Tracking the introduction and spread of SARS-CoV-2 in coastal Kenya

SARS-CoV-2 was first confirmed in Kenya on March 13, 2020, shortly after the WHO declared COVID-19 a pandemic. Despite rapid implementation of public health measures (border closures, curfews, school and gathering restrictions, quarantine, and isolation), cases rose steadily, suggesting established local transmission. Coastal Kenya, with multiple international points of entry (airports, seaport, and land borders) and major transport links, experienced an early and substantial burden, particularly in Mombasa County. The study aims to elucidate the timing, sources, and spread of early SARS-CoV-2 introductions into coastal Kenya, assess the effectiveness of early interventions, and inform regional control and vaccination strategies by integrating genomic and epidemiological data from March to July 2020.

The authors note a lack of prior Kenya-specific genomic epidemiology on SARS-CoV-2 during the early epidemic. Related large-scale investigations elsewhere and in the region examined travel histories and transmission patterns (e.g., Iceland, Europe/North America) and indirect or limited genomic datasets from East and Southern Africa (e.g., Uganda, Ghana, South Africa). Reports from neighboring countries (e.g., Uganda) implicated international truck drivers as common sources of cross-border transmission. Globally, lineage B.1 was a major European lineage early in 2020. This context motivated a focused genomic surveillance effort in coastal Kenya to complement and expand regional understanding.

- Study design and setting: Samples were collected under Kenya MoH protocols by Rapid Response Teams from six coastal counties (Kilifi, Kwale, Mombasa, Taita Taveta, Tana River, and Lamu) between March 17 and July 31, 2020. Testing eligibility criteria evolved over time, initially symptom-based, later including returning travelers and targeted community testing. - Sample collection and diagnostics: Nasopharyngeal and oropharyngeal swabs were collected and processed at KEMRI-Wellcome Trust Research Programme (KWTRP). RNA extraction used commercial kits (e.g., QIAamp). SARS-CoV-2 detection employed RT-PCR using multiple protocol kits (E gene/ORF targets). - Sequencing: Positive samples underwent whole-genome sequencing using the ARTIC nCoV-2019 V3 protocol with Oxford Nanopore technology. Libraries were barcoded and pooled; standard ARTIC end-prep and ligation steps were followed; libraries were quantified and sequenced. - Genome assembly and QC: Reads were processed with the ARTIC bioinformatics pipeline, mapped to the reference (MN908947.3), and consensus genomes generated with IUPAC ambiguity handling. Sequences passing quality thresholds (~≥80% completeness) were retained for analysis; 311 genomes were included for phylogenetics out of 406 RT-PCR positives analyzed. - Lineage assignment: Consensus sequences were assigned using Pangolin (v3.1.2) and Nextclade (2021-10-19). - Contextual global dataset: GISAID sequences collected up to July 31, 2020 were sampled (approximately five per country per month) to create a contextual set (~2252; alternative sets ~2077/2777 noted) for phylogenetic placement. - Phylogenetic analysis: Sequences were aligned with MAFFT; 3′ and 5′ ends trimmed; known problematic sites masked. Maximum likelihood and time-scaled trees were inferred using TreeTime within a modified Nextstrain Augur workflow, rooted on the reference genome. - Estimating introductions: Two approaches: (1) epidemiologically informed heuristic defining introductions as early (March–April) detections of lineages in individuals with recent travel or sampled at points of entry; (2) computational ancestral state reconstruction using PastML (MPPA with biased transition models) on combined local and global datasets using country and regional geographic traits to infer the origins and number of independent introductions. - Statistical analyses were conducted in R (v4.0.2). Ethics approval obtained (KEMRI SERU #4035).

- Sequencing output: 311 high-quality SARS-CoV-2 genomes generated from samples collected March 17–July 31, 2020 (out of 406 positives analyzed; representing <1% of positives tested in the lab during the period). - Multiple introductions: Phylogenetic interspersion with global sequences indicates multiple independent introductions into coastal Kenya. Estimates include a lower bound of at least 24 introductions (heuristic), and computational estimates of at least 34 introductions; an upper-bound range of approximately 34–37 introductions to coastal Kenya and surrounding areas. - Geographic origins: Most introductions were inferred to be of European origin, though introductions via neighboring countries likely occurred, including through land border points of entry. - Dominant lineage: Lineage B.1 accounted for 74% of sequenced cases and showed the only extensive local expansion, especially in Mombasa, during April–July 2020. Other detected lineages (A, B, B.4) were primarily associated with individuals at border screening or recent travelers and did not sustain local transmission. - County patterns: Multiple lineages were observed across counties; Lamu experienced early cases dominated by lineage B.1.1.33, with phylogenetic clustering suggesting a local outbreak possibly due to founder effect; B.4 cases in Mombasa were linked to travelers (including from Lusaka, Zambia) and one local case. - Epidemiology of sequenced cases (n=406 positives underlying dataset): Median age 37 years (range 1–85); 66% male; over half asymptomatic. Recent travel history identified in 26%; 16% were sampled at ports of entry; approximately 41.3% had no travel history, indicating established local transmission. - Mutational profile: Most sequences had 4–16 substitutions relative to reference; D614G (A23403G) in Spike was prevalent, alongside mutations including R203K/G204R in N. - Points of entry: International airports, seaport, and land borders were key conduits for introductions; early interventions (screening, quarantine, isolation, contact tracing, travel restrictions) limited onward transmission for many introduced lineages.

The integration of genomic and epidemiological data indicates that coastal Kenya experienced numerous SARS-CoV-2 introductions in early 2020, largely of European origin but also via neighboring countries through land borders. Despite multiple introductions, only lineage B.1 established widespread local transmission, particularly in Mombasa, suggesting differences in lineage prevalence were influenced by both introduction frequency and potentially transmissibility. The concentration of introductions at points of entry and the limited subsequent spread of many lineages support the effectiveness of early public health measures—border screening, quarantine, isolation, contact tracing, and restrictions on international travel—in curbing transmission chains. Nonetheless, undetected or poorly sampled introductions (including via internal travel) seeded local epidemics, underscoring the need for sustained genomic surveillance and coordinated regional strategies across East Africa (e.g., harmonized screening of truck drivers and border policies). Findings align with reports from neighboring countries and global patterns (dominance of B.1), emphasizing the value of genomic epidemiology to evaluate interventions and inform control policies.

This study provides the first large-scale genomic epidemiology of early SARS-CoV-2 in coastal Kenya, revealing multiple introductions primarily of European origin and highlighting points of entry as critical conduits. While many introductions did not result in established transmission due to early interventions, lineage B.1 became dominant and drove local spread, especially in Mombasa. The work demonstrates the utility of genomic surveillance for tracking introductions and evaluating public health responses. Future efforts should prioritize sustained, systematic sampling, enhanced data sharing across East Africa, and rapid genomic analyses to detect and interrupt emerging transmission chains, including monitoring lineage dynamics as the epidemic evolves and vaccination programs scale.

- Incomplete epidemiological metadata for some cases (e.g., demographics, travel histories). - A substantial number of genomes (n≈91) were incomplete due to amplicon dropout or low viral loads, reducing resolution. - Non-systematic, evolving testing and sampling strategies may introduce biases in lineage representation and temporal inference. - Limited availability of contemporaneous genomes from neighboring countries in sub-Saharan Africa hindered precise inference of introduction sources and routes. - Findings from coastal Kenya may not generalize to other regions without further data.