The first positive SARS-CoV-2 case from KRCC was detected on May 22, 2020, in a refugee camp 2 months after the first case of SARS-CoV-2 was reported in Kenya. This case-patient was tested because he had traveled from Nairobi to the camp. By December 31, 2021, a total of 1,024 SARS-CoV-2 infection cases were reported in KRCC; among those, 664 (63.4%) were refugees and 360 (36.6%) members of the host community. The median patient age was 25 (1–102) years; 498 (51.0%) patients were male. Nationally, a total of 297,155 SARS-CoV-2 cases were reported as of December 31, 2021.

During March 2020‒December 31, 2021, a total of 23 COVID-19 deaths were reported in KRCC and 5,442 nationally. The overall CFRs were 2.25% (95% CI 1.29%–3.20%) for KRCC and 1.83% (95% CI 1.78%–1.87%) for Kenya. Of the 23 deaths in KRCC, 19 deaths (CFR 2.86%, 95% CI 1.52%–4.20%) were reported among refugees, and 4 deaths (CFR 1.11%, 95% CI 0.00%–2.33%) among the host community.

Figure 2

Figure 2. Weekly COVID-19 SARS-CoV-2 testing rates per 1,000 population for the general population and among host and refugee populations in Kakuma Refugee Camp Complex, Kenya, March 16, 2020‒December 31, 2021.

During the same period, a total of 18,864 SARS-CoV-2 tests from Turkana West subcounty were performed. Weekly variation in the testing rates during the pandemic (Figure 2) resulted in part from challenges in obtaining supplies for specimen collection and testing; hence, it is likely that many SARS-CoV-2 infection cases were missed and not reported. The testing rate for SARS-CoV-2 was lower than the WHO recommended minimum weekly testing rate of 1/1,000 population, among all 3 populations in our analysis: refugees (10,689 tests; 0.52/1,000 population), host community members (2,635 tests; 0.11/1,000 population), and nationally (3,260,483 tests; 0.69/1,000 population).

Figure 3

Figure 3. Weekly SARS-CoV-2 infection attack rates per 100,000 population for the general population and among host and refugee populations in Kakuma Refugee Camp Complex, Kenya, March 16, 2020‒December 31, 2021.

The overall SARS-CoV-2 infection attack rates per 100,000 persons was 325.0 (95% CI 300.1–349.9) among refugees , 150.2 (95% CI 134.5–165.9) in the host community, , and 628.8 (95% CI 626.6–631.1) nationally. The COVID-19 pandemic in Kenya occurred in waves. Nationally, there were 5 waves: June‒August 2020, September‒December 2020, February‒May 2021, May‒September 2021, and November‒December 2021 (Figure 3). Similarly, refugee and host communities experienced waves of infections. The highest attack rate among refugees and host community members during the first wave, and the subsequent waves were of a lesser magnitude than the previous waves. The third national wave coincided with an increase in attack rate among refugees and host community. In contrast, during the fourth national wave there was no increase in attack rate among refugees and host community members.

This analysis found that refugees were severely negatively affected by COVID-19 during the pandemic. Refugees experienced a lower attack rate but a higher CFR than the population of Kenya as a whole. The low attack rate among refugees compared with national rates is similar to findings in Jordan’s refugee camps (16), which may suggest that refugees were not responsible for the continued spread of SARS-CoV-2 but instead were affected by external SARS-CoV-2 transmission. We noted that the pandemic occurred when the Kenya government had given a nonnegotiable ultimatum for closure of refugee camps that was not enforced at the time (17); stigmatization, increased restriction, or crackdowns on refugees were possible if they were thought to be responsible for driving infection. At the beginning of the pandemic there were fears of catastrophic effects of COVID-19 among refugee settings because those populations were already vulnerable (18). We found a higher CFR among refugees compared to national rates. The low testing rate among host community members may partly explain a lower attack rate, whereas the lower CFR observed among host community members could be a consequence of poor documentation of deaths among this population in Turkana West subcounty.

We cannot fully explain why the refugees experienced a higher CFR than the host and national populations. One possible reason is that, unlike in the host and Kenya populations, all deaths among refugees were meticulously investigated to determine if the cause of death was COVID-19. This process included collection and testing of nasopharyngeal swabs from deceased refugees at both community and health facility levels, potentially inflating the CFR. In addition, contributing factors such as malnutrition, other diseases, or poor health status could put refugees at higher risk for severe disease and death. Unfortunately, that information was not available for analysis.

Measures put in place to either delay introduction of the virus into the camps or contain transmission within Kenya had positive effects and some unintended consequences. In April 2020, the government of Kenya imposed restriction on movement of persons from SARS-CoV-2‒infected areas and a lockdown in urban areas (19,20). Similarly, UNHCR restricted movement into and out of the camps to delay introducing the virus in the refugee camps to enhance pandemic control. Whereas the national lockdown and restriction of movement from SARS-CoV-2‒infected areas were imposed with good intentions, they may have had unintended consequences on refugees. Refugees who were living on daily wages in urban settings experienced job losses that caused them to migrate from urban settings with high SARS-CoV-2 attack rates to the camps, introducing SARS-CoV-2 into the camps, as happened with the first case in KRCC.

After the introduction of the virus to the camp, the first wave among refugees had very high attack rates at a time when the testing rates were very low. The high attack rates were partly caused by inability to significantly reduce contact with many undiagnosed case-patients living with other refugees in crowded shelters. At that time, the estimated population of refugees in the camp was 196,120 in an expected capacity of <100,000 (5). Also, because most of the cases reported were asymptomatic at that time, there was inadequate information among refugees on SARS-CoV-2. Poor adherence to COVID-19 mitigation measures among refugees was not uncommon, which may partly explain the high attack rates reported. In contrast, the attack rate among host community members was low during the first wave, possibly because it was a sparsely populated rural population, which reduced the risk for transmission.

Although we did not test for the SARS-CoV-2 variants circulating in the refugee camps, it is likely that that the variants that were predominant nationally were also circulating in the refugee camp. The third national wave commenced in January 2021 when the more transmissible Alpha variant B.1.1.7 was detected in Kenya (21); during that time, there was a similar increase in attack rate among refugees and host community members. The Delta variant B.1.617.2 was detected in Kenya in March 2021, co-circulating with the Alpha variant (22). The Delta variant took over as the predominant variant during the fourth national wave. It is unclear why, during the period when the Delta variant was predominant in Kenya, the attack rate increased nationally but not among refugees.

We also found that testing was low for refugees nearly throughout the pandemic; some SARS-CoV-2 infections were undetected and thus the reported attack rates are likely an underestimate. The introduction of Ag RDTs and GeneXpert testing in the camp led to improvement in testing rate of refugees, although it did not reach the minimum test per population rate recommended by WHO. The improved testing capacity came late in the pandemic, when it was no longer tenable to perform contact tracing for all confirmed cases. The low testing rate among refugees in KRCC is consistent with findings among other refugee populations (23).

The first limitation of this descriptive analysis is that we relied on routine data collected by different entities which had gaps, and hence, we had to cross-check with other datasets to have a complete dataset for analysis. To do that, we had to make assumptions because of the unique deficiencies in each dataset: the national SARS-CoV-2 testing repository was established after SARS-CoV-2 cases were detected in the camp; the repository was not routinely checked for errors; multiple tests for the same person were captured even if they were done almost at the same time; and manual transfer of data using Microsoft Excel spreadsheets from the testing laboratory to the national SARS-CoV-2 repository could have led to some errors. In addition, we were unable to identify persons who had multiple tests performed during the same infection or those with prolonged shedding of the virus. Prolonged detection of viral RNA does not indicate infectiousness (24). In our analysis, we reported those infections as separate, thus overestimating the attack rates for refugees and for host and national populations. Those deficiencies likely led to reporting bias. Second, the reported positive cases were based on multiple testing platforms such as RT-PCR, GeneXpert, and Ag RDT, which had different sensitivities and specificities. Ag RDT, for example, had a low sensitivity, potentially resulting in false negatives and resulting in an underestimate of the reported attack rates. Third, since we are comparing attack rates, it would have been better if we adjusted for the varying testing rates among the populations; however, this adjustment was not possible because host and refugee populations were a subset of the national database.

In conclusion, KRCC did not escape the COVID-19 pandemic. Refugees experienced a reported lower attack rate but a higher CFR compared with the national rates. It is unclear if the high CFR among refugees was a result of poor utilization of health services, poor access to healthcare by severe cases, low testing rate, or a general predisposition to ill health because of malnutrition or co-infection with other pathogens.

Testing is an essential component in controlling the pandemic to ensure early isolation, contact tracing, and case management. To provide evidence on the need for behavior change to mitigate the spread of SARS-CoV-2, a robust testing system is critical. Therefore, to prepare for the next pandemic, it is essential to strengthen laboratory capacity that would serve not only refugees but also rural communities hosting the refugees.

The challenges with reliable SARS-CoV-2 data from this refugee population also emphasizes the need for improved integration of migrant populations into national data systems to better understand differences in disease transmission during outbreaks and pandemics. Our findings underscore the need to prioritize COVID-19 mitigation measures, including vaccination for refugees.