Hospital-level patterns of COVID-19 mortality in Ghana: a retrospective analysis of 2020-2022 routine surveillance data

¹School of Public Health, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana, ²German West-African Centre for Global Health and Pandemic Prevention, Kumasi, Ghana, ³Department of Health Care Management, Technische Universität Berlin, Berlin, Germany, ⁴Department of Planetary and Public Health, University of Bayreuth, Bayreuth, Germany, ⁵Center for Global Health, Charité-Universitätsmedizin, Berlin, Germany, ⁶System Modeling Group, Institute of Veterinary Epidemiology and Biostatistics, Freie Universität Berlin, Berlin, Germany, ⁷Faculty of Geo-Information Science and Earth Observation, University of Twente, Enschede, Netherlands, ⁸Public Health Division, Ghana Health Service Headquarters, Osu, Ghana, ⁹Global Health and Infectious Diseases Research Group, Kumasi Centre for Collaborative Research in Tropical Medicine, Kumasi, Ghana, ¹⁰Global One Health Research Group, Bernhard Nocht Institute of Tropical Medicine, Hamburg, Germany

^&Corresponding author
Gideon Kwarteng Acheampong, School of Public Health, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana

Introduction    

The COVID-19 pandemic has resulted in millions of deaths worldwide. As of August 17, 2025, the World Health Organization reported 778,509,414 confirmed COVID-19 cases and 7,100,227 deaths globally [1]. Substantial differences exist in reported mortality rates across various geographic areas, primarily due to variations in demographic structures, disparities in emergency response capacities, and differences in surveillance systems [2-5]. By the end of 2021, the COVID-19 mortality rate in Africa stood at 3.49%, lower than that in North America (4.57%) and Europe (3.74%). Despite this relatively lower mortality rate, healthcare systems in a number of African countries faced challenges during the pandemic, including impaired active surveillance, poor healthcare infrastructure, and shortages of trained professionals, particularly in critical care [6-8].

Ghana recorded the third-highest number of COVID-19 cases and deaths among the 16-member states of the West African sub-region as of February 2022 [9]. In response to the pandemic, Ghana designated primary, secondary, and tertiary healthcare facilities for managing severe COVID-19 cases, while asymptomatic and persons with mild illness without underlying conditions were advised to self-isolate [10]. Despite these efforts, the healthcare system's capacity to handle the increasing number of cases remained a serious concern, including an insufficient number of facilities and their uneven distribution across the country [11], highlighting the need to understand how COVID-19 outcomes varied across different levels of healthcare in Ghana. Studies have examined the distribution of COVID-19 mortality in parts of Europe, Asia, and the Americas using large databases [12-14]. However, the patterns of COVID-19 mortality in Low- and Middle-Income Countries (LMICs) remain poorly understood. Ghana's health surveillance system collects data on demographics, healthcare facility visits, symptom profiles, and underlying medical conditions [15], providing a valuable opportunity to analyse how COVID-19 outcomes were distributed across different healthcare levels in an African context. Understanding mortality patterns across healthcare levels in Ghana offers lessons for other LMICs, where referral pathways, diagnostic capacity and critical care resources are often constrained.

Against this backdrop, this study aimed to analyze the distribution of COVID-19 cases, deaths and case fatality rates (CFRs) across hospital levels of care in Ghana, examine the characteristics of cases and deaths at different levels of healthcare, and assess the distribution of facilities reporting cases and mortalities across Ghana. Specifically, the study addressed the following questions: (i) how are COVID-19 cases and deaths distributed across hospital levels in Ghana? (ii) how do CFRs vary across levels of care? and (iii) what patient characteristics are associated with COVID-19 mortality?

Methods     

Study setting

The study was conducted in Ghana, a West African country which shares borders with Ivory Coast to the west, Burkina Faso to the north, Togo to the east, and has the Gulf of Guinea to its south. In 2022, Ghana's population was 30,829,615, with females comprising 50.7% and males 49.3% [16]. Greater Accra is the most populous region, followed by Ashanti. Ghana's health sector is structured into three administrative tiers: national, regional, and district levels [17]. Healthcare services are also categorized into three levels: primary, secondary, and tertiary care. Primary healthcare, the foundational and entry level of the system, includes Community-Based Health Planning and Services (CHPS) facilities, Reproductive and Child Health (RCH) Units, Maternity Homes, Health Centres, Polyclinics, Clinics, and District Hospitals [18]. Most private and faith-based institutions operate at the primary care level, with only a few extending to secondary care [19]. Secondary healthcare involves specialist services, where primary healthcare facilities refer patients requiring more specialized treatment [19]. Regional hospitals serve as designated secondary care facilities. Tertiary care provides the highest level of expertise, specialized equipment, and advanced medical interventions, with teaching hospitals representing the apex of this system [19]. As of December 2020, amid the COVID-19 pandemic, Ghana had a total of 8,825 health facilities. Approximately 90% of these were primary care facilities, followed by secondary care, with tertiary care facilities being the least common [20].

Study design and data sources

This was an observational study employing a retrospective cross-sectional design using routinely collected COVID-19 surveillance data from the Ghana Health Service between March 2020 and February 2022. A cross-sectional approach was appropriate as it allowed the assessment of the distribution of cases, deaths and case fatality rates across different hospital levels at a national scale within a defined period. The retrospective use of existing surveillance data enabled the inclusion of a large, nationally representative dataset without influencing patient outcomes and ensured comprehensive coverage of reported cases, enhancing the generalizability of findings.

Study population and eligibility criteria

The study population included all laboratory-confirmed COVID-19 cases reported to the Ghana Health Service, Disease Surveillance Department between March 2020 and February 2022. Cases were included if they had a documented outcome (survived or died) and were linked to a reporting health facility, community surveillance structure or laboratory. Cases recorded outside the study period were excluded from the analysis. For CFR analyses, cases with missing outcome data and cases referred to another facility without a confirmed outcome at the receiving facility were excluded from the analysis. A total of 152,896 confirmed COVID-19 cases were included in the analysis.

Sampling and sample size: this study utilized the complete national COVID-19 surveillance dataset for the study period; therefore, no sampling was performed. All eligible cases meeting the inclusion criteria were included in the analysis. As a result, no formal sample size calculation was required.

Variables: the primary outcome variable was COVID-19 mortality (death or survival). Independent variables included age, sex, presence of underlying medical conditions (comorbidities), patient status (in-patient or out-patient), referral status, reporting facility name and hospital level (primary, secondary, tertiary).

Data organization: healthcare facilities were classified as either ‘Non-hospital primary care facility´ (including CHPS, Maternity Homes, RCHs, Health Centres, Clinics, and Polyclinics) or hospital-based. Hospital-based facilities were further categorized into primary, secondary, or tertiary care levels. The classification was guided by a pre-established facility list from the Ghana Health Service-Institutional Care Division and the Health Facility Regulatory Agency (HeFRA) of Ghana.

Bias: potential sources of bias include underreporting of cases and deaths, incomplete data capture, and variations in testing and reporting practices across regions and facility types. Differences in referral patterns across hospital levels may also have influenced CFR estimates.

Data analysis

Data obtained from the National Surveillance Department were reviewed for completeness and internal consistency. Duplicate records had already been removed at source prior to data extraction. Analyses of continuous and categorical variables were carried out using STATA'18 (StataCorp.2023. Stata Statistical Software: Release 18. College Station, TX: StataCorp LLC). QGIS 3.16 (QGIS Development Team, 2009. QGIS Geographic Information System. Open Source Geospatial Foundation. URL http://qgis.org) was used to generate spatial outputs. Frequencies of COVID-19 cases and deaths across facilities were generated. Subsequent analyses focused on case and mortality distribution by hospital level. Frequencies and proportions were generated according to hospital level for the following categorical variables: age group, sex, presence of underlying medical condition or comorbidity, type of facility visited, patient status and outcome. Data on whether a case was referred to another hospital or discharged was available and included in the analysis; however, the name and type of ‘referred hospital´ was absent from our data sources. Consequently, we only computed proportion of out-referrals by referring hospital. For this analysis we included only cases with hospitalization outcome recorded at the reporting facility. Cases referred to another facility without a known outcome at the receiving hospital were excluded from CFR calculations to prevent double counting or incomplete outcome attribution to any facility. This approach ensures that all deaths and recoveries were captured within the dataset, minimizing potential bias from incomplete outcome data. Summary statistics were conducted for "Age".

CFRs were computed as the proportion of deaths among confirmed cases and among hospitalized cases. Ninety-five percent (95%) confidence intervals (CI) were calculated using binomial exact methods. CFR was further computed by case characteristics across hospital levels, and Chi-square tests were conducted to assess associations between COVID-19 CFR and case characteristics also at a 95% significance level. Missing data were assessed for all variables, including age, sex, and presence of comorbidities. The extent of missingness for each variable is presented in Table 1. No imputation was performed given the descriptive nature of the study and the lack of auxiliary variables to inform imputation. Analyses were therefore conducted using non-missing data for each variable and proportions were computed excluding missing values. Sensitivity checks showed that the overall distribution of age and sex among cases with complete data was similar to the full dataset, suggesting limited bias from missingness. Graduated symbol maps were generated to examine the regional distribution of facilities recording COVID-19 cases and deaths in Ghana. A chi-square test was used to examine whether there were statistically significant differences in CFR across categories of underlying medical condition or comorbidity, age group and sex.

Ethical considerations: this study was approved by the Ghana Health Service Ethics Review Committee with reference code: GHS-ERC-010/06/24 as well as the Committee on Human Research, Publication and Ethics (CHRPE) of the Kwame Nkrumah University of Science and Technology (KNUST), Kumasi, Ghana (Reference: CHRPE/AP/517/25). Permission to use the data was granted by the Ghana Health Service.

Results     

Distribution of COVID-19 cases and mortalities across hospital levels

Out of 152,896 confirmed cases recorded nationwide, 1,442 related-deaths occurred between March 2020 and February 2022, corresponding to a population CFR of 0.94%. Most deaths (1,210; 83.9%) and nearly half of all cases (71,404; 46.7%) occurred in health facilities. However, when considered together, standalone laboratories and ‘non-hospital’ healthcare facilities accounted for a higher number of cases (81,492; 53.3%) than hospitals. Primary facilities recorded the highest number of hospital-reported cases (42,340; 59.3%), although they had the lowest number of cases per facility (111.13), compared to secondary (638.35) and tertiary facilities (1,598.00). The highest number of deaths occurred in tertiary facilities (478; 39.5%), which also had the highest deaths per facility (53.11) (Table 1, Figure 1). The geographic distribution of these cases and deaths is illustrated in Figure 2. A concentration of hospitals reporting cases was observed in the southern regions, particularly Greater Accra, Ashanti, Eastern, Central, and Western (Figure 2A). Reported deaths were concentrated in the Ashanti and Greater Accra regions (Figure 2B). Additionally, whereas all regions had at least one primary and secondary health facility reporting COVID-19 cases, the reporting tertiary facilities were exclusively located in the Greater Accra, Ashanti, Central, Northern, and Volta regions, which are the only regions with such facilities in Ghana (Figure 2).

Case Fatality Rates (CFRs) across hospital levels

Among cases with complete hospitalization outcomes, the overall hospital-reported CFR was 1.69% [95% CI: 1.60 - 1.78], while in-hospital CFR among hospitalized patients was 13.11% [95% CI: 12.37 -13.87]. In-hospital CFR ranged from 10.08% in primary facilities to 14.16% in secondary and 16.78% in tertiary facilities (Table 2). COVID-19 out-referrals were most frequent in primary care facilities (1,122; 2.55%) and least common in tertiary facilities (27; 0.26%) (Table 1). Cases in secondary and tertiary facilities were slightly older (median age: 35 years) compared to those in primary facilities (33 years), and a higher proportion had comorbidities (tertiary: 11.33%; secondary: 7.93%; primary: 5.79%). Similarly, the proportion of elderly cases was higher in tertiary (6.48%) and secondary (4.44%) facilities than in primary facilities (3.86%).

Patient characteristics and factors associated with mortality

COVID-19 deaths were more common among males (56.86%). The median age of cases and deaths recorded in hospitals was 34 years (IQR: 26-45) and 62 years (IQR: 49-75), respectively. The highest number of deaths occurred among individuals aged 70 years and above (414; 34.22%). More than half of all deaths (638; 52.73%) had underlying medical conditions/comorbidities (Table 1). In-hospital mortality varied significantly by age group (p  <  0.001), increasing from 4.12% among patients aged 19 years and below, and peaking among persons in the 50-59-year age group (27.93%). Mortality rate also differed significantly by sex (p = 0.021), with males recording a higher in-hospital CFR (16.29%) than females (14.05%). Similarly, mortality was significantly higher among patients with underlying medical conditions (p  <  0.001), with an in-hospital CFR of 29.76% compared to 7.99% among those without comorbidities (Table 2).

Discussion     

This study retrospectively examined COVID-19 CFR as well as case and mortality characteristics according to hospital levels using data from Ghana's National Disease Surveillance System. Previous studies in Ghana have primarily focused on the demographic and clinical characteristics of COVID-19 cases and deaths [21-23] with limited attention to how cases and deaths were distributed across health facilities. To our knowledge, this is the first national-level analysis in Ghana to stratify CFR by hospital level, revealing systematic differences in case severity and mortality risk across tiers of healthcare. We found that, while most confirmed cases were reported in ‘Non-hospital´ facilities and laboratories (53.3%), the majority of deaths occurred in healthcare facilities (83.9%). Amongst hospitalized cases, COVID-19 in-hospital CFR was 13.11% and increased with increasing level of healthcare. In-hospital mortality was also associated with older adults, males, and patients with comorbidities.

In our study, over half of all confirmed cases (53.3%) were recorded from standalone laboratories and primary care facilities. This widespread detection by primary care facilities partly contributed to Ghana´s low population CFR (0.94%) by detecting a large proportion of mild or asymptomatic cases. These findings reinforce the role of primary care systems in the early detection of COVID-19, emphasizing the need to continuously build their capacity to enhance surveillance. The in-hospital CFR (13.11%) observed in our study was substantially higher than the population CFR of 0.94%, reflecting the concentration of more severe cases among hospitalized patients. This pattern is consistent with reports from other settings, where hospital-based CFRs exceed population-level estimates [24]. The high CFR recorded among hospitalized cases in Ghana compared to the population CFR, highlights the need for improved triage systems, strengthening hospital capacity, particularly in secondary and tertiary facilities and early risk stratification to reduce COVID-19 mortality in Ghana. Although lower than estimates reported in some high-income settings [25,26], the observed in-hospital CFR suggests a significant burden of mortality within Ghana´s hospitals, potentially reflecting a combination of late presentation, limited access to critical care, differences in case severity and referral patterns, and the high prevalence of comorbidities among cases.

In the African context, the in-hospital CFR observed in this study falls within the range reported in similar settings, aligning closely with findings from Democratic Republic of Congo (13.2%) [27]. Variations in CFR across countries may reflect differences in disease severity, testing capacity, and mortality surveillance systems. The observed CFR in this study likely reflects the combined influence of these factors, including case detection practices and health system capacity [2,28]. These findings highlight the heterogeneity in facility-level outcomes and underscore the importance of strengthening mortality surveillance systems particularly during health emergencies.

Mortality in our analysis was significantly associated with male sex, older age, and the presence of comorbidities. These findings align with existing evidence from Ghana and other settings, where these factors are consistently identified as key determinants of adverse COVID-19 outcomes [21,29,30]. The higher median age among deaths (62 years) compared to that for cases (34 years), highlights the vulnerability of older populations. Notably, the elevated mortality observed among individuals aged 50 - 59 years suggests that factors beyond advanced age may also contribute to risk in this group. These findings underscore the need for targeted interventions for high-risk populations, including early clinical risk stratification, timely referral, and preventive measures. Among hospitalized COVID-19 cases, CFR increased with higher health facility level. This gradient in CFR likely reflects differences in the demographic and clinical profile of admitted patients across the tiers of hospital care. Specifically, patients admitted to secondary and tertiary facilities were slightly older than those in primary facilities, and a higher proportion had comorbid conditions. The share/proportion of elderly patients followed a similar pattern, with tertiary and secondary facilities admitting more elderly individuals than primary facilities. These findings are consistent with reports from other settings, where CFR increases with level of care due to referral of more severe cases [14,24]. The observed pattern underscores the importance of strengthening capacity in secondary and tertiary facilities to effectively manage severe cases during health emergencies.

COVID-19 cases and mortalities were more concentrated in hospitals in southern Ghana than in the north in our study. This pattern likely reflects higher population density and greater economic and social activity in southern regions, which may facilitate transmission. These findings highlight the need for strengthened surveillance and response strategies in densely populated urban areas. Regional differences in testing intensity may have influenced case detection and CFR patterns. Urban regions, particularly in southern Ghana, likely captured more mild or asymptomatic cases due to higher testing capacity, which may have lowered apparent CFR. In contrast, under-testing in less-resourced areas may have missed milder cases, potentially inflating CFR estimates. As facility-level testing data were not available, these differences could not be accounted for; therefore, comparisons of CFR across regions or hospital levels should be interpreted with caution.

Limitations

The surveillance data obtained from the Ghana Health Service did not include persons re-infected with COVID-19, and therefore the total number of new COVID-19 cases for the period may have been undercounted. Although data on patient referrals were available, the receiving facilities were not recorded, preventing tracking of cases across multiple hospitalizations. Additionally, data completeness was limited because some surveillance records were unavailable. Regional variations in case detection, evolving variants, vaccination, and changing clinical practices may have affected CFR estimates; findings should be interpreted cautiously.

Conclusion     

Our analysis of COVID-19 mortality across different levels of care in Ghana showed that cases were concentrated in facilities located in the urban southern regions. Mortality was significantly associated with patient age, male gender, and the presence of comorbidities. COVID-19 in-hospital CFRs were higher in secondary and tertiary care facilities, which were characterized by having received higher severity cases compared to primary care facilities. These findings highlight the need to strengthen urban surveillance, hospital referral systems, and infrastructure in higher care facilities in Ghana to better manage patients to improve health outcomes. Future studies should examine facility-level clinical processes and time-to-death metrics to identify avoidable mortality factors.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Conceptualization: Gideon Kwarteng Acheampong, John Amuasi, Wilm Quentin, Reinhard Busse, Verena Struckmann, Hanna-Tina Fischer. Data curation: Gideon Kwarteng Acheampong, Emmanuel Nakua, Dennis Odai Laryea, Franklin Asiedu-Bekoe. Formal analysis: Gideon Kwarteng Acheampong, Emmanuel Nakua, Andrzej Jarynowski, Vitaly Belik. Methodology: Gideon Kwarteng Acheampong, John Amuasi, Wilm Quentin, Reinhard Busse, Verena Struckmann, Hanna-Tina Fischer, Emmanuel Nakua. Supervision: John Amuasi, Wilm Quentin, Emmanuel Nakua. Visualization: Ernest Akyereko, Isaac Buertey Gormey. Writing - original draft: Gideon Kwarteng Acheampong. Writing - review & editing: Gideon Kwarteng Acheampong, John Amuasi, Wilm Quentin, Reinhard Busse, Verena Struckmann, Hanna-Tina Fischer, Emmanuel Nakua. All authors read and approved the final manuscript.

Acknowledgments     

This study was supported by the German-West African Centre for Global Health and Pandemic Prevention (G-WAC) scholarship of the German Academic Exchange Service (DAAD) as part of the Global Centres Programme. Additional supervisory support was provided by the Flattening the Curve Project of the Berlin University Alliance (BUA). The authors express their gratitude to the Ghana Health Service, Disease Surveillance Department, for granting access to the COVID-19 surveillance data.

Tables and figures     

Table 1: demographic and clinical characteristics of COVID-19 cases and deaths across hospital levels in Ghana, Mar 2020 - Feb 2022

Table 2: case fatality rates (CFR) of COVID-19 by hospital level and patient characteristics in Ghana, Mar 2020 - Feb 2022

Figure 1: distribution of COVID-19 cases and deaths across healthcare levels in Ghana, Mar 2020 - Feb 2022

Figure 2: geographic distribution of COVID-19 cases and deaths across hospital levels in Ghana, Mar 2020 - Feb 2022 (A,B)

References