EVOLVING KNOWLEDGE FOR UNDERSTANDING & STOPPING PANDEMIC

Does dose (viral load) matter? Certainly, but uncertainties remain. Learn more about Risk Analysis and SRA.

My previous post on the COVID-19 pandemic raised great questions from readers about lower severity of illness reported in areas outside hotspots or epicenters of the pandemic. The questions arose from the figure below from the World Health Organization on the dynamics of crude fatality ratios (CFRs) over space and time in China. CFRs steadily declined for the epicenter (Wuhan) from Jan 1st to Feb 10th. CFRs were considerably lower further from the epicenter.

One reader posed these three hypotheses besides identified risk factors (age, co-morbidities, and limitations of ICU beds or health care capacity). Severe cases might:

contact higher concentrations or titers of the virus in the hotspots (leading to higher inhaled doses or viral loads).
contract virus from multiple exposures or encounters with cases in hotspots (leading to cumulative doses and higher virulence than single doses).
are exposed to mutated, more virulent strains of the virus in hotspots.

Some data already exists to support one hypothesis, from my perspective as a longtime member of the Society for Risk Analysis (SRA) and a past president and current officer of the SRA Dose Response Specialty Group (DRSG). In a blog last fall, I described some of SRA’s specialty groups (SGs), and many of the 16 SGs have expertise to contribute to risk analysis for the pandemic. I will highlight DRSG here, but see appended list of the SGs with links to their websites and a contact person for broader perspectives of the deep and varried interdisciplinary expertise of SRA risk practitioners.

WHAT DO WE KNOW ABOUT DOSE-RESPONSE?

Assessing risk requires data on the doses of a hazard (for example, a chemical, physical hazard, or microbial pathogen like the novel corona virus) that cause illness or harm (adverse responses) in humans (or other organisms or the environment). DRSG members examine scientific data and generate dose-response models to predict the chance or likelihood of adverse responses and uncertainties associated with the response from the data available. Multiple studies on pandemic cases have already been published this year documenting groups of cases or cohorts admitted to hospitals. Most studies reported use of testing methods that are qualitative, reporting presence/absence of the virus from nasal swabs or nasopharyngeal swabs of the throat at the back of the nasal cavity. A few studies used quantitative methods that reported estimates of the viral loads (dose of virus) over the course of illness. These studies suggest that dose IS IMPORTANT in assessing and managing risk of pandemic illness.

Some general principles for describing dose-response relationships are illustrated in a recent study documenting dose-dependencies and time-dependencies for tularemia, a bacterial disease also associated with fever and flu-like symptoms in volunteers administered well-characterized aerosol doses of the pathogen (McClellan et al., 2017). As dose increases:

the chance or likelihood of illness increases;
the severity of illness increases; and
the time before onset of symptoms decreases. (In other words, high doses make people sick sooner than low doses).

What was dramatically demonstrated in a hospital in Nanchang, China (Liu et al., 2020) is that these principles are consistent with emerging data quantifying the novel corona virus load estimated for a group of 76 cases. The estimated dose (initial viral load upon admission) was significantly correlated with severity of disease. Nasopharyngeal swabs were analyzed from 46 cases with mild illness and 30 cases with severe illness. Significantly higher mean viral loads were estimated for the 30 severe cases compared to those from the 46 mild cases (p<0.00001, Mann-Whitney test).

However, higher severity observed in local/regional hot spots with high densities of ill people could be due to the alternative hypotheses: 2) cumulative exposures to multiple repeated doses; or 3) mutation of the virus to higher virulence as it was transmitted from the epicenter of the pandemic.

Regarding the second hypothesis that multiple doses in hotspots cause more severe illness than single doses in more distant regions, I am not aware of definitive supporting evidence for corona viruses at present. It is reasonable to assume that exposures are more frequent near the epicenter. The statistical analysis documented in a recent study of repeated doses of a spore-forming bacterial pathogen (Coleman et al., 2017) may be relevant for designing experiments to test for effects of repeated doses of the novel corona virus in animal or tissue culture models. However, it also seems likely that care givers and medical professionals are exposed to higher doses and possibly repeated higher doses that may influence the likelihood and severity of disease, particularly if protective gear is not available or is not put on (and taken off) following guidance to minimize exposures.

Regarding the third hypothesis that the virus mutates to higher virulence in hotspots, I am not aware of definitive evidence for corona viruses at present. However, it seems unlikely at this point in study of this novel corona virus that higher fatality rates are linked to mutations in the virus based on a recent cohort study of mild and severe cases in Hong Kong (To et al., 2020) and this study on genomic variance (Ceraolo and Giorgi, 2020).

Other recent studies also extend knowledge for predicting the likelihood and severity of disease for the novel corona virus. A study conducted in Beijing (Yu et al., 2020) compared methods for qualitative (presence/absence) detection by reverse transcriptase polymerase chain reaction (RT-PCR) with quantitative estimation of viral load (droplet digital PCR or ddPCR) simultaneously for 95 cases over the full course of disease via samples from the upper and lower respiratory tract, blood, and urine.

Pan and colleagues (2020) reported results of viral loads in throat swabs and/or sputum (indicative of lower respiratory tract) from 82 cases in China, including serial samples of throat swabs, sputum, urine, and stool from 2 cases over the full course of illness and recovery by day 15. Variability in viral load was detected between specimens, with reporting of simple correlative analysis.

Two recent studies conducted in Germany provided the following new findings.

• Viral replication was limited to throat and lung, not blood, urine, and stool from 9 mild cases (Wolfel 2020)

• Person-to-person transmission efficiency appeared to decrease from the original index (primary) case to subsequent contacts (secondary and tertiary cases) in company and household clusters (Bohmer 2020)

I also encourage readers to take a look at this New Yorker article written by scientist and prize-winning author Siddhartha Mukherjee. (How Does the Coronavirus Behave Inside a Patient). The importance of understanding dose-response relationships was also noted by Dr. Mukherjee.

WHAT DON’T WE KNOW?

Great uncertainties remain, and multiple recent studies have identified crucial gaps in knowledge about SARS-CoV-2 that limit our ability for early identification of cases more likely to develop severe illness in order to reduce clinical severity and mortality. These studies (Huang et al., 2020; Joynt and Wu, 2020; Monteiro et al., 2020; Wong et al., 2020) point to critical knowledge gaps about detection, infectivity, and modes of transmission. Each of the gaps highlighted below are dependent on dose, either directly or indirectly.

Dectection of the presence of viral RNA fragments in human clinical samples by PCR methods for one or two viral genes does not demonstrate replication or infectivity of the virus in that sample. For example, it is unclear if infective virus is present in stool or if the fecal-oral pathway is causing human illness in this pandemic.
Feasible modes of transmission are poorly characterized. Transmission by aerosolized droplets and contact with cases are well documented for corona viruses, but the contribution of alternative modes of transmission (e.g., from asymptomatic cases, contact with surfaces, fecal-oral) is unknown.
Potential biomarkers for disease severity are needed for early identification of cases more likely to progress to severe illness due to the broad clinical spectrum and non-specific initial symptoms. High initial dose early in disease onset seems a useful biomarker based on time series studies available.

WHY MIGHT DOSE MATTER?

Deeper understanding of the relationships between dose and likelihood, severity, and timing of adverse affects could support future evidence-based decisions about triage and therapeutic options. For example, quantitation of viral load would provide an early indicator of potential disease severity for allocation of limited supplies of convalescent serum to patients with high viral loads early in their disease cycle, prior to development of systemic potentially fatal complications.

The results of the Liu study are promising for supporting development of more rapid and streamlined triage procedures that could reduce the severity and mortality for future cases. While more data are needed to verify the results of the Liu study and generate a comprehensive statistically sound model predicting the magnitude and timing of adverse affects, the urgency of the need for biomarkers is extreme, considering the high fatality rates in major metropolitan areas around the world described below.

Expansion of the available data for dose-response assessment could lead to improvements in triage for new cases that reduce crude fatality ratios for the pandemic. Liu and colleagues noted that quantitation of viral load may be a useful biomarker for severe disease that could support early diagnosis for potential needs for ICU and critical care. If this study is generalizable to other populations around the world, medical staff could screen patients on admission for viral load, and those with high load might be provided additional therapeutics (e.g., convalescent serum) to more quickly reduce the viral load and lessen the severity of symptoms.

Can public health departments apply such quantitative methods tomorrow? Unlikely, without additional investment and training. Additional data are needed from hospitals and research organizations around the world for other cohorts of cases. The SRA community has the interdisciplinary/transdisciplinary expertise to assist in developing and communicating about reliable models to predict viral doses causing mild and severe disease for use in triage of cases with new or recurring infections. Tools to support earlier diagnosis of cases more likely to require ICU and critical care will be important to manage increasingly limited medical and testing resources in areas around the world where transmission to new cases has not yet dissipated.

PANDEMIC DYNAMICS AFTER MID-APRIL?

As of April 9th, the World Health Organization (WHO) Situation Report listed 1,436,198 confirmed cases from the pandemic, including 85,522 deaths (estimated overall crude fatality ratio approximately 6%). For the Region of the Americas, the report lists 454,710 confirmed cases and 14,775 deaths (estimated overall crude fatality ratio approximately 3%). The continuing toll on people’s lives plus on health care systems and on economic, financial, social, and religious structures is ‘certainly’ not good news, but knowledge is expanding. See the WHO Situation Report for guidelines for religious leaders and faith-based communities regarding the pandemic and other links to more detailed information.

The Corona Virus Resource Center at John’s Hopkins provides data and tools relevant to the pandemic, including data on numbers and rates of confirmed cases, hospitalizations, fatalities, and recoveries. The figure below was generated on April 17 using the Mortality Analyses tool.

https://coronavirus.jhu.edu/data/mortality

One of the seven critical issues that require ‘concerted coordinated attention and action’ identified by Wong and colleagues (2020) from cases and responses in Singapore is relevant to SRA and the disciplines of analysis of risk perceptions and risk communications. The authors note that the ‘medical community needs to collectively find better ways to communicate and engage the public … understandably anxious’. How might these global needs be addressed? SRA is poised to fill this void through the COVID CONVERSATIONS ON RISK.

SRA’S FREE WEBINAR SERIES ON COVID CONVERSATIONS ON RISK

The magnitude of the pandemic prompted SRA leaders to organize a free a bi-weekly webinar series, COVID CONVERSATIONS ON RISK. The series will highlight diverse expertise that the SRA community can bring to bear on the difficult challenges of these trying and unusual times. In particular, multidisciplinary, risk-based perspectives are essential to address the complexities and uncertainties of the pandemic and help provide advice and guidance on evidence-based risk management actions.

On April 9th, SRA President Seth Guikema welcoming two internationally known panelists to this week’s webinar, interdisciplinary risk analyst Dr. Pia-Johanna Schweizer (Institute for Advanced Sustainability Studies, Potsdam, Germany) and Professor of Management Science Gilberto Monitbeller (Loughborough University, UK).

An extremely useful illustration was introduced by Dr. Schweizer, the ancient fable of the Blind Men and the Elephant, as applied to considering the pandemic as a systemic risk problem (see slide below). The fable has been viewed as a warning that partial perspectives of a whole (elephant, system, pandemic) may lead to mistaken images and overreaching interpretations of reality that could cause harm without openness to deliberation and analysis of the complexities.

For some SRA members, the warning of the fable could be interpreted as the need for exercise of Analytic-Deliberative Process (NRC, 1996, Understanding Risk: Informing Decisions in a Democratic Society) to build useful knowledge of the integrated whole, incorporating the different partial perceptions and other relevant data.

Analytic-Deliberative Process was described by a committee convened by the US National Academy of Sciences, the National Academy of Engineering, the Institute of Medicine, and the National Research Council as follows: characterization of 'a potentially hazardous situation in as accurate, thorough, and decision-relevant a manner as possible, addressing the significant concerns of the interested and affected parties, and to make this information understandable and accessible to public officials and to the parties' (stakeholders).

Clearly, we have never seen an ‘elephant’ like this pandemic, and the speakers describe successes, failures, and the need for significant improvements in Risk Communication and Risk Management from their perspectives of systemic risk and portifolio decision analysis.

I highly recommend that you click on the video below for COVID Conversations on Risk (Episode 1), a fascinating and challenging hour of dialogue with these exceptional analysts who presented informative, thought-provoking slides and responded to questions from the SRA audience with unique perspectives from their extensive experience with complex interdisciplinary/transdisciplinary modeling that build discrete scenarios to support decisions about tough tradeoffs. A brief summary of the dialogue is provided after the video.

Brief Summary of Responses to Questions from SRA Audience

The panelists responded with complementary perspectives to many questions, particularly about the need for clarity (transparency in risk jargon) and consistency in Risk Communication when new information about the virus and the pandemic is expanding each day from many sources around the world. Both panelists agreed that one key to effective Risk Communication is that communicating absolute certainties is a grave mistake. Both the virus and public perceptions of the risk of illness/death are continuously evolving, and the media has sometimes blurred the emerging scientific evidence, contributing to escalation of local outbreaks by reporting premature findings with greater confidence than deserved at the time.

Neither panelist was aware of development of a model to date that broadly integrates risk assessment and decision analysis (e.g., systemic risk, simultaneously modeling risk and recovery for public health, security, social, and economic/financial sectors), with feasible scenario modeling of alternative response strategies or policies. Such modeling is very difficult to design and test in this climate of ‘deep uncertainty’ with so many factors and dependencies to continuously update and model across disciplines and around the world. Science seems to progresses incrementally, with fine tuning of ideas and theoretical models based on peer review and additional data and deliberation amongst the professional community. However, when faced with ‘deep uncertainty‘ due to conflicting models and preliminary or ambiguous data, the very strong human temptation is to prematurely select the model predicting the most desirable outcomes rather than to support deliberative processes that provide more reliable decision support. The speakers offered these dramatically different situations to inform future risk-based decision analysis.

In the UK, these factors derailed Risk Communication efforts as described by Prof. Monitbeller. Major problem was the premature selection of a novel theoretical model, perhaps a ‘best-case’ model predicting that the outbreak was waning and not the serious public health problem predicted by the competing model that predicting undesirable outcomes. The decision makers did not seem to understand at the time that the ‘best-case’ model that predicted the desired results was based on preliminary data that oversimplified and mischaracterized real life and death scenarios. This selection bias and faulty risk analysis in the face of ‘deep uncertainties’ likely led to continued escalation of transmission to new cases in the UK.

In Germany, Risk Communication was exceptional as described by Dr. Schweizer, with daily press conferences including both political leaders and scientists who were trained to share evidence as knowledge expanded, with care to openly and fairly represent uncertainties. Daily press conferences proved an essential service to stopping the outbreak in Germany, communicating actual and perceived risks and uncertainties clearly (transparently) with the public and continuously updating and revising what they thought they knew the day as knowledge expanded. Dr. Schweizer also described a ‘miracle’: the Robert Koch Institut had test kits prepared for broad scale, comprehensive testing and research BEFORE the outbreak hit Germany. The Chancellor also clearly communicated that strict countermeasures were needed and built trust, emphasizing that social distancing was a common societal responsibility that depended on compliance of each and every citizen to stop disease transmission. The public trusted and followed this careful Risk Communication and contributed to the successful end of disease transmission in Germany.

To sum up, why did these esteemed risk analysts spend nearly an hour in dialogue about uncertainty in the SRA webinar? Because uncertainty must be framed for all stakeholders to understand, based on data (evidence) and models (describing and extrapolating from the evidence) in order to support reasoned deliberative risk management and evidence-based decision support for selecting, implementing, and evaluating alternative strategies to stop the pandemic.

For unique perspective from SRA Past President Rae Zimmerman (NYU Wanger) on her experience as a risk analyst in the US epicenter of the pandemic, click on the link to the first Let’s Talk Risk podcast of her 30-minute conversation with current SRA President Seth Guikema. Professor Zimmerman also mentioned successes in Germany’s response to the pandemic and the urgency for applying deep knowledge of the principles of risk analysis, particularly Risk Communication and managing competing risks (e.g., personal and societal) and conflicting priorities as the pandemic continues. She notes that NY has survived many catastrophes and disasters in the past, but none quite as broad or disruptive of so many systems, including public health, economics, education, social equity, transportation) for so long. Her third Big Research Question addressed Risk Communication, how to encourage people under stress to make positive behavioral adjustments to adapt to our new reality. Professor Zimmerman reminded us to celebrate our first responders and health care professionals during this crisis, as NY City residents do (#ClapBecauseWeCare) and Italians do, singing from their rooftops and balconies.

BOTTOM LINE

Have hope, but play it smart by following guidance from WHO (DO THE FIVE, below), CDC (Use of Cloth Face Coverings to Help Slow the Spread of COVID-19), and other local public health organizations. Even though WHO recommends 3 meters for social distancing and CDC recommends 6 feet, social distancing and quarantine are effective measures to reduce the likelihood of both exposure and illness, particularly severe illness.

Feel free to ask questions in the Comment section below and I will respond within a few days. My next blog will be posted after the second episode of the SRA biweekly COVID Conversations on Risk webinar series on April 23rd. The current state of the pandemic as of April 9th is illustrated below from the WHO Situation Report 80 (Figure 1).