adults (aged 15–49 years) who work in a manual labour profession such as construction worker and motor vehicle driver | 0.00–15.11 | 201436 | 1×1 km |
2 | Detergent availability | Percent of households where no soap/detergent was observed | 0–97.46 | |||||
3 | Car ownership | Percent living in a household that does not own a private car | 53.90–99.96 | |||||
4 | Place for handwashing | Percent living in a household with no place for handwashing | 4.92–95.14 | |||||
5 | Education attainment | Mean years of school/education attainment | 2.95–21.73 | 201537 | 5×5 km | |||
6 | Unimproved water source | Proportion of households without access to improved water sources | 0.73–95.91 | 201438 39 | ||||
7 | Malnutrition | Prevalence of stunting among children | 21.87–51.34 | 201540 | ||||
8 | Poor households | Proportion of households within the poorest and poorer wealth quintile | 0.44–97.62 | DHS 2014 47 | Subcounty(B) | |||
9 | Shared sanitation facilities | Percentage in the households sharing a toilet facility | 4.07–95.67 | |||||
Population characteristics | 10 | Informal settlements | Percentage of people living in informal settlements and IDP camps | 0.00–81.52 | 201986–88 | See footnote | ||
11 | Elderly population | Percentage of the population aged 65+ years | 0.61–6.30 | 201941 42 | 1×1 km | |||
12 | Single-parent families | Percentage of the population headed by a single parent | 4.56–35.32 | DHS 2014 47 | Subcounty (B) | |||
13 | Crowded households | Percentage in the population with 3+ persons per bedroom | 17.06–88.97 | |||||
14 | Log Population density | Log of the total population per unit area | −0.24 to 5.10 | 201916 | See footnote | |||
15 | Urban population | Proportion of population living in urban areas | 0.00–100.00 | |||||
Access to services | 16 | Access to hospitals | Proportion of population outside 2 hours travel of a hospital | 0.00–100.00 | 201943 | 1×1 km | ||
17 | Health workforce | Number of clinicians and medical officers per population | 0.00–152.2 | 201916 89 | See footnote | |||
18 | Hospital beds | Number of hospital beds per population | 0.00–152.06 | |||||
19 | Access to urban areas | Travel time to the nearest urban centre with at ≥5000 people | 0.00–3641 | 201544 | 1×1 km | |||
EVI | Epidemiological factors | 20 | HIV | HIV prevalence among adults | 0.62–22.67 | 201745 | 5×5 km | |
21 | Smoking | Percent of households with least a daily or weekly smoker | 0.70–28.92 | 201436 | 1×1 km | |||
22 | Obesity | Percentage of adults categorised as obese | 0.00–38.62 | NCD survey 2015 48 | County (B) | |||
23 | Diabetes | Percentage of adults diagnosed with diabetes | 0.00–17.71 | |||||
24 | Hypertension | Percentage of adults diagnosed with high blood pressure | 0.00–53.70 |
The methods used to create each dataset are detailed in the references provided in table 1 and fall into three broad categories. (1) Methods used to create gridded surfaces (denoted with a spatial resolution of either 1×1 km or 5×5 km), using model-based geostatistics that produces probabilistic inference on a spatially continuous phenomenon based on data collected over a finite set of geo-referenced locations90 or the use of a cost distance algorithms to define travel time (spatial access) to either the nearest urban area or health facility.43 44 (2) Generated from household surveys using small area estimation (SAE) techniques (denoted by letter B). SAE smooths estimates by borrowing strength from adjacent units and weights predictions towards the estimated prevalence of neighbouring areas. The spatial dependence was represented through a queen adjacency91 and modelled via the Besag-York-Molliè 2 conditional autoregressive model. (3) The remaining datasets required a combination of steps to derive them. Lists of informal settlements were extracted from online portals86 87 and their extents digitised using Google Earth. Population density maps88 were used to define the proportion of the population living in the digitised settlements by subcounty. Similarly, the proportion of the population living in urban areas by subcounty was derived based on a list of urban areas according to Kenya’s 2019 census.16 Population per unit area (population density) were available by subcounty,16 whereas the number of health workers and hospital beds per population available from the 2018 Kenya harmonised health facility assessment89 were divided by the total population per subcounty16 to derive health workers and beds per 10 000 population. We considered hospitals bed instead of ICU beds that are very few in Kenya.31 When hospital beds are augmented with oxygen supply, they offer an essential requirement for the management of patients with COVID-19 in resource-constrained countries such as Kenya.31
EVI, Epidemiological Vulnerability Index ; ICU, intensive care unit; IDP, internally displaced people; NCD, non-communicable disease; SEVI, Social Epidemiological Vulnerability Index ; SVI, Social Vulnerability Index.