Revisiting the initial COVID-19 pandemic projections

Early projections of the COVID-19 pandemic prompted federal governments to action. One critical report, published on March 16, 2020, received international attention when it predicted 2 200 000 deaths in the USA and 510 000 deaths in the UK without some kind of coordinated pandemic response.¹ This information became foundational in decisions to implement physical distancing and adherence to other public health measures because it established the upper boundary for any worst-case scenarios.

However, the authors derived these projections from best available estimates at the time. The evolving nature of empirical knowledge about COVID-19 provides current estimates with more accurate information than what would have been available merely weeks after first discovery of the virus—plus the benefit of hindsight. For example, asymptomatic transmission has been said to be the Achilles' heel of public health strategies to control the pandemic,² and several factors about asymptomatic cases remained uncertain during the early days. The report assumed that asymptomatic individuals were 50% as infectious as symptomatic cases,¹ whereas the current US Centers for Disease Control and Prevention (CDC) estimates suggest a 75% infectiousness rate for asymptomatic individuals.³ A more important difference is the infection fatality ratio as originally projected in the Imperial College London (London, UK) report¹ versus current estimations. A high ratio of asymptomatic individuals might have inflated the perceived mortality of the disease given the limited testing supplies and attention to symptomatic cases.

These analyses explore the original projections in light of this new information, which might be especially important when comparing the efficacy of any public health measures undertaken to prevent loss of life. Three sources of information were referenced for these assumptions: the original Imperial College London report,¹ the best estimates of current CDC pandemic planning scenarios,³ and a seroprevalence study covering 45 countries.⁴ The CDC estimates largely come from a separate study,⁵ but the values are used here as presented by available CDC information. Population information was collected from the 2010 US census and 2011 UK census as the most complete and recent government surveys of those populations. Actual calculations used the most refined infection fatality ratio and age groups available, although data presented in the table truncate these values to the CDC limits for ease of presentation after adjusting for age and population differences. The percentage of population infected, 81%, comes from the original estimation in the Imperial College London report.

Table.

Projected deaths based on age-adjusted infection fatality ratios for the USA and UK

		Age 0–19 years	Age 20–49 years	Age 50–69 years	Age >70 years	Total
USA
Population		83 267 556	126 429 144	71 216 117	27 832 721	308 745 538
Projected deaths
	Imperial College London report	2733	89 358	725 232	1 532 044	2 349 367
	CDC estimations	2023	20 482	288 425	1 217 403	1 528 333
	Seroprevalence	1359	48 638	259 235	1 070 381	1 379 612
UK
Population		15 098 000	26 193 000	14 533 000	7 359 000	63 183 000
Projected deaths
	Imperial College London report	493	18 744	159 069	402 318	580 624
	CDC estimations	367	4243	58 859	321 883	385 351
	Seroprevalence	250	10 184	56 653	279 548	346 637

Data are from the initial Imperial College London report¹ and two more recent parameter estimations from the CDC³ and a retrospective study with data from 45 countries (seroprevalence).⁴ CDC=US Centers for Disease Control and Prevention.

This simplified assessment arrives at a comparable approximation of the original report—2 349 367 projected deaths in the USA and 580 624 deaths in the UK. Applying age-adjusted infection fatality ratio rates to the census population values reveals a striking difference from CDC estimates and seroprevalence reporting. CDC estimates place total deaths at 1 528 333 in the USA and 385 351 deaths in the UK, whereas seroprevalence estimates total deaths at 1 379 612 in the USA and 346 637 deaths in the UK. For the US estimates, the differences produce a 54–70% overestimation of approximately 1 million deaths. For the UK estimates, the differences produce a 51–68% overestimation of approximately 200 000 deaths.

Such overestimations remind us of several lessons learned over the course of the pandemic. First, the initial projections were never going to be 100% accurate with a novel coronavirus. Initial projections built worst-case scenarios that would never happen as a means of spurring leadership into action. This upper boundary of possibility then demonstrates a functional value of modelling efforts for unmitigated pandemic progression. Second, asymptomatic cases inflated perceived mortality ratios in addition to complicating any containment challenges. Third, consensus predictions underscore the value of public health coordination—especially early in a novel outbreak. When information is scarce, information sharing from multiple sources becomes crucial to attaining the clearest prediction possible. Last, in democracies, these public health crises will be politicised, and it is incumbent upon guardians of the public trust in health-care institutions and services to remain apolitical—to remain focused on scientific knowledge and the needs of public health, just as the US Department of Defense remains apolitical and focused on the needs of national defence.

Ultimately, the relative value of mask wearing and physical distancing, and the economic consequences of lockdowns will be analysed retrospectively. These evaluations will use worst-case scenarios of unmitigated progression as the measuring stick to describe the merit of different public health interventions. Still, initial projections were commendable efforts that brought about public action despite more than 2 million deaths in the USA and more than 500 000 deaths in the UK being a significant overestimation.