Coronavirus: The Story of Risk and Resilience (Part 2)
What will be our exit strategy after the “enhanced community quarantine”?
This article (Part 2 of the series Coronavirus: The Story of Risk and Resilience (Part 1, Part 2, Part 3) follows the “Coronavirus: Ang Maso at ang Sayaw,” an adaptation of “Coronovirus: The Hammer and the Dance.” The Filipino adaptation is dedicated to the heroes of our trying times, our valiant “frontliners.” We salute you. This article now focuses on the situation in the Philippines as of April 30, 2020.
Summary. The plot of our story thickens. We have to extinguish the spark to prevent the “fire” of infections spread out to a larger population. Our government starts crafting policy for our exit strategy after this enhanced community quarantine. Over the coming weeks, the decisions our leaders will make based on the best available data, useful models, and expert advice will come under fire. We need clear data-driven strategies and insights as we cannot suppress this COVID fire blindfolded. We thus need metrics for risk and resilience to gain much-needed clarity on our relevant actions to win this war against the invisible adversary.
What is our current situation?
Are we flattening the curve?
Chart 1 shows the epidemic curve in terms of cumulative cases and daily cases of the Philippines in comparison with other countries. As you can see from Chart 1, we have a daily growth rate (r) of COVID-19 cases of 0.023 or 2.3 %, which is equivalent to around 30 days of case doubling time (CDT = ln(2)/r). We can notice that some countries like China, New Zealand, Iceland, and Serbia are effective in suppressing the spread of the first wave of infections as you can see the drop of daily cases in their trajectory. On the other hand, the US and other parts of Europe are starting in the deceleration phase of the epidemics. Likewise, the trend based on the reported confirmed cases in the Philippines suggests an encouraging scenario that our suppression strategy is effective.
However, let’s interpret our data with caution as we may be underreporting these statistics because of limited testing, inefficient bureaucracy, and delayed reporting. We thus need to develop more robust metrics, i.e., innovative ways of collecting and processing data, and measuring how we are progressing in winning this war against an invisible enemy.
Here is my fearless forecast.
The next couple of weeks and months will become more interesting to monitor the development of the story of coronavirus in our country and the whole wide world. These words “risk” and “resilience” will become the most favorite buzzwords of our policymakers, politicians, and experts in this VUCA (volatile, uncertain, complex and ambiguous) world ignited by the spark of COVID-19.
April 30 was the scheduled lifting of the Luzon-wide Enhanced Community Quarantine (ECQ) which started last March 16. On April 24, the government of the Philippines announced that ECQ would be extended from May 1 until May 15 to those areas of “high risk” such as Metro Manila. In the following days, some of the ECQ-designated cities or provinces were also reclassified to General Community Quarantine (GCQ), as shown in Chart 2.
While the heightened presence of uniformed personnel enforcing quarantine protocols are present in places under ECQ, GCQ requires uniformed personnel and quarantine personnel to be present at border points. GCQ also limits people’s movements to accessing necessities and work according to the level of risk the areas are classified. However, the residents in the area under GCQ can go outside from home without quarantine pass. According to IATF, those areas placed under GCQ are those communities deemed as “low-risk” to “moderate-risk” on potential resurgence or second wave of COVID-19 infection.
The Department of Health (DOH) and Inter-agency Task Force(IATF) messaging to the public is that their decisions are evidence-based and updated with the facts on the ground; thanks to the group of volunteer experts and a team of scientists who provide insights based on the simulations of their epidemiological models. For example, the risk level that IATF is using is based on the growth rate of cases or its doubling time (CDT) and the capacity of our healthcare system (e.g., hospitals) to cope up with the increasing confirmed cases of COVID-19 (see Chart 3). Would such decision tool using these indicators good enough to describe the risk level of these LGUs?
As the data becomes more available from the aggressive testing efforts and rigorous contact tracing, the accuracy of the models, including the methods for risk stratification of the communities, will be improved hopefully. Otherwise, we cannot navigate the path to an effective exit plan blindfolded.
In this article (Part 1), I underscore the importance of understanding the concept of risk and resilience in the context of this COVID-19 epidemic. Hundreds or even thousands of excess deaths from COVID-19 are not the only adverse consequence we are anticipating from such risk. Because our hospitals are overburdened by this crisis, some of these deaths may not even be COVID-19 related. Some excess deaths may be caused by common diseases and conditions such as heart attacks, strokes, among others that are not adequately treated during this crisis. We have to consider as well the collateral damage to the economy and society, such as the loss of jobs, livelihood, tax revenues, and even our accustomed freedom. We should also not forget its impact on our psycho-social and mental health. These are key factors to our personal resilience. Indeed, risk and resilience are just two sides of the same coin.
Playing with the “fire” metaphor, these are the questions we have to ask ourselves:
“When will the fire of infections die out? Will there be another spark of “burning” hotspot within and outside the borders of the Luzon island? What is the risk of getting burned from this fire? If we get burned, can we heal promptly and recover from this burn soon? Are we resilient enough to bounce back and be better?”
Are we slowing down the spread of the “fire” of COVID-19 infections?
In Chart 4, we can see those areas that are still under ECQ until May 15, 2020, whereas the rest of the country with reported cases and categorized as low to moderate risk, will be under GCQ. It is noticeable from Chart 4 that the municipality of Pateros has shown a smaller number of cases. Still, because of the geographic proximity with other burning hotspots in NCR, it is also placed under ECQ. On the one hand, IATF recommended that Albay, which is one of the provinces for extended ECQ before, will transition to a more relaxed GCQ on May 1. But some local government units (LGUs) in Albay oppose this plan, and they intend to extend enhanced community quarantine measures in the province considering the high number of probable cases.
It is somehow clear now why some places will still be in ECQ, or some restrictions will be lifted under GCQ on May 1. But the devil is in the detail. Our local government units (LGUs) need data-driven and localized measures for them to cope up and allocate their limited resources properly amid this crisis.
They need as well to be future-focused with clarity on actionable insights as they improve their resilience in the so-called “new normal” of the postCOVID era.
Again, context is everything. Every city or province will be different. Without an understanding of interdependencies and vulnerabilities, the disruption and chronic stresses brought by this pandemic pose not only such risk to public health but also the potential to cascading disruptions to other sectors of our society. Economic decline is already expected, and the long-term stresses can take in many forms.
This epidemic reveals the weakness of our institution and infrastructure. It can also reveal the strengths of our citizens and community to recover and adapt from such disaster. We need to understand the factors that contribute to this disaster risk and our resilience to bounce back to be better prepared in the next emerging threat. Thus, this motivates me and keeps me awake during the night.
How can we gain from this crisis and use the situation as an opportunity to learn more and be prepared for the next threat? This forms the basis of CO-INFORM.
What is CO-INFORM?
CO-INFORM is a framework to develop an epidemic risk index as a tool to provide insights to public health and intervention decisions at the national, regional, and community levels by classifying different areas in the country in terms of the risk level. There are still minimal studies in the Philippines on developing a metric to assess the risk and resilience from exposure to biological hazards considering these epidemics and pandemics are also areas of concern for global health security and disaster risk management.
We need to understand disaster risk and measure its risk level to manage it properly. Risk indices are typically used to describe and communicate the level of risk to the intended audience by summarizing risk in terms of numbers or categories such as words or color. A risk index can also be used to compare among different risk, identify the most severe risk, translate how risk changes over time, and measure the effectiveness of risk mitigation strategies. Thus, CO-INFORM can provide such metrics, e.g., the COVID-19 risk index based on the composite index methodology used by InfoRM.
InfoRM (Index for Risk Management) model is a global multi-hazard disaster risk assessment tool that identifies countries at risk of disaster and humanitarian crisis. This tool is developed by the European Commission’s Joint Research Commission (JRC) to support disaster risk management on prevention, preparedness, and response.
The risk concepts used in the model are based on several publications in scientific literature and considers the three dimensions of risk, namely, Hazard & Exposure, Vulnerability, and Lack of Coping Capacity. Though this framework does not define the interactions among these dimensions, this approach allows for a simple and transparent calculation of disaster risk.
The epidemic risk is also recently integrated with the InfoRM model through collaboration with the World Health Organization (WHO) (Doherty et al., 2018). Each dimension includes different categories where each category can be broken down by a reliable set of indicators. However, this index is non-specific in the context of the COVID-19 epidemic in the country. Some indicators may not be available or even relevant to the need of our decision-makers at the local level.
Chart 5 illustrates the proposed conceptual framework based on the concepts introduced in Promentilla (2019). We can use this as a guide to identifying indicators as metrics to manage risk and resilience. It is an adaptive framework and has the advantage of being modular. The framework can be updated by adding or removing elements in each dimension upon consultation with stakeholders or end-users.
For Hazard & Exposure dimension, biohazard that can cause epidemics or pandemics can be categorized as natural or human-caused. For COVID-19 coronavirus, there is scientific evidence that it is not bio-engineered, implying it is most likely of natural origin. Accordingly, epidemic disease susceptibility of a region in terms of exposure of the population to infectious agents such as that of person-to-person (P2P) transmission route, or other environmental pathways is considered under the Hazard&Exposure dimension. As for COVID-19, we will focus on person-to-person (P2P) transmission and its potential spread to the population.
This can be determined primarily from the epidemiological models providing quantitative indicators such as the likelihood of outbreak and the number of infected cases or fatalities. The emphasis is on the people at risk due to exposure of infectious agents such as that of coronavirus. The Hazard and Exposure are coupled into one dimension referred to as Hazard & Exposure since there is no risk if there is no exposure, no matter how severe the hazard event is. If there are physical exposure and physical vulnerability, the “hard” risk can be measured, and it is considered as hazard dependent.
In contrast, the “soft” risk can be measured from the second dimension using the concept of vulnerability due to the fragility of the socio-economic system, including the susceptibility associated with the low level of awareness, nutritional, and health status. These are the social determinants of health. This is hazard independent.
On the one hand, the physical vulnerability due to inherent predispositions of an exposed population to be affected or susceptible to COVID-19 disease is categorized as hazard dependent. The severity of such an epidemic in terms of human health would cause a disruption in society, which will result in a ripple effect on health, trade, and social order collectively. Thus, the vulnerability dimension represents the susceptibility of the individual, societal group, or community due to the social, economic, migration, and behavioral characteristics of the population. The more vulnerable the area or population is, the higher the risk will be.
The third dimension, which is made explicit in this proposed model is the concept of resilience. The definition of resilience is described in Part 1 of this article. Though coping or adaptive capacity is typically incorporated in vulnerability analysis, a separate measure of resilience capacity will be useful in tracking the results of risk mitigation measures, primarily if these are localized.
Resilience is thus operationalized and defined by physical infrastructure, health system capacity, institutional and resource management capacity. Conceptually, better epidemic management means higher resilience capacity, which translates to a lower level of risk. Since higher indicator values in risk index refer to worse conditions, the resilience capacity dimension is transformed into an index describing the lack of resilience when risk index is computed.
Poor governance with no concrete disaster or emergency response plan and implementation would pose a higher risk. This is attributed to the lack of capacity to recover due to a weak institution. This is hazard independent as the COVID-19 epidemic does not cause the institution to be more vulnerable. Instead, the crisis at hand only reveals the weakness of the institution. The lack of institutional capacity leads to higher soft risk.
On the other hand, the epidemic can overwhelm the healthcare capacity as this creates a surge of demand for critical care beds, ventilators, personal protective equipment (PPE), laboratory facilities, among others. The higher this resource gap is, the higher the risk is as well. This is hazard dependent.
Each dimension of the framework encompasses different categories to compute the risk index. Any individual indicator may not fully capture the categories/sub-categories, but these are mostly user-driven. The category can then be broken down into components/subcomponents and are presented with a carefully chosen set of indicators.
Data processing involves transforming all the data for indicators into a dimensionless number, e.g., through the max-min normalization method. This approach rescales all the data into scores ranging from 0 to 100, wherein the higher score indicates a worse condition or higher risk level. If a higher value of indicator would mean lower risk, inversion of the dataset is executed during normalization to make all values presented with the notion that higher is worse. For details of the composite index methodology, please refer to the document provided by InfoRM (De Groeve et al., 2014).
Chart 6 describes an application of using the CO-INFORM Framework for rapid risk assessment, i.e., to measure the risk of COVID-19 disaster. For the sake of illustration, Charts 7a and 7b describe sample output from CO-INFORM using the initial set of indicators identified from consultation with experts and stakeholders. These are still contingent considering the available data we have so far.
Chart 7b describes the risk stratification of cities/provinces based on the CO-INFORM risk index, wherein the three dimensions are aggregated with equal weighting. From Chart 7a, the metric for each dimension is described separately. The Hazard & Exposure index uses indicators such as the likelihood of an outbreak, and the predicted mortality rates provided by the UPLB Biomathematics Team lead by Dr. Jomar Rabajante. It is not surprising that the reddish region on the map is NCR, which is the epicenter of the outbreak in the country.
As for the vulnerability index, indicators such as the percentage of elderly and poverty incidences in the population are used. The reddish region on the map is found mostly in Mindanao, where poverty incidences are found to be relatively high.
As for the resilience capacity, indicators such as health expenditure per capita and projected maximum critical care bed demand are used as indicators for illustration. The map on lack of resilience describes the lack of coping capacity of healthcare systems for some regions because of the expected surge of critical care demand with respect to the baseline capacity of hospitals during the peak of the epidemic curve. These are all preliminary results and will be updated once the data for the set of indicators become available.
This is just one of the products being developed by our team from the LEADS for Health Security and Resilience Consortium. LEADS stands for Leading Evidence-Based Actions Through Data Science.
Who are we?
Through the “lead” of the Philippine Society of Public Health Physicians (PSPHP), we are a consortium of professionals from the academe, the civil society, and the private sector; volunteering and coming together in the wake of the evolving COVID19 pandemic. Capitalizing on our diverse but synergistic disciplines, we aim to harness the intersection of public health and data science to catalyze data-driven solutions toward transforming public health security and resilience, nationwide (Chart 8).
Concluding remarks.
The general guidelines provided by DOH are clear: prevent, detect, isolate, and treat (Chart 9). Resilience is mentioned in the said document but it focuses on living a healthy lifestyle and taking care of the vulnerable population. Our government is implementing several non-pharmaceutical interventions (NPIs) to address this crisis, as shown in Chart 10.
Are we doing enough to “fortify” our vulnerable communities, our institutions, and infrastructure, and prevent the re-emergence of the next outbreak? Are we doing enough testing and contact tracing in a rigorous and timely manner? Are we getting the right information at the right time from our leaders and the government? As a concerned citizen, we have to demand these things.
There are still many questions wherein answers are not also very clear yet, such as:
- Which provinces, cities, and municipalities are at higher risk from the COVID-19 epidemic, requiring additional support from the national government or international humanitarian assistance?
- What are the underlying factors leading to the crisis in a locality that amplifies the risk from this COVID-19 epidemic?
- How can we build community resilience through time to cope up and learn from this COVID-19 epidemic, and hopefully be prepared for the next threat or disruption?
CO-INFORM and the other tools being developed by LEADS hope to answer some of these questions.
Over the coming weeks and months, decisions will be made based on the best available data, useful models, and expert advice. These decisions will come under fire. We have to remember that all models are inherently wrong. But some are useful for policymaking. It is imperative to know how these models are made and the assumptions on which they are built.
The transparency of any data-driven models should not be in conflict with our desire to comply with data privacy regulations and ensures that the data is used ethically even in times of crisis. If the assumptions, including data used in the models, are not made explicit, and if the modelers are not open for scrutiny, then these models can also be abused and misused.
We are not only facing this pandemic COVID threat, but we also need to be wary of “infodemic” of hoaxes, conspiracy theories, honest misunderstandings, and politicized scientific debates.”
Models can be used for disinformation. Models can be sources of misinformation. Are we spreading disinformation or misinformation or pure B.S.?
Let me end this part of the article by quoting the evolutionary biologist Prof. Carl Bergstrom who co-authored a forthcoming book “Calling Bullshit: The Art of Scepticism in a Data-Driven World” :
“When these models get treated as if they’re oracles, then people both over-rely on them and treat them too seriously — and then turn around and slam them too hard for not being perfect at everything.”
***
I am neither a health professional nor an expert in epidemiology. I am just like you, a concerned citizen who would like to understand our situation right now. Use your critical reasoning to scrutinize this article and let me know your comments. Stay safe, healthy, and well informed.
Thank you for reading!
Please do share if you think this article or any similar one is informative and would help others to understand the situation at hand or change peoples’ opinions. The time to advocate for health systems resilience is NOW.
The author is a Professor at the De La Salle University, teaching risk assessment and management in the Environmental Engineering Graduate Program. He was also a 2018–2019 fellow at the Philippine Council for Health Research and Development under the ASEAN Science and Technology Fellowship program. His views are independent of those of his affiliations.
Special thanks to LEADS: Dr. Jomar Rabajante, Dr. Gelo Apostol, Dr. Peter Cayton, and Robert Leong, for the data and suggestions for CO-INFORM, Doc Dominic Ligot, for the implementation of CO-INFORM’s first prototype online, and to Dr. Mike Salazar and Dr. Jace Alacapa for convening the group despite most of us have not yet met face to face. Thanks also to my volunteer research assistants who help me in data encoding and processing, creating GIS maps and infographics: April Ann Tigue, Ithan Dollente, and JM Chan.