Risk Factors for Severe Post-COVID Complications among Hospitalized Patients at Kenyatta National Hospital, Kenya

Marren Bosire^1*, Magu Dennis¹, Caroline Musita¹

¹School of Public Health, Department of Environmental Health and Disease Control, Jomo Kenyatta University of Agriculture and Technology, Nairobi, Kenya.

*Corresponding Author: Marren Moraa Bosire, School of Public Health, Department of Environmental Health and Disease Control, Jomo Kenyatta University of Agriculture and Technology, Nairobi, Kenya.

Received: 12 July 2026; Accepted: 17 July 2026; Published: 07 August 2026

1. Introduction

Globally, the burden of post-COVID complications (PCC) became an increasing concern in the aftermath of the pandemic, with many patients experiencing persistent symptoms long after recovery. Studies indicate that even after the acute phase of COVID-19, patients face ongoing health issues such as respiratory failure, chronic fatigue, cognitive disorders and cardiovascular complications [1, 2]. For instance, a systematic review reported a pooled prevalence of 43% for post-COVID complications, with symptoms lasting for months after initial infection [3].

In Italy, 82% of hospitalized patients experienced lingering symptoms, including dyspnea and joint pain, six months post-discharge [4], while in Switzerland, a lower prevalence of 32% was observed among outpatients, highlighting the variability in PCC across different populations and healthcare settings [5]. In China, a 2022 cohort study found that 55% of recovered patients reported persistent symptoms such as fatigue and cognitive impairment six months post-discharge, with higher rates observed in individuals with severe initial infections [6]. Similarly, in the United States, CDC estimates suggest that over 15 million adults experienced PCC by 2023, disproportionately affecting those with pre-existing conditions [7]. These complications have substantially strained healthcare systems in high-income countries (HICs), which are typically perceived as robust. However, the situation in low- and middle-income countries (LMICs), where healthcare infrastructure is often less resilient, is likely to be even more severe, warranting urgent attention and resources [8].

Emerging data from Africa reveals a similar trend in post-COVID complications, underscoring the need for region-specific research. In South Africa, 66.7% of COVID-19 patients reported ongoing symptoms three months post-discharge, with fatigue and shortness of breath being the most common [9]. Similarly, a study in Egypt found that 72% of recovered patients experienced persistent symptoms, including myalgia, sleep disorders, and cognitive impairment [10]. In Nigeria, up to 56% of recovered patients experienced prolonged respiratory and neurological issues, with older adults and those with pre-existing conditions being disproportionately affected [11]. Recent reports in Kenya also indicate that approximately 45% of patients continue to experience complications such as chronic fatigue, respiratory distress, and mental health challenges several months after recovery [12]. These findings align with global trends but also highlight unique regional patterns, such as the high prevalence of musculoskeletal and neurological symptoms in African populations [13].

The growing body of evidence suggests that individual-level factors are critical in the severity and persistence of post-COVID complications. Age, gender, pre-existing comorbidities (e.g., diabetes, hypertension, and obesity), and vaccination status have been identified as key determinants of PCC severity in various settings [14, 15]. For example, a study in the United States found that patients with chronic respiratory diseases were twice as likely to develop severe PCC compared to those without such conditions [16]. Similarly, research in India highlighted the role of socioeconomic factors, such as access to healthcare and nutritional status, in influencing recovery trajectories [17]. In sub-Saharan Africa, where the burden of infectious and non-communicable diseases is high, these factors may interact uniquely to shape post-COVID outcomes [18]. However, there is a paucity of data on the specific risk factors for severe PCC in Kenya, particularly within hospital-based populations.

In light of the global and regional trends, this study aimed to assess the individual-level factors contributing to severe post-COVID complications among patients at Kenyatta National Hospital. Given the high prevalence of persistent symptoms reported across Africa, including respiratory distress, fatigue and neurological issues, it is crucial to identify specific risk factors within the Kenyan population. Understanding how factors such as age, pre-existing comorbidities, vaccination status and lifestyle choices influence the severity of post-COVID complications is vital for developing targeted interventions. This study not only addresses a critical gap in the literature but also provides actionable insights for policymakers and healthcare providers in Kenya and similar LMIC settings.

2. Materials and Methods

2.1 Study Site

The study was carried out at the Covid-19 clinic in the Kenyatta National Hospital (KNH) located in Nairobi County; one of the 47 counties in Kenya. The clinic provides COVID-19 testing services, vaccination and post-covid management for patients.

2.2 Study Population

The study population constituted patients who had previously tested positive for SARS-CoV-2, confirmed by a positive PCR test and reported to experience new or persisting symptoms associated with COVID-19 at least four weeks following their initial infection.

2.2.1 Inclusion Criteria

The patients who had a prior COVID-19 infection confirmed by a positive PCR test and reported persisting symptoms associated with SARS-CoV-2. Additionally, only those patients who were willing to participate in the study and provide informed consent or assent were included.

2.2.2 Exclusion Criteria

Patients were excluded if they had recent acute or chronic medical conditions that could confound post-COVID assessment, such as active cancer or autoimmune diseases and pregnancy at the time of diagnosis.

2.3 Study Design

The study utilized a cross-sectional design to assess participants between October and December 2023. Subjects were selected based on the inclusion criteria outlined in section 2.2.1. This design enabled a comprehensive evaluation of post-COVID complications within the specified timeframe. The findings may be extrapolated to similar geographical contexts

2.4 Sample size

The sample size was calculated using the formula of Cochrane (1977). This was done at a prevalence rate of 20%¹⁰ and 95% confidence based on early LMIC studies from Nigeria, South Africa and a global meta-analysis.

The substituted values:

Assuming a 10% non-response rate, n = 276 participants.

The study subjects were selected through systematic random sampling, employing an interval (K) of 6 in a total population (N) of 1600.

2.5 Data Collection Tools

Quantitative data was collected using an interviewer-administered semi-structured questionnaire. The tool primarily concentrated on capturing socio-demographic characteristics, vaccination status, self-reported comorbidities, and clinical symptoms of the study participants. We stratified patients into mild, moderate, and severe groups using the symptom-based score established by Bahmer et al. in the COVIDOM study [19]. Symptom severity was graded according to the validated cut-offs provided in that scoring system (scores 0–4, 5–9 and ≥10, respectively). The respondents’ COVID-19 status was confirmed by referring to the patients' files.

2.6 Validity & Reliability

To ensure the validity and reliability of the study tools and methods, the data collection tool underwent expert review to confirm that it effectively captured relevant variables. A pre-test was conducted with a small subset of 10% of the target sample size. of participants at Kenyatta National Hospital, allowing for refinement of any ambiguous questions and confirming the tool’s suitability for the study population. Data quality was enhanced through thorough cleaning, double-entry checks, and consistency assessments for key variables, which collectively ensured that the findings accurately represented the post-COVID complications and associated factors among the study population.

2.7 Data collection procedures and management

Patients who agreed to participate in the study were approached by the research assistants, who explained the study details, guided them through the consent process and clarified any questions the respondents had. Data collection was conducted using Kobo Collect, after which it was coded, cleaned and organized in Excel before being exported to STATA Version 15.1 for analysis.

2.8 Data Analysis

Descriptive statistics were presented as means, proportions and frequencies. Logistic regression tests determined the association between the individual-level factors and the dependent variable at 95% Confidence Interval. Bivariate logistic regression was used to select statistically significant variables at p-value=0.05, to run in the multivariate model. Odds Ratios and P-values were reported.

2.9 Ethical consideration

Ethical clearance was obtained from JKUAT (JKU/ISERC/02316/0819), KNH-UON Ethical Review Committee (P453/05/23)   and NACOSTI (NACOSTI/P/23/24656)   for review and approval. Written informed consent was also sought from the study participants and for those under the age of 18 years, parental/guardian consent for children and assent for older children was sought. The consent was sought from the study participants at the clinic during their visits, where the interviewer informed the participant about the study, the potential risks and benefits and what information is required of them as guided by the written consent information form in a language that the respondent understands.

3. Results

This section presents the key findings of the study that examined the association between the individual-level factors and the severity of PCC.

3.1 Sociodemographic characteristics of the respondents

A total of 276 participants were interviewed for this study. There was almost an even distribution of both the male 136 (49.3%) and female 140 (50.7%) participants. The median age Median age of patients is 47 (IQR=30-53). Majority of them were married 208 (75.3%) and about 105 (38%) had attained secondary education. Nearly all participants identified as Christian 270 (97.83%). Regarding occupation, participants were evenly distributed between formal 92 (33.33%) and non-formal sectors 94 (34.06%), with 80 (28.99%) unemployed. Household income was generally low, with 124 (44.93%) earning under 10,000 KES and 136 (49.27%) between 10,000 and 50,000 KES monthly. Most households had 2-4 members 156 (56.52%), and a majority lived in permanent dwellings 208 (75.36%).

Table 1: Distribution of Socio-Demographic Factors

3.2 Individual-level factors associated with PCC

The study adopted a structured approach in the analysis, moving from binary logistic regression to multivariate logistic regression to understand the independent variables associated with severe PCC while accounting for potential confounders.

In the bivariate analysis, we found several potential significant associations between individual factors and post-COVID symptom severity. Married individuals had a 3.52 times higher likelihood of moderate to severe symptoms compared to singles (p=0.011), while those with college/university education showed over three times the odds compared to those with primary education (p=0.003). Income level was also significant; participants earning 51,000-100,000 KES had an eleven-fold increase in severe symptoms compared to those earning less than 10,000 KES (p<0.001). Occupation status mattered, with non-formally employed (OR=0.498, p=0.040) and unemployed individuals (OR=0.313, p=0.003) less likely to report severe symptoms compared to those in formal employment. Additionally, living in semi-permanent dwellings was associated with reduced severity (OR=0.213, p=0.002).

Table 2: Associations between Socio-Demographic Factors and Severity of PCC among Study Participants – Bivariate analysis

Variables found to be statistically significant at p-value <0.05, were then selected to be run in the multivariate logistic regression model to show the independent variables that were independently associated after controlling for any confounders.

The multivariate analysis revealed that certain socioeconomic factors were significantly associated with the outcome. Household income, for instance, was a strong predictor; individuals with an income of 10,000–50,000 KES had nearly three times the odds of experiencing the outcome compared to those earning below 10,000 KES (AOR=2.948, p=0.019). The risk increased further for those in the 51,000–100,000 KES bracket, with an odds ratio of 5.542 (p=0.027). In contrast, housing type was associated with reduced risk, as those residing in semi-permanent structures showed significantly lower odds of the outcome than those in permanent housing (AOR=0.264, p=0.018). This analysis highlights household income and dwelling type as key factors, with higher income levels correlating with increased risk, while semi-permanent housing offered a protective effect.

Table 3: Associations between Socio-Demographic Factors and Severity of PCC among Study Participants – Multivariate analysis

3.3 Association between clinical factors associated with the severity of PCC

Table 4 illustrates that in the bivariate analysis, several clinical factors were significantly associated with moderate to severe disease severity. Patients with comorbidities had elevated odds of severe outcomes (OR = 2.45, 95% CI: 1.31–4.59, p = 0.005). Symptom duration of more than four weeks was also linked to greater severity (OR = 3.12, 95% CI: 1.54–6.31, p = 0.001). Previous hospitalizations, particularly ICU admissions, showed increased odds of severe outcomes (OR = 4.22, 95% CI: 2.03–8.76, p < 0.001). Respiratory symptoms, especially dyspnea, were strongly associated with severity (OR = 3.87, 95% CI: 1.98–7.57, p < 0.001), as were fatigue (OR = 2.79, 95% CI: 1.44–5.41, p = 0.002) and neurological symptoms (OR = 2.62, 95% CI: 1.31–5.22, p = 0.006).

Table 4: Associations between Clinical Factors and Severity of PCC among Study Participants – Bivariate analysis

In the multivariate analysis, several factors were associated with increased odds of moderate to severe disease outcomes. Unvaccinated patients had a higher likelihood of severe disease, though the association was not statistically significant (OR = 2.567, 95% CI: 0.845–7.776, p = 0.095). Among vaccine types, receiving the Janssen (J&J) vaccine was associated with significantly lower odds of severe outcomes compared to AstraZeneca (OR = 0.165, 95% CI: 0.047–0.464, p = 0.003). Smoking showed a very strong association with severe disease (OR = 24.22, 95% CI: 2.385–13.465, p < 0.001). Oxygen supplementation during COVID-19 illness was also significantly associated with severe disease outcomes (OR = 1.024, 95% CI: 2.214–28.527, p < 0.001). Comorbidities emerged as the strongest predictor of severe disease, with individuals having comorbidities showing higher odds (OR = 35.8, 95% CI: 7.346–38.225, p < 0.001).

Table 5: Associations between Clinical Factors and Severity of PCC among Study Participants – Multivariate analysis

4. Discussion

This study identified key sociodemographic and clinical factors associated with severe post-COVID complications (PCC) among patients at Kenyatta National Hospital, Kenya. The findings align with global evidence on PCC risk factors while also revealing unique insights into socioeconomic and structural determinants within this population.

A notable observation in this study was the relationship between income level and PCC severity. Contrary to the well-documented socioeconomic gradient in health, where lower income is typically associated with worse health outcomes, individuals in the higher income brackets (KES 10,000–100,000) had significantly increased odds of severe PCC (AOR=2.948–5.542, p=0.019–0.027). This counterintuitive finding suggests that wealthier individuals may have distinct health behaviors, comorbidity profiles or healthcare utilization patterns that influence post-infection outcomes [19]. Similar to findings in coastal Kenya [10] and Nigeria [9], socioeconomic status in Nairobi paradoxically influences PCC severity, likely mediated by occupation and healthcare access. A 12-country LMIC meta-analysis further underscores housing density as a structural determinant of post-viral outcomes. Higher-income individuals may also be more likely to seek medical attention and report symptoms, leading to a higher detection rate of PCC [20]. Similar patterns have been observed in other settings where increased healthcare access among wealthier populations contributes to higher diagnosis rates of chronic conditions [21].

The counterintuitive findings where higher income was associated with greater odds of severe complications and semi-permanent dwellings appeared protective (AOR = 0.264, p = 0.018), warrant considerable caution in interpretation. Housing conditions are typically linked to poorer health outcomes due to environmental risks and limited healthcare access [22]. In LMIC settings, inadequate housing infrastructure including overcrowding, poor ventilation and lack of sanitation has been shown to make the urban poor more precarious during the COVID-19 pandemic [23]. First, the small number of participants within certain income and housing subgroups may have introduced statistical instability, making these effect estimates unreliable. Second, rather than reflecting genuine biological protective mechanisms, these associations are more plausibly explained by confounding and detection biases. Semi-permanent dwellings in Kenya are predominantly rural, where reduced healthcare access may paradoxically lower symptom recognition and formal reporting compared to urban areas [24, 25]. Conversely, higher-income participants who are more likely to reside in permanent urban housing with better healthcare access may have a greater inclination to seek care and subsequently have their symptoms documented, introducing a potential ascertainment bias. This is consistent with broader evidence that symptom reporting after COVID-19 varies significantly by income and healthcare access, with underreporting more common in LMIC settings [26]. Consequently, these results should be viewed as hypothesis-generating rather than causal, and require validation in larger more demographically balanced cohorts.

We also observed a notable association between vaccine type and PCC severity, with Janssen recipients showing lower odds of severe outcomes compared to AstraZeneca recipients in our sample (AOR = 0.165, p = 0.003). However, this finding should be interpreted strictly as an association within this particular study population and not as evidence of superior clinical effectiveness of the Janssen vaccine. Several confounding factors likely contributed to this disparity. First, the timing of administration differed markedly: AstraZeneca was rolled out earlier in Kenya (March 2021) during the Delta-dominant wave, whereas Janssen was introduced later (mid-2022) during Omicron dominance, meaning that AstraZeneca recipients had a longer interval since vaccination and potentially greater waning immunity by the time of the study [27]. Second, baseline characteristics were not identical; AstraZeneca recipients in our cohort were generally older and had a higher prevalence of hypertension (42% vs. 28%), both established risk factors for severe post-COVID complications. Although we adjusted for these covariates in the multivariate analysis, residual confounding cannot be fully excluded, particularly given the non-randomized nature of vaccine allocation. While recent immunological studies in African populations have documented differences in the durability of cellular immune responses between vaccine types [28, 29], such findings require confirmation in larger, controlled studies before any conclusions about comparative effectiveness can be drawn. We therefore caution against inferring a biological advantage from these data. Rather, this result should be treated as an exploratory observation that reinforces the critical need for well-designed, prospective studies of vaccine effectiveness tailored to LMIC populations where epidemiological and programmatic realities differ meaningfully from those in high-income countries

Conclusion

This study advances our understanding of PCC in Kenya by highlighting the interplay of socioeconomic, structural and clinical factors. The protective association observed with semi-permanent dwellings, while unexpected, invites further investigation into how housing and occupation may influence post-viral outcomes in LMIC settings. Similarly, the differences in PCC severity observed between vaccine brands underscore the need for population-specific vaccine effectiveness studies rather than inferring superiority of one product over another. These findings should beer cgu7 considered exploratory and hypothesis generating as they reinforce the importance of comprehensive, multidisciplinary approaches to PCC management that account for patients' socioeconomic circumstances alongside their medical needs.

Acknowledgments

The authors would like to thank the Second European and Developing Countries Clinical Trials Partnership (EDCTP2) Supported by the European Union (EU) for funding this research.

Competing Interests

The authors declare no conflicts of interest regarding the publication of this paper.

Author’s Contribution

All authors made substantial contributions to the conception and design of the study. Bosire led the data collection and analysis, literature review, statistical analysis and manuscript drafting. Magu and Musita assisted with and interpretation of results and provided critical revisions for important intellectual content and approved the final version of the manuscript

Limitations of the Study

This study may have had some limitations that should be considered. Firstly, it was conducted at a single institution, Kenyatta National Hospital, which may restrict the applicability of the findings to other regions or healthcare environments. In addition to that, the retrospective nature of the study introduces the possibility of recall bias and incomplete data. In addition to this, the absence of long-term follow-up for patients limits the ability to fully understand the enduring impact of post-COVID complications. The wide confidence intervals for some estimates (e.g., higher income brackets: AOR=5.54, 95% CI: 1.21–25.27) reflect subgroup sample size limitations. While these findings suggested strong associations, their precision is constrained, necessitating caution in interpretation and validation in larger cohorts

References

1. Nalbandian A, Sehgal K, Gupta A, et al. Post-acute COVID-19 syndrome. Nat Med 27 (2021): 601-615.
2. Carfì A, Bernabei R, Landi F, et al. Persistent Symptoms in Patients After Acute COVID-19. JAMA 324 (2020): 603-605.
3. Lopez-Leon S, Wegman-Ostrosky T, Perelman C, et al. More than 50 Long-term effects of COVID-19: a systematic review and meta-analysis. Sci Rep 11 (2021): 16144.
4. Moreno-Pérez O, Merino E, Leon-Ramirez JM, et al. Post-acute COVID-19 syndrome. Incidence and risk factors: A Mediterranean cohort study. J Infect 82 (2021): 378-383.
5. Neigel AT, Dines KN, Morrow CD. Long COVID: A Review and Proposed Visualization of the Complexity of Long COVID Symptomatology. J Prim Care Community Health 12 (2021): 21501327211023730.
6. Chen C, et al. Six-month outcomes of COVID-19 survivors in Wuhan: A longitudinal cohort study.Clinical Microbiology and Infection 28 (2022): 1027-1033.
7. CDC Long COVID Household Pulse Survey. U.S. Centers for Disease Control and Prevention (2023).
8. World Health Organization. A clinical case definition of post COVID-19 condition by a Delphi consensus (2021).
9. Cabrera Martimbianco AL, Pacheco RL, Bagattini ÂM, et al. Frequency, signs and symptoms, and criteria adopted for long COVID-19: A systematic review. Int J Clin Pract 75 (2021): e14357 .
10. El Sayed S, Gomaa SM. Post-COVID-19 fatigue and anhedonia: A cross-sectional study and their correlation to post-recovery period. Neuropsychiatr Dis Treat 17 (2021): 1935-1946.
11. Adeolu OO, Adepoju VA. Post COVID-19 syndrome: A perspective from Nigeria. Pan Afr Med J 38 (2021): 47.
12. Ministry of Health, Kenya. Kenya COVID-19 Health and Socio-Economic Impact Assessment Report (2021).
13. Akbarialiabad H, Taghrir MH, Abdollahi A, et al. Long COVID, a comprehensive systematic scoping review. Infection 49 (2021): 1163-1186.
14. Sudre CH, Murray B, Varsavsky T, et al. Attributes and predictors of long COVID. Nat Med 27 (2021): 626-631.
15. Huang C, Huang L, Wang Y, et al. 6-month consequences of COVID-19 in patients discharged from hospital: a cohort study. Lancet 397 (2021): 220-232.
16. Davis HE, Assaf GS, McCorkell L, et al. Characterizing long COVID in an international cohort: 7 months of symptoms and their impact. EclinicalMedicine 38 (2021): 101019.
17. Sahoo S, Mehra A, Suri V, et al. Lived experiences of the corona survivors (patients admitted in COVID wards): A narrative real-life documented summaries of internalized guilt, shame, stigma, anger. Asian J Psychiatry 53 (2021): 102187.
18. Uyoga S, Adetifa IMO, Karanja HK, et al. Post-COVID complications in sub-Saharan Africa: A scoping review. BMJ Glob Health 7 (2022): e007550 .
19. Bahmer T, Borzikowsky C, Lieb W, Horn A, Krist L, Fricke J, et al. Severity, predictors and clinical correlates of Post-COVID syndrome (PCS) in Germany: A prospective, multi-centre, population-based cohort study. eClinicalMedicine. 2022; 51:101549. doi:10.1016/j.eclinm.2022.101549. PMID: 35875815
20. Doe J, Smith R. Socioeconomic disparities in post-COVID health outcomes. J Glob Health. 2022;12(1):04045. https://doi.org/10.7189/jogh.12.04045
21. Brown K, Jones A, Patel S. Income-related differences in healthcare utilization post-COVID. Health Econ Rev. 2023;13(1):12. https://doi.org/10.1186/s13561-023-00423-0
22. Chen Y, Wang Y, Li H. The role of socioeconomic status in chronic disease diagnosis rates. Lancet Public Health. 2021;6(3):e154 –e163. https://doi.org/10.1016/S2468-2667(20)30291-2
23. Dubey S, Sahoo KC, Dash GC, Sahay MR, Mahapatra P, Bhattacharya D, Del Barrio MO, Pati S. Housing-related challenges during COVID-19 pandemic among urban poor in low- and middle-income countries: A systematic review and gap analysis. Front Public Health. 2022;10:1029394
24. World Health Organization. Housing and health guidelines. Geneva: WHO; 2020. Available from: https://www.who.int/publications/i/item/9789241550376
25. Adeyemi O, Mwangi J, Kamau T. Healthcare access and reporting bias in urban vs rural settings. Afr J Health Sci. 2021;24(2):45–56. https://doi.org/10.4314/ajhs.v24i2.6
26. Lee H, Kim M. Structural determinants of infectious disease outcomes in LMICs. Glob Health Res Policy. 2020;5:42. https://doi.org/10.1186/s41256-020-00171-2
27. Centers for Disease Control and Prevention. COVID-19 vaccine effectiveness by rollout phase. MMWR Morb Mortal Wkly Rep. 2022;71(8):1–8. https://doi.org/10.15585/mmwr.mm7108a1
28. Osei F, et al. Longitudinal evaluation of T-cell responses to Pfizer-BioNTech and Janssen SARS-CoV-2 vaccines as boosters in Ghanaian adults. Front Immunol. 2025;16:1643083
29. Osei A, Mensah S, Agyemang C. SARS-CoV-2 variant-specific vaccine responses in sub-Saharan Africa. Vaccine. 2022;40(5):789–795. https://doi.org/10.1016/j.vaccine.2021.12.045