Assessment of Suspected COVID-19 Deaths in Nonhealthcare Settings in Côte d'Ivoire,March 11 to July 31,2020

Abstract

According to hospital records, 5 months after reporting its first case of COVID-19, Côte d'Ivoire reported only 102 deaths. We conducted a community mortality survey in the 13 districts where 95% of COVID-19 cases were reported to assess COVID-19 mortality in nonhealthcare settings. To identify suspected COVID-19 deaths in communities, we used data from social and administrative institutions, such as police and fire departments, funeral homes, and places of worship, whose functions include providing services related to deaths. Our survey identified 54 (17.6%) suspected COVID-19 deaths, which is more than half of the official reported number. Our study showed that in areas with low access to healthcare and poorly functioning death notification and registration systems, community-based data sources could be used to identify suspected COVID-19 deaths outside of the health sector. They can provide early warning data on events, such as an unusual number of community deaths or diseases.

Introduction

Côte d'Ivoire reported its first confirmed COVID-19 case on March 11, 2020.¹ By July 31, 2020, the country reported 15,813 confirmed cases and 99 deaths.²

The initial low weekly number of confirmed cases could be attributed to insufficient testing. At the time, the Institute Pasteur of Côte d'Ivoire was the only laboratory capable of running COVID-19 tests, but it did not have enough test kits to run more than a few hundred tests each day. The weekly average number of tests that were run between March and July 2020 was between 950 and 1,000.³ The 7-day moving average of confirmed COVID-19 cases was 113, with an average positivity rate of 15% between April and July 2020 (range between 4.9% and 40.8%). The low number of COVID-19 deaths reported was assumed to be an undercount because it included only hospitalized cases and did not consider deaths that might have occurred in communities. Due to the lack of a functional system of registration of vital statistics in Côte d'Ivoire, records for deaths that occur outside of healthcare settings are both inaccessible and incomplete. Although death reporting is compulsory by law in Côte d'Ivoire, most deaths go unreported. According to police sources in Abidjan, 90% of deaths that they investigate are not reported. Causes of deaths are rarely investigated except for deaths from accidents, homicides, or suspicious deaths, for which prosecutors order investigations by a medical examiner or a coroner.⁴

Additional factors may have further contributed to the low number of confirmed cases, limiting the number of deaths that could have been attributed to COVID-19. For example, the stigma attached to COVID-19 may have prevented people from seeking healthcare who could have otherwise been found positive and hospitalized. At the same time, the reliance on traditional medicine to treat COVID-19 signs and/or symptoms or lack of trust in the health system—leading many people to self-treat—are additional factors that limit the number of people who access care and die at hospitals. Between 30% of people in urban areas and 63% in rural areas use traditional healing in Côte d'Ivoire.⁵

We used community-based data to assess suspected COVID-19 deaths in adults that occurred between March 11 and July 31, 2020, in the 13 districts of the greater Abidjan area, where 95% of COVID-19 cases reported occurred. In this article, we discuss the implications for strengthening community-based surveillance and the capacity for early detection of suspected COVID-19 cases and other emerging diseases and public health emergencies in Côte d'Ivoire. Considering that the total population of Côte d'Ivoire is approximately 26 million inhabitants,⁶ and these 13 districts have 23% of Côte d'Ivoire's population, according to local government estimates, we estimated the overall study area population to be 6 million inhabitants. With about half of the population younger than 18 years of age,⁷ the adult population in these 13 districts was estimated to be 3 million inhabitants in 2020.

Methods

To collect data on the occurrence and causes of death, we used a variety of approaches, including record reviews and interviews with families and officials. We used several data sources from the public and private sectors to create a database of deaths. The sources included records from police departments, fire departments, city halls, funeral homes, and places of worship, such as churches and mosques. Data sources were obtained from community organizations whose functions include the provision of formal or informal services in relation to deaths (Table 1). All data were merged and cross-checked with police records to obtain a uniform cross-sectional retrospective list of 1,131 deaths of all causes outside of healthcare settings reported in the study area between March 11 and July 31, 2020. The research team also collected telephone number(s) of persons listed as contacts for information about the deceased persons, including their close relatives or family members.

Table 1.

Types of Community Organizations Used as Sources of Data on Deaths of All Causes, Greater Abidjan, Côte d'Ivoire, September 2021

Organizations and Institutions	Services Provided Related to Deaths
Funeral homes	Organize funeral ceremonies, including burials
Places of worship (eg, mosques, churches, temples)	Conduct prayers for the dead
Police departments	Protect against deadly force, prevent violence
Fire departments	Rescue from life-threatening situations
City and town halls	Register deaths
National Institute of Statistics	Conduct population surveys
National Institute of Legal Medicine	Investigate causes of suspicious deaths, generally because of an accident or homicide; work with public prosecutors

Telephone numbers were available for 1,085 (96%) family contacts of the decedents. However, of those, only 561 of the telephone numbers (52%) were working. Of these, we successfully reached 467 (83%) contacts. We conducted face-to-face interviews with 309 (66%) contacts about the deaths in their family. Three of the family contacts were for deaths of children younger than 18 years of age and were removed from the sample, resulting in a final sample size of 306. Of the 158 (34%) contacts that we did not interview, 29 (18%) refused to be interviewed mainly because they had no desire to recall a painful memory or denied that their relative's death had anything to do with COVID-19, and they stopped answering our phone calls as a result; 96 (61%) kept postponing their appointment times and were ultimately unavailable; and 33 (21%) said they were notified about the death of the deceased but were unable to provide any information about the person or the circumstances of the death. To explore other plausible explanations of missed mortality, we asked our interview respondents if they had observed any unusual number of deaths in their communities since the beginning of the COVID-19 pandemic.

Our sample of 306 deaths, which includes only adults 18 years and older, is 27% of the 1,131 total deaths from all causes identified from our sources during the study period. This sample size is larger than the 287 needed for a probability level of significance at P=.05 according to the following formula: $\begin{matrix} n & = \frac{t^{2} \cdot N \cdot p (1 - p)}{N \cdot δ^{2} + t^{2} \cdot p (1 - p)} \\ = \frac{0.9 5^{2} * 1131 * 0.5 (1 - 0.5)}{1134 * \cdot 0.0 5^{2} + 0.9 5^{2} * 0.5 (1 - 0.5)} = 287 \end{matrix}$

where t is the confidence level, p is the prevalence of deaths (estimated at 50% since it is unknown), δ is the margin of error, N is the total population size, and n is sample size.

We conducted interviews between August 24 and September 13, 2020. We used the World Health Organization 2016 Verbal Autopsy Instrument⁸ to elucidate the circumstances of deaths and identify suspected COVID-19 deaths. We used the definition of a suspected case of COVID-19 that the Côte d'Ivoire Institut National d'Hygiene Publique used at the time of our study (Isaac Tiembre, professor and deputy director for Immunization Services, National Institute of Public Hygiene, Côte d'Ivoire, personal communication, April 5, 2020):

A patient with acute respiratory infection (fever and at least one sign/symptom of respiratory disease, eg, cough, shortness of breath), and having no other suspected etiology that could explain the clinical presentation and a history of travel or residence in a zone or territory of Co^te d'Ivoire or a country other than Co^te d'Ivoire reporting local transmission of the COVID-19 disease during the 14 days before onset of symptoms

A patient with acute respiratory infection and who had been in contact with a laboratory-confirmed or probable case of COVID-19 in the last 14 days before the onset of symptoms.

A patient with a severe acute respiratory infection and requiring hospitalization and with no other suspected etiology that could explain the clinical presentation.

Data were entered in an electronic form during interviews, compiled into a Microsoft Windows 10 Excel file, and exported to Stata 14 (StataCorp, College Station, TX) for analysis. We used chi-square statistical methods to test for association with categorical sociodemographic variables among suspected and nonsuspected COVID-19 cases. These variables include age, sex, and professional occupation.

The study was approved by the National Committee of Ethics of Life Sciences and Health of Côte d'Ivoire (Ref: 086-20/MSHP/CNESVS-kp). The protocol of the study was not reviewed by the US Centers for Disease Control and Prevention (CDC) Global Health Center. Because the study was led by the Côte d'Ivoire Institut National d'Hygiene Publique and the role of the CDC scientists was limited to providing technical support, it was not deemed necessary to submit the protocol for review by CDC Global Health Center for research determination.

Results

Of the 306 deaths, our analysis found that 54 (17.6%) were suspected to be community deaths due to COVID-19. Seventeen (31.5%) of all 54 suspected COVID-19 deaths had epidemiological links with a laboratory-confirmed COVID-19 case. Of the total number of deaths, 151 (49.3%) were male (Table 2). Of the 54 people who died of suspected COVID-19, 40 (74.1%; P=.01) were male; 18 (33.3%; P=.34) occurred among persons older than 60 years of age; more than a quarter (n=14, 26.0%) were employed as salespersons, traders, or business owners; and 10 (18.5%; P=.67) were unemployed.

Table 2.

Sociodemographic Characteristics Among Community COVID-19 Suspected Deaths and Other Deaths, 13 Districts of Greater Abidjan, Côte d'Ivoire, March 11 to July 31, 2020

Characteristic	Suspected COVID-19 Deaths (n=54) n (%)	Other Deaths (n=252) n (%)	Total (N=306) n (%)	P Value
Sex
Male	40 (74.1)	111 (44.0)	151 (49.3)	.01
Female	14 (26.0)	141 (56.0)	155 (50.7)
Age group, years
18 to 35	11 (20.3)	37 (14.7)	48 (15.7)
36 to 60	25 (46.3)	101 (40.1)	126 (41.2)
Over 60	18 (33.3)	112 (44.4)	130 (42.5)	.34
Professional categories
Unemployed	10 (18.5)	75 (29.8)	85 (27.8)
Handicraft salespersons, small trader/business owner	14 (26.0)	68 (27.0)	82 (26.8)
Retired	6 (11.1)	32 (12.7)	38 (12.4)
Self employed	6 (11.1)	27 (10.7)	33 (10.8)	.67
Mid-level employee	7 (13.0)	22 (8.7)	29 (9.5)
Day laborer	3 (5.6)	12 (4.8)	15 (4.9)
Government employee with university degree	4 (7.4)	10 (4.0)	14 (4.6)
Farmer	4 (7.4)	6 (2.4)	10 (3.3)

The most common comorbidities among the 54 people who died of suspected COVID-19 included hypertension (n=15, 27.8%; P=.96), diabetes (n=9, 16.7%; P=.63), tuberculosis (n=9, 16.7%; P=<.001), malnutrition (n=7, 13.0%; P=.63), and cancer (n=6, 11.1%; P=.32) (Table 3). Tuberculosis was the only medical condition found to be significantly associated with COVID-19 deaths, with 16.7% in the group of suspected COVID-19 deaths and 4.0% in the group of other deaths (P=<.001). The absence of any reported medical condition was not associated with COVID-19 suspected death.

Table 3.

Medical Conditions Among Community Suspected COVID-19 Deaths and Other Deaths, 13 Districts of Greater Abidjan, Côte d'Ivoire, March 11 to July 31, 2020

Conditions	Suspected COVID-19 Deaths (n=54) n (%)	Other Deaths (n=252) n (%)	Total (N=306) n (%)	P Value
Any medical conditions	35 (64.8)	143 (56.7)	178 (58.2)
Hypertension	15 (27.8)	75 (29.8)	90 (29.4)	.96
Diabetes	9 (16.7)	31 (12.3)	40 (13.1)	.63
Malnutrition	7 (13.0)	27 (10.7)	34 (11.1)	.63
Cancer	6 (11.1)	17 (6.7)	23 (7.5)	.32
Tuberculosis	9 (16.7)	10 (4.0)	19 (6.2)	<.001
Asthma	3 (5.6)	13 (5.2)	16 (5.2)	.41
Kidney disease	4 (7.4)	9 (3.6)	13 (4.2)	.43
Hepatitis	2 (3.7)	9 (3.6)	11 (3.6)	.97
HIV/AIDS	2 (3.7)	9 (3.6)	11 (3.6)	.97
Epilepsy	1 (1.9)	5 (2.0)	6 (2.0)	.58
No medical conditions	19 (35.2)	99 (39.3)	118 (38.6)	.23

Note: Persons with more than 1 medical condition are counted more than once.

Discussion

Using community mortality data sources, we found 54 suspected COVID-19 deaths that were not previously reported as COVID-19-related. Assuming that the rate of suspected COVID-19 deaths of 17% in our sample of 306 deaths also holds for all 1,131 deaths, we estimate an additional 192 suspected deaths that could be attributed to COVID-19 during the study period in the community. Our assessments suggest that the actual number of deaths due to COVID-19 is higher than the official number of laboratory-confirmed COVID-19 deaths reported. The estimated number of 192 excess deaths in the community is likely to be an underestimation. In the early months of the outbreak, we knew very little about COVID-19 and testing capacity was very limited. As a result, many people who had the disease but who were not tested may have died from it and never had their deaths reported. On the other hand, the number of COVID-19 deaths in Sub-Saharan Africa, in general, has always been unclear, with some sources suggesting that the numbers were lower than expected.^9,10

Using verbal autopsy to identify probable COVID-19 deaths, another study found that 20% of the deaths occurring at home were related to COVID-19.⁸ However, that study also included deaths from hospitals, skilled nursing facilities, and other healthcare facilities, and the investigators used an algorithm to determine COVID-19-related deaths.

A comparison of deaths in the study area during the same months (March 11 to July 31) in 2019 using hospital-based death records showed 88% more hospital deaths in 2020 than in 2019. To our knowledge, there were no major events other than the COVID-19 pandemic that could explain such an increase in deaths. This increase supports the hypothesis that the burden of the COVID-19 disease may have been higher than officially recorded during the study period. In any case, the increase in the number of deaths could mean that community COVID-19-associated deaths were underreported. Another explanation for the large increase in deaths reported from 2019 to 2020 could be that hospitals might have felt more pressure to report deaths during the COVID-19 pandemic.

Our study shows that COVID-19 deaths in Côte d'Ivoire were likely higher than what has been officially reported. According to the World Health Organization, the mean excess deaths associated with the COVID-19 pandemic from all causes in Côte d'Ivoire in July 2020 was estimated at 1,552.¹¹

Implications for Community Surveillance and Public Health Events

The lack of a functional system of notification and registration of deaths in all settings with high coverage and completeness in Côte d'Ivoire makes it difficult to determine where and when COVID-19 deaths occur. In many low- and middle-income countries, the coverage and completeness of civil registration of deaths is often below 20%, and hospitals are the main source of cause-of-death data.¹² In some countries, up to 70% of deaths might occur in the community and are therefore out of reach of any likely COVID-19 testing or clinical case detection.¹³ Côte d'Ivoire is no exception. In addition, hospital death data are limited due to low utilization of healthcare services. Under these circumstances, it is challenging to find a reasonable number of cases of COVID-19 deaths and investigate their causes. In our opinion, difficulties accurately quantifying COVID-19-related mortality resulted in an underappreciation of the pandemic's true impact in Côte d'Ivoire and undermined the ability of public health officials to advocate for more urgent interventions by policymakers.

The social and administrative institutions we used as mortality data sources provided critical information to identify the deaths that we investigated and assess their causes. By and large, these institutions could assist in the early detection and reporting of epidemics and unusual events that have health impacts, especially in the urban settings where they are found. In rural areas, where these institutions may not exist, key community leaders or influential figures could fill these roles if they are trained as community health workers. Based on our findings, these individuals could support community-based disease surveillance and early detection of epidemics in Côte d'Ivoire.¹³ Engaging these community institutions and community leaders to report deaths outside of healthcare settings could help expand data sources to build a more robust vital statistics database. When community organizations are connected to the surveillance systems and can use formal channels to systematically report unusual deaths when they occur, they create an opportunity for public health authorities to conduct community investigations of suspicious diseases or outbreaks and respond in a timely manner before they become a large-scale outbreak. In the case of COVID-19, early investigations of suspected deaths could have resulted in early identification of contacts, testing, and isolation. The use of community data sources is strengthened when they occur within a legal framework, are integrated into the health system, and complement it. The value of the data from these sources would have to be demonstrated prior to being accepted by health and response authorities, especially among epidemiologists and medical staff who make the final determinations of outbreaks and causes of death.

The government of Côte d'Ivoire has responded favorably to the study and its implications. As a result, technical experts are considering ways to implement this approach as an operational early warning system about epidemics and other unusual public health events that could include community deaths in the study districts.

The 54 COVID-19 deaths we identified with the help of community informants were not officially identified or reported as being related to COVID-19. They may be considered sources of undetected community transmission before their deaths and may have contributed to the spread of the disease.

Limitations

Our study has several limitations. Our method revealed only 3 deaths in persons younger than 18 years of age, indicating that the estimate could be applied only to adults. Contacts of deceased persons who did not participate in the survey may have reported different answers to the survey questions. Also, because of the stigma associated with the disease in Côte d'Ivoire, contacts of deceased persons who had COVID-19-like signs and/or symptoms may have deliberately refused to participate, which could have generated more suspected COVID-19 deaths. Moreover, the absence of randomization may impact the extrapolation of the proportion of suspected COVID-19 deaths in our sample to the entire community.

For reasons that were not explained to us, we were forbidden by police to use any information related to the names of institutions that participated in the study, their number, or the number of deaths they reported.

Lastly, because we started interviewing the contacts of deceased persons in late August 2020, it was impossible to conduct postmortem laboratory tests to confirm COVID-19 as the cause of death among our sample.

Conclusion

Our community mortality survey shows that an estimated 192 (17%) deaths outside of hospital settings were attributable to COVID-19 between March 11 and July 31, 2020, in the 13 districts of the greater Abidjan area. During the same period, only 102 COVID-19 deaths were officially reported in the same survey area. Based on our findings, the official number of COVID-19-related deaths was clearly an undercount.

In Côte d'Ivoire, institutions related to death services can be used to identify deaths not reported by outside health facilities and could be useful sources of mortality surveillance, which send alerts about an unusual number of deaths or unexplained deaths. Similarly, settings with limited death notifications and registration capacity may want to rely more on these institutions to obtain a more accurate estimation of deaths.

Acknowledgments

The authors want to thank the family members who volunteered to respond to our questions while still mourning the loss of their loved ones, as well as religious, traditional, and administrative officials who collaborated in the data collection. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the US Centers for Disease Control and Prevention.

References

Ivory Coast confirms first case of coronavirus: health ministry. Reuters. March 11, 2020. Accessed June 14, 2023. https://www.reuters.com/article/us-health-coronavirus-ivorycoast/ivory-coast-confirms-first-case-of-coronavirus-health-ministry-idUSKBN20Y2BT

United Nations Children's Fund. Côte d'Ivoire: COVID-19 situation report - #15. Reporting period: 18 to 31 July 2020. Accessed June 8, 2023. https://reliefweb.int/report/c-te-divoire/cote-d-ivoire-covid-19-situation-report-15-reporting-period-18-31-july-2020

World Health Organization Côte d'Ivoire Country Office. Base de données de surveillance des cas de COVID 19 de Côte d'Ivoire. March-July 2020.

Main courante des commissariats de police d'Abidjan de Mars à juillet 2020. Capital. https://www.capital.fr/votre-argent/deposer-une-main-courante-1343304

Ministère de la Santé et de la Lutte Contre le SIDA (MSLS). République de Côte d'Ivoire. Politique Nationale de Promotion de la Médecine Traditionnelle. Abidjan: MSLS; 2014. Accessed August 19, 2022. https://www.sante.gouv.ci/userfiles/file/POLITIQUE%20NATIONALE%20DE%20PROMOTION%20DE%20LA%20MT.pdf

The Word Bank. Population, total - Cote d'Ivoire. Accessed August 19, 2022. https://data.worldbank.org/indicator/SP.POP.TOTL?locations=CI

Demographic Dividend. Cote d'Ivoire. Accessed August 19, 2022. https://demographicdividend.org/cote-divoire/

World Health Organization (WHO). Verbal Autopsy Standards: The 2016 WHO Verbal Autopsy Instrument. Geneva: WHO; 2016. Accessed August 19, 2022. https://www.who.int/publications/m/item/verbal-autopsy-standards-the-2016-who-verbal-autopsy-instrument

Nguimkeu

, Tadadjeu

. Why is the number of COVID-19 cases lower than expected in Sub-Saharan Africa? A cross-sectional analysis of the role of demographic and geographic factors. World Dev. 2021; 138:105251.

10.

Adams

, MacKenzie

, Amegah

, et al. The conundrum of low COVID-19 mortality burden in sub-Saharan Africa: myth or reality?. Glob Health Sci Pract. 2021; 9(3):433-443.

11.

World Health Organization. Global excess deaths associated with COVID-19 (modelled estimates). Updated May 5, 2022. Accessed March 3, 2023. https://www.who.int/data/sets/global-excess-deaths-associated-with-covid-19-modelled-estimates

12.

Vital

Strategies

, World Health

Organization

. Revealing the Toll of COVID-19: A Technical Package for Rapid Mortality Surveillance and Epidemic Response. New York: Vital Strategies; 2020. Accessed August 19, 2022. https://www.vitalstrategies.org/wp-content/uploads/Vital-Strategies-Revealing-the-Toll-of-COVID-19.pdf

13.

Ministère de la Santé, de l'Hygiène Publique et de la Couverture Maladie Universelle. Données journalières sur la COVID-19. Institut National d'Hygiène Publique. July 31, 2020. Accessed August 19, 2020.