The COVID-19 Pandemic in Sri Lanka: Contextualizing it geographically – II

By Dr. Nalani Hennayake and
Dr. Kumuduni Kumarihamy

(Continued from yesterday)

We need to focus on the geographies within or rather internal geographies of the pandemic. Geographically speaking, this pandemic is now fully localized. The map prepared by the Epidemiological Unit of the Ministry of Health, based on the prevalence of positive cases within the ‘last 14 days,’ provides us with three categories of risk areas, based on MOH divisions (see Map 01, 02, and 03). They are (a) MOH areas where a single case is not reported (yellow areas), (b) Low-risk MOH areas where cases were reported before 14 days (green areas), and (C) High-risk MOH areas where positive cases were reported within 14 days (red areas). These maps generally show how the disease has spread across the country over time.

However, according to this criterion, even when a single COVID-19 patient is found ‘within the last 14 days’, that entire MOH area is automatically designated as high risk. It would have been better, if possible, to combine the prevalence of positive cases within the last 14 days with the number of cases found. In the early days of the second wave, the Epidemiological Unit produced such risk maps daily, but they are no longer available to the public. Instead, the same information is presented in a chart marked in red, green, and yellow, thus avoiding a spatial representation – thus a visual effect – of the pandemic.

As geographers, careful observation of these maps yields two interpretations of possible scenarios, given the fact that it will take a while for the high-risk areas (red areas) to become no-risk areas. First, given how the coronavirus is spreading, and since cases have been reported before 14 days, the low-risk areas (green areas) can transfer into high-risk areas (red areas) at any time. Second, the current no-risk areas (yellow areas) can turn into high-risk (red areas) at any time. Since these are the two obvious patterns that can be observed, it is only a matter of time before the entire Sri Lanka is depicted in red! These kinds of maps engender the fallacy of ‘modifiable areal unit problem’ (MAUP). The MAUP occurs when a single observation measure is aggregated to represent an entire district or a large-scale spatial category. Here, even if a single positive case is found, the entire MOH area would be categorized as high risk, resulting in misrepresentation of the phenomena. Dr. Harith Aluthge, of GMOA, from the very beginning, has been consistently pointing out that it is essential to identify the COVID-19 infected at a smaller scale unit such as of GN divisions. This concern indicates that the COVID-19 pandemic as a public health issue must be dealt with as a ‘local’ issue in the days to come. We were quite intrigued by his repeated attempts to make this simple fact heard by the decision-makers.

Precise and accurate geo-visualization of COVID-19 cases is vital at this stage. However, the official information on COVID-19 is typically published as district-level aggregated data. The most frequently used geo-visualization method of this aggregated data is district-level choropleth maps. For example, Wikipedia has published a map to visualize the spatial distribution of COVID-19 cases as of December 18, 2020 (Map 04). According to this map, Colombo, Gampaha, Kalutara, Kurunegala, Galle, Rathnapura, and Kandy districts are highly vulnerable. If the data are available at spatial units such as Divisional Secretary Divisions (DSDs) or Grama Niladari Divisions (GNDs), the spatial impression will differ. To illustrate this, we have drawn a map for Kandy district (Map 05) using the MOH level data for a specific period, available at the official Facebook page of the Regional Director of Health Services (Map 05). By December 18, 2020, 757 cases have been reported from Kandy, and the most vulnerable DSDs/MOH areas are Akurana (224) and Kandy MC (178). Ududumbara is reported with no cases where Minipe, Panvila, Nawalapiliya, Hataraliyadda, and Galagedara have reported less than 10 cases.

These two maps (Maps 04, 05) demonstrate how choropleth maps depicted at different spatial units can lead to different and, at times, even erroneous conclusions. Spatial heterogeneities of COVID-19 within districts are mostly unknown as the published data are aggregated and generalized to the district level. As mentioned, this kind of statistical bias occurs due to spatial aggregation, commonly known as MAUP. It must be emphasized that the best spatial unit to visualize COVID-19 is the GN division since hidden variations can be depicted at a lower scale within a district.

(To be continued)

(Dr. Nalani Hennayake (nalihennayake@gmail.com) teaches a range of Human Geography courses; Dr. Kumudini Kumarihamy (kumudinik@gmail.com) teaches GIS and Health at the Department of Geography, University of Peradeniya

Peradeniya)