Prospective Clinical Surveillance for Severe Acute Respiratory Illness in Kenyan Hospitals during the COVID-19 pandemic

Background There are limited data from sub-Saharan Africa describing the pattern of admissions to public hospitals with severe acute respiratory infections during the COVID-19 pandemic.

We conducted a prospective longitudinal hospital-based sentinel surveillance between May 2020 and December 2022 at 16 public hospitals in Kenya. All patients aged above 18 years admitted to adult medical wards in the participating hospitals were included. Demographic and clinical characteristics, COVID-19 infection and vaccination status and outcome data were collected.

Of the 52,714 patients included in the study, 18,001 (35%) were admitted with severe acute respiratory illness (SARI). The mean age was 51 years. Patients were equally distributed across sexes. Pneumonia was the most common diagnosis at discharge. Hypertension, HIV and diabetes mellitus were the most common comorbidities. COVID-19 test results were positive in 2,370 (28%) of the 8,517 (47%) patients that underwent testing. Overall in-patient case fatality for SARI was 21% (n=3,828). After adjusting for age, sex and presence of a comorbidity, SARI patients had higher in-patient mortality compared to non-SARI patients regardless of their COVID-19 status (aHR 1.31, 95% CI 1.19 – 1.46). COVID-19 positive SARI patients had a higher in-patient mortality rate compared to their negative counterparts (aHR 1.31, 95% CI 1.12 - 1.54, p value < 0.0001). COVID-19 vaccine effectiveness against mortality due to SARI after adjusting for age, sex and presence of a comorbidity was 34% (95% CI 11% - 51%).

We have provided a comprehensive description of the pattern of admissions with respiratory illnesses in Kenyan hospitals during the COVID-19 pandemic period. We have demonstrated the utility of routine surveillance activities within public hospitals in low-income settings which if strengthened can enhance the response to emerging health threats. Key Words COVID-19; Hospital Surveillance; Severe Acute Respiratory Illness; Vaccine Effectiveness; Kenya . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 3

Unexpected disease outbreaks such as the COVID-19 pandemic pose unique challenges in characterising the clinical pattern of disease due to their novelty as well as the evolving patterns of clinical manifestation based on pathogen strains(1). By 5th May 2023 when the COVID-19 pandemic was declared over by the World Health Organisation (WHO), Kenya had reported more than 340,000 confirmed cases and over 5,000 deaths due to COVID-19 against a global tally of over 770 million cases and almost 7 million deaths(2). Three years into the pandemic, the pattern of disease among patients admitted to hospital with the disease is still being described with key differences in severity and outcome by specific groups such as age, comorbidities or ethnic groups being apparent(3, 4, 5, 6, 7). However, data on these patterns in Kenyan and other sub-Saharan populations is limited, despite indications that the clinical pattern of illness in these settings was significantly different to that in high-income countries (8). Globally, hospital-based surveillance systems have been key in generating timely data on the evolving clinical pattern of COVID-19 and informing clinical practice(4, 5, 9, 10, 11, 12). These systems have been key in evaluating trends in mortality, admissions and disease burden as well as evaluating vaccine effectiveness against severe disease and mortality (9). Kenya’s national COVID-19 vaccination programme begun on 5th March 2021. As at December 2023, over 23 million vaccine doses of six licensed vaccine products (Moderna, Pfizer/BioNTech, Gamaleya, Janssen(Johnson & Johnson), Oxford/AstraZeneca and Sinopharm) (13) have been delivered (2). However, vaccine coverage in the country remains low at about 36% among adults (2, 14) while no local data on the effectiveness of these vaccines exist. Working with Ministry of Health (MOH) officials in Kenya, we developed a hospital-based surveillance system using existing infrastructure from a clinical information network (CIN)(15, 16, 17) to conduct surveillance for admissions with severe acute respiratory illness (SARI) and COVID- 19 in Kenya. We used this platform to evaluate the effectiveness of licensed COVID-19 vaccines against admission with severe acute respiratory disease in Kenya. This study describes the characteristics of adults admitted to the medical wards of 16 public hospitals with Severe Acute Respiratory Illness (SARI) in Kenya between 5th May 2020 and 31st December 2022. We explore risk factors for poor outcomes overall and among those who tested positive for COVID-19. Additionally, we describe the effectiveness of licensed COVID-19 vaccines against admission with COVID-19 as well as mortality due to SARI. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 4

Study Design We conducted a prospective multi-hospital clinical surveillance between 5th May 2020 and 31st December 2022 in 16 public hospitals in Kenya (Supplementary Figure 1). The study sites were level four and level five hospitals part of a clinical information network (CIN) established in 2013(17). They are spread across the country’s densely populated western, central and coastal regions(18). The surveillance was aimed at monitoring the evolution of the pandemic through review of admissions and management of patients with SARI in the adult medical wards. Additionally, it allowed for the estimation of COVID-19 vaccine effectiveness using a test-negative case-control study approach. Data were collected conducted during the course of routine delivery of care and we did not influence the clinical management of patients in the wards or provide any support for laboratory testing of COVID-19. This study was approved by the Scientific and Ethical Review Unit (SERU) of the Kenya Medical Research Institute (KEMRI). Study Patients Data from all adults (age ≥18 years) patients discharged from medical wards in the participating sites during the study period were included. The discharge date and diagnosis were used to define in- patient episodes to ensure consistency in populations(18). We collected data on the demographic and clinical characteristics, COVID-19 vaccination status, discharge diagnoses and clinical outcome of all included patients. Those with SARI had additional data on their clinical symptoms, laboratory results and management collected. Using Kenyan MOH guidelines, SARI was defined as the presence of fever, cough or shortness of breath that required hospitalisation, in the absence of an alternative diagnosis that fully explains the clinical presentation(19, 20). Confirmed COVID-19 cases were those that had a positive real-time reverse transcription-polymerase chain reaction (RT-PCR) or antigen detection rapid diagnostic test (Ag-RDT) for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in a respiratory sample during or before admission(20, 21). Admission and discharge diagnoses were recorded based on the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes. Data Collection The data collection and analysis procedures for the CIN sites are reported elsewhere(22). In summary, trained clinical officers abstracted data from patient records to an electronic data capture tool at each site daily. De-identified data were then synchronized to KEMRI-Wellcome Trust Research . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 5 Programme (KWTRP) servers where a central data team conducted regular data checks. Additional checks included cross-validation of monthly tallies with daily bed returns reported at the hospitals(18). A dynamic structured data collection tool was adapted for use. The tool was updated regularly in- keeping with emerging knowledge and changes in global and national guidelines for COVID-19 patient management. Key changes included a.) recording of steroid use among patients admitted with SARI following the WHO recommendation of dexamethasone use in management of COVID-19 in June 2020(20, 23); b.) use of Ag-RDT results in addition to RT-PCR results to confirm positive COVID-19 cases(21) and, c.) recording of patient vaccination status following the commencement of the COVID-19 vaccination programme in Kenya in March 2021(24). Statistical Analysis Mortality among patients admitted to the adult medical wards We described the characteristics of all patients in the study. We determined factors associated with mortality during the hospital admission using a forward-built stepwise Cox regression model. Patients entered the risk set at admission and exited at the earliest of either date of death or date of discharge from hospital. The final model included age, sex and presence of a comorbidity as risk factors for mortality. Predictors of mortality among patients with SARI and patients with COVID-19 We used a forward stepwise approach to build a Cox regression model to determine the factors associated with death among SARI patients in general and among SARI patients who tested positive for COVID-19. This analysis was limited to patients who had complete data on all potential risk factors, including age, sex, presence of a comorbidity, leucocyte count, haemoglobin level, temperature oxygen saturation at admission and COVID-19 test results. These variables were defined as confounders if the crude and adjusted rate ratio differed by >10% and/or the stratum specific rate ratios differed by >10%. COVID-19 Vaccine Effectiveness We used a test-negative case control design to estimate COVID-19 vaccine effectiveness (VE). We restricted these analyses to patients admitted after 5th March 2021 when COVID-19 vaccines were available in Kenya. Patients were considered vaccinated if they had received a COVID-19 vaccine at least 14 days before their admission. We used logistic regression, to estimate COVID-19 VE against the following outcomes: (a) admission with COVID-19, and (b) mortality due to SARI (disregarding COVID-19 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 6 status because of low COVID-19 testing at the hospitals). We in addition computed VE estimates by (a) vaccine product and (b) the predominant circulating SARS-CoV-2 variant. For VE against admission with COVID-19, we considered as cases SARI patients who tested positive for COVID-19 while those who tested negative were considered controls. For VE against mortality due to SARI, we considered SARI patients who died during admission as cases while those who were discharged alive were considered controls. Estimates were adjusted by age, sex and presence of a comorbidity. VE was calculated as 1 minus the adjusted odds ratio multiplied by 100. We could not estimate VE by full/partial vaccination status and VE against mortality due to COVID- 19 because of limited COVID-19 testing, incomplete vaccination records, difficulty confirming the vaccination status for some participants from the national vaccination registry and mixing of vaccine products. All data were analysed using Stata/IC™ 16.0 (Stata Corp, College Station, Texas, USA).

General characteristics of study patients We collected data from 52,714 patients admitted across the 16 hospitals. Of these, 18,001(34.2%) were patients classified as having SARI. Of all SARI patients, 4,883(27.1%) had complete data for inclusion in the analysis for predictors of mortality among SARI patients. The mean age of patients was 51 years (Standard deviation (SD) ±8 20). Distribution of patients by sex was equal. The median length of stay in hospital was 5 days (Interquartile range (IQR) 3 to 9 days). Pneumonia was the most common discharge diagnosis with heart failure and pulmonary tuberculosis being the second and third, respectively. One or more comorbidities was present at admission in 16.5% (n=8,683) of all patients. Hypertension (6.5%, n= 3,755) was the most common comorbidity. HIV infection was present in 2,690 patients (5.1%), while 1,975(3.8%) of all patients had Diabetes Mellitus. The proportion of patients receiving specialised care as part of their management i.e., receiving ventilation, IV fluids and/or oxygen was highest during the second quarter of 2021 (13.1%; n=1,342) and lowest during the last quarter of 2022 (2.8%; n=284). Overall, 16.8% (n=8,848/52,714) of the included patients died in hospital (Table 1). COVID-19 vaccination histories were available for 22,943 (43.5%) eligible patients. Of these, 4.0% (n=919) had received at least one dose of a COVID-19 vaccine (Table 1). A COVID-19 test sample was collected from 52.8% (n=8,738) of all SARI patients and results were available for 97.5% (n=8,517) of these (Table 1). . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 7 Mortality among patients admitted to the adult medical wards Figure 1 displays cumulative in-patient mortality for SARI and non-SARI patients. Overall, those with SARI had higher mortality rates compared to those without SARI. Case fatality ratio (CFR) for SARI patients with COVID-19 was 531/2,368 (22.4%) compared to 3,297/15,615 (21.1%) for SARI patients without COVID-19. Case fatality among patients without SARI was 14.5% (5,020/34,675). Using non-SARI patients as the reference category, the crude hazard ratio (HR) for mortality was 1.31 (95% CI 1.25 - 1.38; p value < 0.0001) for SARI patients who were negative for COVID-19. and 1.46 (95% CI 1.33 - 1.61; p value < 0.0001) for SARI patients who tested positive for COVID-19. The difference was slightly reduced after adjusting for age, sex and presence of a comorbidity (COVID-19 negative SARI aHR 1.22, 95% CI 1.15 – 1.30; COVID-19 positive SARI aHR 1.31, 95% CI 1.19 – 1.46; p-value < 0.0001). Male sex, presence of a comorbidity and increasing age were all associated with a higher rate of mortality. Table 2 displays the predictors of mortality due to SARI after adjusting for COVID-19 status, age group, presence of a comorbidity, temperature, oxygen saturation, white blood cell count and haemoglobin levels at admission. Overall, male sex and leucocytosis were the strongest predictors of mortality. COVID-19 Vaccine Effectiveness Of all vaccinated participants included in the VE analysis (n =356), 20.5% (n=73) had received the AstraZeneca, 8.4%(n=30) had received the Janssen, 3.4%(n=12) had received the Moderna and 7.6%(n=27) had received the Pfizer vaccine product as their first dose. The rest, 60% (n=214) reported being vaccinated but did not have a vaccine certificate to verify the vaccine product they’d received. COVID-19 vaccine effectiveness (VE) against mortality due to SARI after adjusting for age, sex and presence of a comorbidity was 34%, 95% CI 11% - 51%, Table 3). Adjusted VE against admission to hospital with COVID-19 was not statistically significant overall (i.e., vaccinated versus unvaccinated) adjusted VE 26%, 95% CI -11,50.5. When stratified by period, VE against admission with COVID-19 was highest during the delta period (61%, 95% CI 13,82) (Table 3). . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 8

The value of surveillance systems during the COVID-19 pandemic cannot be overstated (9). Insights from these systems and other hospital-based studies have been useful in identifying those at increased risk for severe outcomes and informing both pharmaceutical and non-pharmaceutical interventions against the disease(6, 25, 26, 27, 28, 29). Our study found that patients admitted between May 2020 and December 2022 with SARI had a higher mortality rate, especially those with COVID-19, compared to their non-SARI counterparts. As in other studies, our study identified being advanced in age(5, 7, 10, 30, 31, 32, 33, 34), male(3, 5, 6, 7, 10, 32, 33), having leukocytosis(30, 32, 35, 36), anemia and/or a low oxygen saturation(34) at admission as risk factors for death in patients admitted with COVID-19 in Kenya. Similar findings have been reported elsewhere (5, 6, 7, 37). Various contextual factors may explain the results presented. First, inadequate access to specialized care like invasive ventilation and disruptions in the supply of medical commodities may have prevented their use in patients who were at increased risk for poor outcomes. This may have contributed to the higher mortality observed among patients with SARI (38, 39). Secondly, variable testing and recording of co-infection with other infectious diseases like malaria and HIV which are endemic to our region limited the study’s ability to explore mortality among these subgroups. For example, routine testing for malaria was low with only 24.9% of adults tested. Unfortunately, routine testing for malaria was one of the services disrupted by the pandemic in Kenya and may have caused some of the patients with co-infection to miss out on treatment(40). While we report a reduction in hospitalisations for adults and in-hospital mortality during the COVID-19 period compared to the pre- COVID period elsewhere(18), our study still observed the disproportionate burden of mortality due to COVID-19 among the elderly and male patients compared to younger and female patients that has been reported globally. Estimating vaccine effectiveness using observational studies, especially those in low- and middle- income settings such as ours, is challenging and not without shortcomings(41). We report a modest COVID-19 vaccine effectiveness of 34% against mortality due to SARI during the COVID-19 pandemic in Kenya. This VE appeared to be higher among patients who received the AstraZeneca vaccine product (64%). However, we could not definitively determine statistically significant differences in VE by vaccine products due to limited numbers of patients that received the different vaccines. Overall, these VE estimates are lower than those reported elsewhere in sub-Saharan Africa (42, 43, 44, 45, 46). Our estimates, while informative, should be interpreted with caution as they are reported against a backdrop of extensive community spread of COVID-19 before vaccine introduction(47), limited access to and availability of COVID-19 tests and low vaccine coverage with variable availability of the vaccine(14). For example, COVID-19 vaccine coverage among the study participants was 4% compared to a national vaccine coverage of 36%. Difficulty in data linkage across national, regional and local COVID-19 testing, vaccination registries and hospital admissions . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 9 limited the patients included in these analyses. Consequently, the estimates reported are not without bias(48). Specifically, the effect of selection bias, healthy vaccinee bias, misclassification, waning immunity and differential depletion of susceptibles on the final estimates cannot be overlooked. Additionally, limited capacity to verify verbal vaccination histories and mixing of vaccine products due to variable availability of specific vaccine products hindered our study’s ability to estimate VE by all licensed vaccine products. Strengths and Limitations: A major strength of our study is the large sample size achieved during the study period. Where data on clinical outcomes in patients with COVID-19 in our region are limited(49), these results provide key insights on the pandemic within our setting. Our study developed a dynamic data collection tool that allowed real-time review of management practices and clinical outcomes with evolving knowledge on the disease. Whilst the study was not designed to evaluate emerging COVID-19 treatment modalities, its findings on the use of steroids in treating SARI have allowed for further studies to be conducted (50). Despite including multiple sites spread across densely populated areas of the country and having a large sample, the results of this study may not be fully generalizable to the country. Our sample of COVID-19 positive patients represents only about 1% of all cases reported in the country. However, the results are still relevant both nationally and globally. Relying on routine clinical data and documentation practices which differ by individual and by site limited the study’s ability to explore the effect of other known risk factors for death in COVID-19 such as multiple comorbidities at admission, social habits e.g., smoking, HIV infection or exposure to previous medication (6, 10, 29). While completion of demographic parameters was high, poor documentation practices and missing data for clinical parameters and investigations limited the number of patients included in the multivariable analysis. This variability in documentation may have introduced bias to our estimates. Additionally, variable documentation limited our capacity to describe the changing clinical pattern of COVID-19 by dominant SARS-CoV-2 variant or to explore the outcome in COVID-19 patients by evolving treatment practices. Despite these limitations, the study identifies at-risk populations and highlights the need for improved documentation practices, access to specialized care and medical supplies, and more comprehensive testing for co-infections to improve the outcomes of COVID-19 patients in Kenya. Future of study The COVID-19 pandemic has highlighted the need for effective hospital surveillance systems in Kenya(11). By monitoring disease activity within the population in real-time, using signs and symptoms, or preliminary diagnoses, hospital surveillance systems can indicate changes in disease trends that require immediate attention(51). . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 10 Standardization of the minimum clinical data documented at health facilities is necessary to effectively monitor disease activity. Developing a standardized admission tool or collecting key priority information from hospitals and transmitting these data to a national repository would enable data linkage and analysis, thus improving surveillance and planning. Additionally, expanding our study to include real-time on-demand analysis(10) could further enhance the effectiveness of the hospital surveillance system. The future of VE studies nested upon observational studies in low- and middle-income countries like Kenya is bright. More efforts are increasingly being undertaken to optimise methodological approaches and build capacity to conduct these studies. Lessons from the COVID-19 pandemic will be key in developing key framework and designing longitudinal observational studies that can be rapidly adapted to conduct vaccine effectiveness studies in the setting of disease outbreaks.

Our study has generated valuable clinical data on mortality among patients admitted to adult medical wards in the country and estimates of COVID-19 vaccine effectiveness during the COVID-19 pandemic. These data were useful in providing situational awareness during the first three years of the pandemic in Kenya and informing the national response measures. We also demonstrate the potential for routine clinical surveillance systems to inform public health activities in low resource settings. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 11

1. Krämer A, Hossain Khan MM. Global Challenges of Infectious Disease Epidemiology. Modern Infectious Disease Epidemiology: Concepts, Methods, Mathematical Models, and Public health. 2010:23-38. 2. WHO COVID-19 Dashboard [Internet]. 2020 [cited 01 February 2023]. Available from: https://covid19.who.int/. 3. Peckham H, de Gruijter NM, Raine C, Radziszewska A, Ciurtin C, Wedderburn LR, et al. Male sex identified by global COVID-19 meta-analysis as a risk factor for death and ITU admission. Nature communications. 2020;11(1):6317. 4. Garg S, Patel K, Pham H, Whitaker M, O’Halloran A, Milucky J, et al. Clinical trends among US adults hospitalized with COVID-19, March to December 2020: a cross-sectional study. Ann Intern Med. 2021;174(10):1409-19. 5. Maximiano Sousa F, Roelens M, Fricker B, Thiabaud A, Iten A, Cusini A, et al. Risk factors for severe outcomes for COVID-19 patients hospitalised in Switzerland during the first pandemic wave, February to August 2020: prospective observational cohort study. Swiss Med Wkly. 2021;151:w20547. 6. Kim L, Garg S, O'Halloran A, Whitaker M, Pham H, Anderson EJ, et al. Risk Factors for Intensive Care Unit Admission and In-hospital Mortality Among Hospitalized Adults Identified through the US Coronavirus Disease 2019 (COVID-19)-Associated Hospitalization Surveillance Network (COVID-NET). Clin Infect Dis. 2021;72(9):e206-e14. 7. Cardemil CV, Dahl R, Prill MM, Cates J, Brown S, Perea A, et al. COVID-19-Related Hospitalization Rates and Severe Outcomes Among Veterans From 5 Veterans Affairs Medical Centers: Hospital-Based Surveillance Study. JMIR Public Health Surveill. 2021;7(1):e24502. 8. Rice BL, Annapragada A, Baker RE, Bruijning M, Dotse-Gborgbortsi W, Mensah K, et al. Variation in SARS-CoV-2 outbreaks across sub-Saharan Africa. Nature Medicine. 2021;27(3):447-53. 9. Manuel DG, van Walraven C, Forster AJ. A commentary on the value of hospital data for covid-19 pandemic surveillance and planning. Int J Popul Data Sci. 2021;5(4):1393. 10. Williamson EJ, Walker AJ, Bhaskaran K, Bacon S, Bates C, Morton CE, et al. Factors associated with COVID-19-related death using OpenSAFELY. Nature. 2020;584(7821):430-6. 11. Papadomanolakis-Pakis N, Maier A, van Dijk A, VanStone N, Moore KM. Development and assessment of a hospital admissions-based syndromic surveillance system for COVID-19 in Ontario, Canada: ACES Pandemic Tracker. BMC Public Health. 2021;21(1):1230. 12. Jassat W, Cohen C, Kufa T, Goldstein S, Masha M, Cowper B, et al. DATCOV: a sentinel surveillance programme for hospitalised individuals with COVID-19 in South Africa, 2020. Johannesburg, South Africa: National Institute for Communicable Diseases. 2020;10. 13. Team VGC-VT. Kenya COVID-19 Vaccines Tracker 2021 [Available from: https://covid19.trackvaccines.org/country/kenya/. 14. Muchiri SK, Muthee R, Kiarie H, Sitienei J, Agweyu A, Atkinson PM, et al. Unmet need for COVID-19 vaccination coverage in Kenya. Vaccine. 2022;40(13):2011-9. 15. Ayieko P, Ogero M, Makone B, Julius T, Mbevi G, Nyachiro W, et al. Characteristics of admissions and variations in the use of basic investigations, treatments and outcomes in Kenyan hospitals within a new Clinical Information Network. Arch Dis Child. 2016;101(3):223. 16. Tuti T, Bitok M, Malla L, Paton C, Muinga N, Gathara D, et al. Improving documentation of clinical care within a clinical information network: an essential initial step in efforts to understand and improve care in Kenyan hospitals. BMJ Glob Health. 2016;1(1):e000028. 17. English M, Irimu G, Akech S, Aluvaala J, Ogero M, Isaaka L, et al. Employing learning health system principles to advance research on severe neonatal and paediatric illness in Kenya. BMJ Glob Health. 2021;6(3). 18. Ogero M, Isaaka L, Mumelo L, Kimego D, Njoroge T, Mbevi G, et al. Effects of the COVID-19 pandemic on hospital admissions and inpatient mortality in Kenya. medRxiv. 2022. 19. Health) GoKMo. Case Definition for Novel Coronavirus Disease (COVID-19). 2020. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 12 20. Ministry of Health K. INTERIM GUIDELINES ON MANAGEMENT OF COVID-19 IN KENYA. 2020. 21. Ministry of Health K. COVID-19 Antigen Rapid Diagnostic Testing; Interim Guide. Kenya2020. 22. Tuti T, Bitok M, Paton C, Makone B, Malla L, Muinga N, et al. Innovating to enhance clinical data management using non-commercial and open source solutions across a multi-center network supporting inpatient pediatric care and research in Kenya. Journal of the American Medical Informatics Association. 2016;23(1):184-92. 23. Organization WH. Corticosteroids for COVID-19. 2020 2 September 2020. Contract No.: WHO/2019-nCoV/Corticosteroids/2020.1. 24. Ministry of Health K. Guidelines on management of COVID-19 in Kenya. Kenya2021. 25. Silal S, Groome M, Govender N, Pulliam J, Ramadan O, Puren A, et al. Leveraging epidemiology as a decision support tool during the COVID-19 epidemic in South Africa. South African Medical Journal. 2022;112(5b):361-5. 26. Acosta AM, Mathis AL, Budnitz DS, Geller AI, Chai SJ, Alden NB, et al., editors. COVID-19 investigational treatments in use among hospitalized patients identified through the US Coronavirus Disease 2019–Associated Hospitalization Surveillance Network, March 1–June 30, 2020. Open forum infectious diseases; 2020: Oxford University Press US. 27. Llupià A, de la Torre-Pérez L, Granés L, Olivé V, Baron-Miras L, Torà I, et al. SARS-CoV2 hospital surveillance and control system with contact tracing for patients and health care workers at a large reference hospital in Spain during the first wave: An observational descriptive study. Health Sci Rep. 2022;5(2):e513. 28. Jarovsky D, Fongaro GF, Zampol RM, de Oliveira TA, Farias CGA, da Silva D, et al. Characteristics and clinical outcomes of COVID-19 in children: A hospital-based surveillance study in Latin America's hardest-hit city. IJID Reg. 2022. 29. Funk T, Innocenti F, Gomes Dias J, Nerlander L, Melillo T, Gauci C, et al. Age-specific associations between underlying health conditions and hospitalisation, death and in-hospital death among confirmed COVID-19 cases: a multi-country study based on surveillance data, June to December 2020. Euro Surveill. 2022;27(35). 30. Hsu JY, Liu PY, Tseng CH, Liu CW, Yang WT, Huang WH, et al. COVID-19 Screening for Hospitalized Patients: The Role of Expanded Hospital Surveillance in a Low Prevalence Setting. J Multidiscip Healthc. 2021;14:3027-34. 31. Ranucci M, Parati G, Di Dedda U, Bussotti M, Agricola E, Menicanti L, et al. When Outcomes Diverge: Age and Cardiovascular Risk as Determinants of Mortality and ICU Admission in COVID-19. J Clin Med. 2022;11(14). 32. Ombajo LA, Mutono N, Sudi P, Mutua M, Sood M, Loo AM, et al. Epidemiological and clinical characteristics of patients hospitalised with COVID-19 in Kenya: a multicentre cohort study. BMJ Open. 2022;12(5):e049949. 33. Ogboghodo EO, Osaigbovo, II, Obarisiagbon OO, Okwara BU, Obaseki DE, Omo-Ikirodah OT, et al. Facility-Based Surveillance Activities for COVID-19 Infection and Outcomes among Healthcare Workers in a Nigerian Tertiary Hospital. Am J Trop Med Hyg. 2021;104(3):1034-40. 34. Gupta RK, Marks M, Samuels THA, Luintel A, Rampling T, Chowdhury H, et al. Systematic evaluation and external validation of 22 prognostic models among hospitalised adults with COVID- 19: an observational cohort study. European Respiratory Journal. 2020;56(6):2003498. 35. Ghaebi M, Tahmasebi S, Jozghorbani M, Sadeghi A, Thangavelu L, Zekiy AO, et al. Risk factors for adverse outcomes of COVID-19 patients: Possible basis for diverse responses to the novel coronavirus SARS-CoV-2. Life Sci. 2021;277:119503. 36. Zheng Y, Huang Z, Yin G, Zhang X, Ye W, Hu Z, et al. Study of the lymphocyte change between COVID-19 and non-COVID-19 pneumonia cases suggesting other factors besides uncontrolled inflammation contributed to multi-organ injury. MedRxiv. 2020:2020.02. 19.20024885. 37. Garg S, Kim L, Whitaker M, O’Halloran A, Cummings C, Holstein R, et al. Hospitalization rates and characteristics of patients hospitalized with laboratory-confirmed coronavirus disease 2019— COVID-NET, 14 States, March 1–30, 2020. Morbidity and mortality weekly report. 2020;69(15):458. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 13 38. Barasa EW, Ouma PO, Okiro EA. Assessing the hospital surge capacity of the Kenyan health system in the face of the COVID-19 pandemic. PloS one. 2020;15(7):e0236308-e. 39. Barasa E, Kazungu J, Orangi S, Kabia E, Ogero M, Kasera K. Indirect health effects of the COVID-19 pandemic in Kenya: a mixed methods assessment. BMC Health Serv Res. 2021;21(1):740. 40. Kiarie H, Temmerman M, Nyamai M, Liku N, Thuo W, Oramisi V, et al. The COVID-19 pandemic and disruptions to essential health services in Kenya: a retrospective time-series analysis. The Lancet Global Health. 2022;10(9):e1257-e67. 41. Feikin DR, Higdon MM, Abu-Raddad LJ, Andrews N, Araos R, Goldberg Y, et al. Duration of effectiveness of vaccines against SARS-CoV-2 infection and COVID-19 disease: results of a systematic review and meta-regression. The Lancet. 2022;399(10328):924-44. 42. Zhang Y, Belayachi J, Yang Y, Fu Q, Rodewald L, Li H, et al. Real-world study of the effectiveness of BBIBP-CorV (Sinopharm) COVID-19 vaccine in the Kingdom of Morocco. BMC Public Health. 2022;22(1):1584. 43. Chanda D, Hines JZ, Itoh M, Fwoloshi S, Minchella PA, Zyambo KD, et al. COVID-19 Vaccine Effectiveness Against Progression to In-Hospital Mortality in Zambia, 2021-2022. Open Forum Infect Dis. 2022;9(9):ofac469. 44. Collie S, Nayager J, Bamford L, Bekker L-G, Zylstra M, Gray G. Effectiveness and Durability of the BNT162b2 Vaccine against Omicron Sublineages in South Africa. New England Journal of Medicine. 2022;387(14):1332-3. 45. Collie S, Champion J, Moultrie H, Bekker LG, Gray G. Effectiveness of BNT162b2 Vaccine against Omicron Variant in South Africa. N Engl J Med. 2022;386(5):494-6. 46. Gray G, Collie S, Goga A, Garrett N, Champion J, Seocharan I, et al. Effectiveness of Ad26.COV2.S and BNT162b2 Vaccines against Omicron Variant in South Africa. New England Journal of Medicine. 2022;386(23):2243-5. 47. Kagucia EW, Ziraba AK, Nyagwange J, Kutima B, Kimani M, Akech D, et al. SARS-CoV-2 seroprevalence and implications for population immunity: Evidence from two Health and Demographic Surveillance System sites in Kenya, February-June 2022. medRxiv. 2022:2022.10. 10.22280824. 48. Griffith GJ, Morris TT, Tudball MJ, Herbert A, Mancano G, Pike L, et al. Collider bias undermines our understanding of COVID-19 disease risk and severity. Nature Communications. 2020;11(1):5749. 49. Madhi SA, Nel J. Epidemiology of severe COVID-19 from South Africa. The Lancet HIV. 2021;8(9):e524-e6. 50. Lucinde R, Abdi A, Orindi B, Mwakio S, Gathuri H, Onyango E, et al. A pragmatic randomized controlled trial of standard care versus steroids plus standard care for treatment of pneumonia in adults admitted to Kenyan hospitals (SONIA) [version 1; peer review: awaiting peer review]. Wellcome Open Research. 2022;7(269). 51. Nsubuga P, White ME, Thacker SB, Anderson MA, Blount SB, Broome CV, et al. Public health surveillance: a tool for targeting and monitoring interventions. Disease Control Priorities in Developing Countries 2nd edition. 2006. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 14 Declarations Ethical approval This study was approved by the Ministry of Health of Kenya (MOH), Scientific and Ethical Review Unit (SERU) of the Kenya Medical Research Institute (KEMRI/CGMR-C/203/4085), the Oxford Tropical Research Ethics Committee (OxTREC 44-20), and the London School of Hygiene and Tropical Medicine Research Ethics Committee (LEO 26950). Additional approval was sought from all participating hospitals and their respective county health departments. Individual consent for access to de-identified patient data was not required. Data Availability Data are available upon reasonable request. Publishing For the purpose of Open Access, the author has applied a CC-BY public copyright license to any accepted manuscript version arising from this submission. Funding This project was funded by the Wellcome Trust (grants 20991/Z/20/Z and 203077/Z/16/Z), a DFID/MRC/NIHR/Wellcome Trust Joint Global Health Trials Award (MR/R006083/1) and The Bill & Melinda Gates Foundation (OPP1131320). Competing interests All authors declare no competing interests.

We are grateful to the hospital teams, including staff, patients and caregivers, and Ministry of Health colleagues, for supporting the Clinical Information Network and all clinical surveillance activities of the Kilifi HDSS. This article is published with the permission of the Director of the Kenya Medical Research Institute. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 15 Table 1: Distribution of patients by key demographic and clinical parameters Category (*n: all, SARI) All patients n (%) SARI patients n (%) Age (52,711; 18,000) 18 – 30 years 8,369 (15.9%) 2,020 (11.2%) 31 – 40 years 9,292 (17.6%) 2,952 (16.4%) 41 – 50 years 8,465 (16.1%) 2,799 (15.6%) 51 – 60 years 7,784 (14.8%) 2,965 (16.5%) 61 years and above 18,801 (35.75%) 7,264 (40.4%) Sex (52,703; 17,995) Female 26,453 (50.2%) 8,492 (47.2%) Male 26,250 (49.8%) 9,503 (52.8%) Presence of comorbidity (52,714; 18,001) None 44,031 (83.5%) 9,404 (52.2%) One 7,048 (13.4%) 6,978 (38.8%) Two 1,512 (2.9%) 1,496 (8.3%) Three or more 123 (0.23%) 123 (0.7%) Length of stay (52,616; 17,978) Less than 5 days 22,632 (43.0%) 6,792 (37.8%) 5 – 7 days 8,552 (16.3%) 2,972 (16.5%) 1 week – 2 weeks 14,290 (27.2%) 5,405 (30.1%) More than 2 weeks 7,142 (13.6%) 2,809 (15.6%) Outcome (52,658; 17,983) Dead 8,848 (16.8%) 3,828 (21.3%) Alive 43,810 (83.2%) 14,155 (78.7%) Temperature (a ,15,837) Hypothermia ( 37.7 oC) a 1,867 (11.8%) Oxygen Saturation (a , 16,674) Less than 90% a 6,519 (38.8%) 90% to 94% a 4,150 (24.7%) More than 95% a 6,123 (36.5%) White Blood Cell count (a , 10,979) Leucopenia ( 12*109/L) a 2,884 (26.0%) Hemoglobin levels (a , 13,841) Anemia (< 11.5 g/dL) a 5,526 (39.6%) Normal a 8,426 (60.4%) Malaria results (a ,4,456) Negative a 3,998 (89.7%) Positive a 458 (10.3%) COVID results (a , 8,517) Negative a 6,147 (72.2%) Positive a 2,370 (27.8%) *n for each parameter varies due to variable documentation of these data a Data not collected . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 16 Table 2: Predictors of mortality among patients with SARI Mortality rate per 100 person days Crude Rate ratio 95% CI Adjusted rate ratio 95% CI LRT p value COVID status Negative 1.5 0.0012 Positive 1.9 1.30 1.12, 1.53 1.31 1.12, 1.54 Age 18 – 30 years 1.5 0.0049 31 – 40 years 1.7 1.12 0.81, 1.53 1.05 0.77, 1.44 41 – 50 years 1.4 1.02 0.74, 1.40 0.92 0.67, 1.27 51 – 60 years 1.6 1.00 0.73, 1.37 0.90 0.65, 1.22 61 years and above 1.9 1.38 1.05, 1.81 1.25 0.94, 1.66 Sex Female 1.2 0.0001 Male 1.9 1.35 1.16, 1.56 1.37 1.18, 1.59 Presence of a comorbidity None 1.8 0.4891 At least one 1.6 1.02 0.88, 1.17 1.05 0.91, 1.22 Temperature Normal 1.6 0.0766 Hypothermia ( 37.7 oC) 1.8 1.08 086, 1.35 1.11 0.89, 1.39 Oxygen Saturation Less than 90% 2.0 < 0.0001 90% to 94% 1.5 0.68 0.56, 0.83 0.69 0.57, 0.84 More than 95% 1.3 0.64 0.54, 0.75 0.64 0.54, 0.77 White Blood Cell count Normal 1.5 0.0002 Leucopenia ( 12) 2.1 1.37 1.17, 1.61 1.38 1.18, 1.62 Hemoglobin levels Normal 1.6 < 0.0001 Anemia (< 11.5) 1.8 1.08 0.94, 1.25 1.26 1.08, 1.47 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 17 Table 3: COVID-19 Vaccine Effectiveness Vaccine effectiveness against mortality due to SARI Vaccinated patients (n=356) Unvaccinated patients (n=7275) COVID-19 Vaccine Effectiveness (VE) Alive (controls) Dead (cases) Alive (controls) Dead (cases) Crude VE % 95% C. I Adjusted VE % 95 % C. I Vaccination status Vaccinated 306 50 5858 1417 32 8, 50 34 11, 51 Vaccine Product (first dose) Unknown a 177 37 5858 1417 14 -20, 40 17 -19, 42 AstraZeneca 67 6 5858 1417 63 14, 84 64 16, 84 Jannsen (Johnson & Johnson) 27 3 5858 1417 54 -52, 86 54 -54, 86 Moderna 11 1 5858 1417 64 -191, 95 63 -187, 95 Pfizer 24 3 5858 1417 48 -172, 84 46 -81, 84 Predominant variant during vaccination Alpha 31 3 5858 1417 60 -31, 88 64 -19, 89 Delta 103 17 5858 1417 32 -14, 59 34 -12, 60 Omicron 172 30 5858 1417 18 -7, 51 29 -6, 52 Vaccine Effectiveness against admission with COVID-19 SARI Vaccinated patients b (n=186) Unvaccinated patients (n=3652) COVID-19 Vaccine Effectiveness (VE) Negative (controls) Positive (cases) Negative (controls) Positive (cases) Crude VE % 95% C. I Adjusted VE % 95 % C. I Vaccination status Vaccinated 156 30 2866 786 30 -4, 53 26 -11, 50.5 Vaccine Product (first dose) Unknown a 94 22 2866 786 15 -36, 47 8 -47, 43 AstraZeneca 37 5 2866 786 51 -26, 81 49 -30, 80 Jannsen (Johnson & Johnson) 8 1 2866 786 54 -264, 94 40 -386, 93 Moderna 8 1 2866 786 54 -264,94 59 -230, 95 Pfizer 9 1 2866 786 59 -220, 95 61 -211, 95 Predominant variant during vaccination Alpha 18 3 2866 786 39 -107 39 -110, 82 Delta 66 7 2866 786 61 15 61 13, 82 Omicron 51 2 2866 786 -1 -67 -11 -84, 33 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 18 a These patients reported being vaccinated and had SMS confirmation of the same (with the date of vaccination) but their vaccination records were not verifiable on the national vaccine registry b Number of vaccinated patients included in the estimation of VE against admission with COVID-19 SARI was reduced because of some vaccinated patients missing COVID-19 test results . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 19 Figure 1: In-patient mortality among SARI and non-SARI patients . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 20 Supplementary Figure 1: Surveillance study sites . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 21 Supplementary Figure 2: Proportion of Patients Treated with Steroids The figure below summarizes the proportion of severe acute respiratory illness (SARI) patients receiving steroids as part of their treatment regardless of their COVID-19 status. After the recommendation of steroid use in the treatment of COVID-19 was made by the world health organization (WHO) in May 2020, patients with SARI were increasingly managed with steroids regardless of their COVID-19 status . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint 22 Supplementary Table 1: Distribution of SARI Patients by Year The table below summarizes the distribution of SARI patients’ outcome and management by year and quarter of the year during the study period Year Quarter of Year a Patient Characteristics % Deaths (n = 8,848) % Managed in Isolation (n = 2,835) % Required Specialised Care (n = 10,274) 2020 *May to June 2020 7.1 (627) 4.3 (123) 7.4 (756) July to September 2020 11.9 (1,051) 16.1 (455) 12.2 (1,257) October to December 2020 8.5 (753) 20.5 (581) 10.7 (1,097) 2021 January to March 2021 8.9 (788) 12.5 (354) 9.3 (953) April to June 2021 12.6 (1,114) 18.6 (527) 13.1 (1,342) July to September 2021 13.3 (1,176) 18.2 (515) 9.7 (991) October to December 2021 7.0 (617) 4.7 (133) 5.3 (549) 2022 January to March 2022 8.0 (704) 3.8 (108) 11.1 (1,135) April to June 2022 9.7 (861) 0.1 (2) 9.2 (943) July to September 2022 10.2 (906) 0.7 (20) 9.4 (967) October to December 2022 2.8 (251) 0.6 (17) 2.8 (284) aChi square test of difference in distribution of patients by quarter in all three groups had a p value < 0.0001 *Data for April 2020 not included in this analysis as the study had not begun . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprintthis version posted March 4, 2024. ; https://doi.org/10.1101/2024.03.03.24303690doi: medRxiv preprint