Undercounting mortality in the Covid-19 pandemic
This comment concerns the most common reasons mortality has been undercounted during the novel coronavirus pandemic, to raise awareness in the general public and help researchers to analyze mortality data with caution. Different countries’ different reporting systems may have different problems; no data are flawless. This comment offers six general issues that apply to most countries, before proceeding to focus on the peculiarities of Covid-19.
- Vital statistics are not universal: In building their census and vital registration systems, national statistical offices must gather information for the entire population in a timely manner, satisfying statistical confidentiality. Covering an entire territory at all times requires enormous financial and technical effort by governments. The resources for such efforts are unavailable in most developing nations or in rural or remote places in most other countries, as a result of which statistics are never really universal. Not to mention extreme cases, like Bolivia or Haiti, where only censuses are available, and no other records of the vital registration system exist.1
- A broken system from the start: Birth certificates are the starting point of the vital registration system; anyone without a birth certificate cannot have a death certificate. Unregistered births are still common in developing nations and for older cohorts around the world. UNICEF reports that only one quarter of the world’s population live in countries with complete vital records.
- Mortality definitions are not comparable across countries: Age, sex, and causes of death are all part of such definitions. For instance, differences in the concept of maternal mortality, one of the targets of the United Nations’ Sustainable Development Goal 3, led to the creation of the Maternal Mortality Estimation Inter-Agency Group, which produced a comparable measure only in 2014.2
- “Hard-to-reach populations”: Accounting for mortality in some populations is almost impossible: residents of isolated places, undocumented migrants, workers in illegal activities, victims of displacement, refugees, those persecuted by social or legal norms, or people in other complicated situations who do not want to be reached by the statistical system for fear of persecution or stigmatization.
- No demand for a death certificate means no accountability: If the deceased leaves no bequest, or if the recipients have no rights to manage those assets, there is no pressure or need for a death certificate. This was the rule not so long ago for many women and remains so in some places around the world. Thus, it is not surprising that in most Latin American countries, women are still more undercounted than men.3
- Lack of trained medical personnel: Most death certificates require a Medical Doctor’s statement, but medical coverage is not universal. Even when and where available, doctors sometimes do not fill in all the required information, perhaps because doing so is a burden for them or simply because they do not value statistical collection. By 2014 near 90% of death certificates in most Latin American countries were filled out by an MD, but near 15% of causes of death are not well defined.
Since the World Health Organization declared a pandemic on 11 March 2020, Covid-19 has aggravated the problems of undercounting mortality for the following reasons.
- Pandemic conditions: An unexpected, increased, and sudden demand for medical services puts pressure on the system of death registration, and in some circumstances may cause it to collapse. Doctors are too busy saving lives to fill out forms, while statistical and registry offices may have had to close temporarily as part of pandemic-mitigation measures or are overloaded as well.
- An unknown cause: Because Covid-19 is a new disease, early cases were, most likely, counted as another respiratory disease. Only once clearly defined as an official cause of death will systems start counting deaths from Covid-19 properly.
- Testing for Covid-19: Some official counts of Covid-19 deaths include only those who tested positive for the disease, so anyone without a test may be excluded. And, as is widely known, testing is not universal – particularly in countries with insufficient resources to test on a massive scale. In addition, Covid-19 tests are new and imperfect. False-negative rates could be at least 15%,4 so some false-negative patients may die without being recorded as having Covid-19. Countries with too many cases and not enough tests are already accounting for “probable” Covid-19 deaths, such as in Ecuador, where those with certain symptoms are included in the figures even without a test result.
- Underlying causes of death: Many countries record both main and underlying cause(s) of death. Demographers use them all. Currently, most records account only for the main cause of death, leaving out those whose underlying cause may have been Covid-19.
- Comorbidity: Another source of underreporting is asymptomatic patients who develop a complication related to Covid-19 in another, comorbid disease (i.e., diabetes, hypertension, acute respiratory diseases) and die. Asymptomatic patients who were never tested could add as many as 18% to 33% additional infected cases.5
- Location of death: Some countries measure Covid-19 deaths primarily at medical facilities and do not always cover deaths at home or in other facilities, such as nursing homes. As the largest fraction of Covid-19 deaths is in the elderly, this will present an additional challenge of undercounting in some countries, such as is the case in Spain, where the Ministry of Health is recording only 75% of total deaths in nursing homes.
- Indirect effects: While some causes of death may decline during the pandemic, such as homicides and traffic accidents, deaths from other causes could increase because the medical system is overrun. Anyone in need of emergency care may receive a below-average response if their hospital is swamped with Covid-19 cases. People may also avoid attending hospital for the care they need for fear of catching Covid-19, perhaps leading to deaths from illnesses that could be easily treated.
- Political reasons: There is an element of politics in all decision about what data to collect. Data collection during a pandemic is no exception. A government may be in no rush to produce mortality data if it shows them to be doing a poor job at handling the pandemic, while others may avoid publishing data to try to minimize the scale of the problem. Take, for instance, the case of Brazil, where the Ministry of Health stopped publishing Covid-19 data on 6 June. Publication was resumed a few days later after a petition to the country’s Supreme Court.
In summary, death records are always imperfect. Under pandemic conditions, all typical issues with vital records are exacerbated, potentially leading to worse reporting for all causes of death, including Covid-19. Therefore, current data should be read with caution. Once the pandemic passes, demographic techniques will allow us to estimate an appropriate correction, as mortality under-registration has been well-studied for more than six decades.6 Hopefully, this can be done soon.
About the author
B. Piedad Urdinola is associate professor in the Department of Statistics, Universidad Nacional de Colombia, Bogotá.
- CELADE- Latin-American and the Caribbean Center for Demographic Studies. Demographic Observatory of Latin America 2017: Life tables. 2017
- World Health Organization. Trends in maternal mortality: 1990 to 2013: estimates by WHO, UNICEF, UNFPA, The World Bank and the United Nations Population Division; 2014.
- Urdinola, B. Piedad, Francisco Torres Avilés, y Jairo Alexander Velasco. “The Homicide Atlas in Colombia: Contagion and Under-Registration for Small Areas.”. 2017.Cuadernos de Geografía: Revista Colombiana de Geografía 26 (1): 101-118. doi: 10.15446/rcdg.v26n1.55429.
- LI, Yafang, et al. Stability Issues of RT‐PCR Testing of SARS‐CoV‐2 for Hospitalized Patients Clinically Diagnosed with COVID‐19. 2020. Journal of Medical Virology.
- Mizumoto, K., Kagaya, K., Zarebski, A., & Chowell, G. Estimating the asymptomatic proportion of coronavirus disease 2019 (COVID-19) cases on board the Diamond Princess cruise ship, Yokohama, Japan. 2020. Eurosurveillance, 25(10).
- Moultrie, T. A., Dorrington, R. E., Hill, A. G., Hill, K., Timæus, I. M., & Zaba, B. Tools for demographic estimation. International Union for the Scientific Study of Population. 2013