Surveillance factors

There are a variety of ways in which surveillance factors impact case and mortality data. A greater number of tests conducted will tend to result in a greater number of positive test results. Figures may be adjusted to account for the number of positive cases per one million of the population, for instance, but this will not address the other many variations that are encountered. The frequency of testing will impact the data: more frequent testing, particularly of a previously confirmed positive case, may tend to inflate case incidence data. Where testing is repeatedly and frequently performed on persons who test positive for COVID-19, multiple cases could be attributable to a single infected person (Kanamoto et al., Reference Kanamoto, Tobe, Takazawa and Saito2020).

Mandatory testing regimes are more likely to detect asymptomatic or mildly symptomatic cases than voluntary testing regimes or testing regimes limited to hospitalized cases. This affects overall case and mortality data. There is evidence that COVID-19 mortality rate is negatively associated with test number. For instance, it has been discovered that one additional test per 100 people is associated with an 8% decrease in mortality rate, even after adjusting for various factors (Liang et al. Reference Liang, Tseng, Ho and Wu2020). If countries using mandatory testing measures detect more asymptomatic cases than those using voluntary testing, then the former will probably have a smaller ratio of deaths than the latter.

Additionally, viral test techniques and their correct usage have an important influence. The main tests for COVID-19 include assays for detection of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) nucleic acid or antigen and serological assay for detection of SARS-CoV-2 antibody (Centers for Disease Control and Prevention, 2020). The relative diagnostic accuracy and reliability of those tests will tend to positively or negatively impact case and mortality data. Even within a single testing category, such as reverse transcription polymerase chain reaction (RT-PCR) testing, accuracy and reliability may vary by manufacturer, and different manufacturers often recommend a different number of amplification cycles to determine the presence of SARS-CoV-2. International variations in testing technique, consistent and correct usage and choice of manufacturer will therefore impact case and mortality data. To confound the matter further, insufficient assessment of diagnostic performance has been reported (Axell-House et al., Reference Axell-House, Lavingia, Rafferty, Clark, Amirian and Chiao2020), in addition to reported manufacturer inflation of performance characteristics (Fitzpatrick et al., Reference Fitzpatrick, Pandey, Wells, Sah and Galvani2020).

Classification factors

National differences in classification of case and mortality data are considerable. The definition of cases differs from country to country; some include presumptive cases, whereas others only recognize cases confirmed by RT-PCR testing (Jamison et al., Reference Jamison, Lau, Wu and Xiong2020). Some countries, such as the UK, include asymptomatic positive test results in their case incidence data, with a case defined as a person with one or more specimen tests that proves positive for the presence of SARS-CoV-2, including positive results not confirmed by a laboratory (UK Government, 2021a). Other countries, such as China, have excluded such results from their case incidence data (Cyranoski, Reference Cyranoski2020). This would tend to inflate UK case incidence data relative to the Chinese data.

Similarly, the methodology for compiling mortality data will tend to positively or negatively impact the published mortality rate in a given country. Definitive criteria may include COVID-19 as the sole or principal certified cause of death, COVID-19 mentioned on the death certificate but without a requirement that it be the sole or principal cause of death, or excess mortality adjusted for non-COVID-19 deaths. Excess mortality refers to the number of deaths from all causes during a crisis, above and beyond ‘normal’ conditions. It is a more comprehensive measure of the overall impact of the pandemic on mortality than the confirmed COVID-19 death count alone. It reflects not only confirmed COVID-19 deaths, but also COVID-19 deaths that were misclassified, as well as deaths from other causes that are attributable to the pandemic. However, a notable limitation of excess mortality comparisons is that many countries, especially low- and middle-income countries, do not have statistical agencies and infrastructural frameworks sufficient to reliably monitor and report the incidence of death on a daily, weekly or even monthly basis (Checchi & Roberts, Reference Checchi and Roberts2005).

Classification methodologies vary in terms of causation requirements and can skew the data either positively or negatively. The main UK classification for a COVID-19 death was the death of a person within 28 days of that person’s first positive COVID-19 test (UK Government, 2021b). There was no strict requirement for causation between COVID-19 infection and death, thus including patients already severely or terminally ill with other diseases, patients with significant comorbidities and patients that died of unrelated reasons but who carried SARS-CoV-2, even asymptomatically. This crude measure is notoriously over-broad. It should also be noted that national classification methodologies can change over time (Raleigh, Reference Raleigh2020), and can also vary on a sub-national level (Tsang et al., Reference Tsang, Wu, Lin, Lau, Leung and Cowling2020).

Classification can also be influenced by systematic factors such as national medical standards, practices and consistency. COVID-19 infections may be undiagnosed, misdiagnosed and attributed to other diseases with similar clinical presentation such as influenza (World Health Organization, 2020), such that gaps between observed increases in all-cause mortality and reported COVID-19 mortality may widen (Jamison et al., Reference Jamison, Lau, Wu and Xiong2020). Conversely, suspected COVID-19 cases may be misdiagnosed and attributed to COVID-19 instead of other diseases with similar clinical presentation, such as influenza or pneumonia (Budhram et al., Reference Budhram, Kobza and Mohammed2020), such that reported COVID-19 mortality may be overstated.

Virological factors

SARS-CoV-2 has developed many thousands of different variants (Koyama et al., Reference Koyama, Platt and Parida2020). Among the more widely reported variants thus far are the B.1.1.7 variant first detected in the UK, the B.1.351 variant first detected in South Africa, the B.1.617.2 (commonly known as ‘Delta’) variant first detected in India and the P.1 variant first detected in Brazil and Japan. Yet these variants are reported to be more or less transmissible than other variants of SARS-CoV-2, which probably affects case incidence; and/or be more or less virulent, which probably affects mortality. For example, the B.1.1.7 variant estimated to have first emerged in the UK in September 2020 and likely to be more prevalent in the UK was reportedly 74% more transmissible than the original (probably zoonotic) SARS-CoV-2 (Volz et al., Reference Volz, Mishra, Chand, Barrett, Johnson and Geidelberg2021), which may tend to inflate UK case incidence relative to data in countries where less transmissible variants are prevalent. Moreover, the variants are unequally distributed around the world and their presence and prevalence will vary over time, with the B.1.617.2 and B.1.1.7 variants presently more widely distributed around the world than, for example, the P.1 variant, in turn more widely distributed than more localized variants such as B.1.1.30 or B.1.1.31 (O’Toole et al., Reference O’Toole, Scher, Jackson, McCrone, Colquhoun and Hill2021). The unequal distribution of more or less transmissible and virulent variants will tend to skew case and mortality data, further complicating international comparisons as a basis for NPI decision-making.

Ecological factors

Case and mortality data will tend to be impacted by the local prevalence of comorbidity factors. Obesity, for example, may be one of the major comorbidities associating with COVID-19 mortality, with it being reported that COVID-19 patients with obesity are more severely affected and had a worse clinical outcome than COVID-19 patients without obesity (Yang et al., Reference Yang, Hu and Zhu2021). However, obesity is unequally distributed among national populations. Yet among the adult populations with the highest prevalence rates of obesity are several island nations where NPIs focused on more intrusive border and quarantine measures are potentially easier to implement or yield greater impact on case incidence data, such as Nauru (60.7%), Cook Islands (55.3%), Tuvalu (51%), Marshall Islands (52.4%) and Tonga (45.9%). Conversely, among the adult populations with the lowest prevalence rates of obesity are a number of developing countries where testing and reporting regimes may be less adequately resourced, such as Vietnam (2.1%), Timor-Leste (2.9%), Bangladesh (3.4%), Cambodia (3.5%) and Ethiopia (3.6%) (World Health Organization, 2017). In each situation, potentially adverse factors may be at work on case and mortality data.

The local prevalence of pre-existing immunity to COVID-19 may be unequally distributed among countries. This includes the prevalence of pre-existing T cell reactivity (Doshi, Reference Doshi2020) and other innate resistance or cross-protection from exposure to seasonal coronaviruses (Lourenço et al., Reference Lourenço, Pinotti, Thompson and Gupta2020). Pre-existing immunity to COVID-19 infection will tend to negatively impact case and mortality rates in comparison to rates in countries with relatively less such immunity. This further confounds national comparisons of case and mortality data. Moreover, it is suspected that COVID-19 is seasonal in whole or in part, with factors such as temperature and maximum daily ultraviolet light affecting infection incidence (Merow & Urban, Reference Merow and Urban2020). This would caution against crude comparisons of data between countries with latitudinal, hemispheric and climactic differences, in addition to economic disparities (Broadbent et al., Reference Broadbent, Walker, Chalkidou, Sullivan and Glassman2020).

In addition, infrastructural and demographic factors will be relevant to local epidemiology. Higher population density (Bhadra et al., Reference Bhadra, Mukherjee and Sarkar2021) and connectivity (Hamidi et al., Reference Hamidi, Sabouri and Ewing2020) may tend to exacerbate infection rates and mortality. Older persons appear to be at greater risk of severe COVID-19 and requiring hospitalization if infected (Clark et al., Reference Clark, Jit, Warren-Gash, Guthrie, Wang and Mercer2020), thus tending to inflate case and mortality rates in countries with older populations.

Political factors

Bureaucratic and infrastructural disparities will probably affect the accurate, efficient, complete and timely gathering and reporting of case and mortality data. This is in itself multifactorial, but less developed countries may under-report case and mortality data due to sub-optimal bureaucracies and infrastructures. Differing national, regional and local testing policies may also tend to affect case incidence data, such as the existence and form of mandatory testing regimes and any financial incentives for testing (Government of the Hong Kong Special Administrative Region, 2020; Lee et al., Reference Lee, Kwak and Kim2021). Local social and cultural variations may also affect compliance behaviours and case incidence rates (Dryhurst et al., Reference Dryhurst, Schneider, Kerr, Freeman, Recchia and van der Bles2020).

Polity type may also affect case and mortality data in multifaceted ways. More authoritarian, autocratic and closed regimes may tend towards greater corruption, concealment and statistical manipulation, and less transparency, relative to more democratic, competitive and open regimes. Even though there is evidence of global inclinations towards authoritarianism in the adoption of COVID-19 NPIs (Thomson & Ip, Reference Thomson and Ip2020a), there are diverse governmental models practised throughout the world. The ability of well-resourced authoritarian regimes to mobilize resources and populations quickly to combat the outbreak of infectious diseases is tempered by information politics in such regimes, which can undermine a rapid response to those outbreaks (Kavanagh, Reference Kavanagh2020). Authoritarian and semi-authoritarian governments tend to manipulate information and stifle dissent in an effort to convince populations of their competence (Guriev & Treisman, Reference Guriev and Treisman2019), behaviours that are not conducive to good pandemic management.

Extensive complicating factors therefore arise in relation to the variegated impact of polity type on the gathering and reporting of COVID-19 case and mortality data. It may be that there is a positive relationship between governmental transparency and the severity of reported data, such that the relative reported ‘success’ of certain countries cannot be taken at face value. It is telling that North Korea and Turkmenistan – among the lowest scoring countries in the Global Democracy Index 2020 (Economist, 2021) and the outright lowest scoring countries in the 2021 World Press Freedom Index after Eritrea (Reporters Without Borders, 2021) – continue to report zero COVID-19 cases and zero COVID-19 deaths (World Health Organization, 2021a, 2021b). Statements of the kind that ‘death tolls don’t lie’ (Mahbubani, Reference Mahbubani2020) should be subject to far greater contextual scrutiny in serious scientific and policy discourse.