UST CoV-2 Model
Introducing the UST CoV-2 Model for the COVID-19 Pandemic in Metro Manila
An epidemiological model is a mathematical tool that tries to predict the future behavior of a viral pandemic by examining how it has spread in the past. In this project, we describe our efforts to model the COVID-19 pandemic in Metro Manila.
We have chosen to focus on the National Capital Region of the Philippines, not only because it is the nation’s demographic heart where over a tenth of the country’s population live, but also because it has been, by far, the epidemiological epicenter of the Philippine pandemic.
We began with the DELPHI Model, a machine-learning capable epidemiological model developed last month at M.I.T. in the United States and adapted it so that it could be used to investigate the pandemic and the impact of quarantine measures in the National Capital Region.
-Bernhard Egwolf and Nicanor Austriaco, O.P.
Evaluating the COVID-19 Pandemic Surge in the National Capital Region
In response to the numerous reports of increasing numbers of COVID-19 positive cases in the National Capital Region (NCR), we analyzed the current pandemic data provided by the Department of Health up to and including July 8, 2020, to evaluate the extent of the surge.
The current epidemic curve for the NCR is shown in Figure 1. It confirms reports that the number of COVID-19 positive cases is on the rise. As shown in Figure 2, however, the number of daily deaths from COVID-19 continues to fall in the NCR. This is expected since the number of deaths from COVID-19 usually lags behind the number of cases of COVID-19 by two weeks.
The increase in positive cases can be attributed either to an increase in testing capacity and/or an increase in the spread of the virus. As shown in Figure 3, testing capacity has also dramatically risen in the NCR so we need to take this into account.
To determine which of these options best explains the data, we examined both the positivity rate and the hospitalization rate for the NCR. The positivity rate is the percentage of COVID-19 tests that are positive. It is an indication of the extent of community transmission in a population. The hospitalization rate reveals the change in COVID-19 bed occupancy in a geographic region. Both of these increase when the pandemic is spreading through a local community.
As shown in Figure 4, the positivity rate for the NCR has been increasing gradually for the past two weeks. Just over two weeks ago, the positivity rate hovered between 6% to 8% of tests. Today, the positivity rate is around 12% and has exceeded 10% for the past week or so. To maintain control of their pandemics, the WHO recommends that countries try to reduce their positivity rates to below 5%.
As shown in Figure 5, the hospitalization occupancy for COVID-19 patients has also spiked in recent weeks. Both the percentage of occupied total COVID-19 and ICU COVID-19 beds have increased. Mercifully, the percentage of occupied ventilators has not spiked as dramatically, supporting claims in the news media that the current COVID-19 patient population is younger and experiencing milder cases of the disease.
Together, the rise in the positivity rate and in the hospitalization rate indicate that the NCR is undergoing a real surge in the pandemic which needs to be addressed immediately by public health authorities.
To determine if we could identify hotspots for the pandemic surge in the NCR, we broke down the daily cases in June at two-week intervals and assigned each of them to one of the component cities of the NCR.
As shown in Figure 6, though there have been noticeable increases in cases in several of the component cities of the NCR, a significant percentage of the surge can be attributed to a dramatic rise in the number of positive cases in the DOH category labelled “Unknown City.” These are positive cases that are not assigned to a particular city but are known to be in the NCR.
It is troubling that the surge is occurring in this “unknown” geographical category. The number of COVID-19 positive persons is disproportionately increasing in this category relative to the other geographical regions in the NCR. Statistically, this cannot be explained by random clerical or encoding errors.
Instead, it may suggest that LGUs may be withholding geographical information when they suspect that individuals in their barangays are positive for COVID-19. However, without proper geographical identification of positive cases, it will be difficult for public health authorities to properly understand the extent of the surge and to control the pandemic through contact tracing, tracking, and isolation.
Finally, we chose to break down the hospitalization occupancy data by component city to determine hotspots where hospital capacity may be overwhelmed by the local pandemic. As shown in Figures 7 and 8, on July 8, 2020, several component cities of the NCR were exceeding the 70% “Danger” occupancy limit established by the DOH. We repeated the analysis for July 12, 2020, and the Figures 9 and 10 show that more component cities have reached their “Danger” occupancy limit established by the DOH.
Cities that were exceeding this “Danger” limit for total COVID-19 beds on July 12, 2020, included Navotas, Makati, Malabon, Mandaluyong, Las Pinas, Manila, Pateros, Muntinlupa, Quezon City, Pasig, and Taguig. These are cities in danger of being crushed by the growing pandemic in the NCR.
- We strongly recommend that the government maintain the NCR in the GCQ until the current surge has been brought under control. Evidence for this will include a two week period of decreasing positivity and hospitalization rates. Positivity rates should also be stable below 10%, and ideally below 5%, and hospitalization occupancy rates should be below 50%.
- We strongly recommend that the DOH enforce public health reporting policies that will lower the number of “unknown” positive cases that lack a geographical identifier. Without proper geographical identification of positive cases, it will be difficult for public health authorities to properly grasp the extent of the surge and to control the pandemic through contact tracing, tracking, and isolation.
- We strongly recommend that LGUs continue to enforce local and targeted lockdowns in their respective geographical areas. As in the case of Cebu City where local authorities were hesitant to go into lockdown, we urge the national government to properly supervise this process to ensure that zoned lockdowns are enacted in a timely manner. Furthermore, targeted lockdowns should be maintained for at least two weeks – which is one incubation period of the virus – to mitigate viral spread. It is troubling to read news reports that reveal that some of the targeted lockdowns have only been kept in place for several days. This is not enough time to allow asymptomatic carriers to become non-infectious.
- As we have recently witnessed in the city of Melbourne in Australia, localized lockdowns can quickly fail if they are not quickly and securely enforced. Because localized lockdowns failed in Melbourne, the entire city has just entered a six-week total lockdown to protect their health care capacity. If localized lockdowns are unable to slow the spread in the NCR, the government must be ready to return the component cities of the NCR to an MECQ or even an ECQ to protect our health care system and to minimize cases and deaths.
- We strongly recommend that the national government restrict the movement of all LSIs and returning OFWs who wish to leave the NCR until the current surge in the pandemic has been brought under control. This should be done to protect the fragile health care system in the provinces that will not be able to handle a surge in their locations. Furthermore, we recommend that provincial LGUs implement mandatory 14 day quarantines for all returning LSIs and OFWs, regardless of their testing result in the NCR. It has become a concern that rapid testing in the NCR has not been able to detect all the asymptomatic carriers of COVID-19 who have then transported the disease back to their home provinces.
A wicked problem is defined as a social challenge that is difficult to approach for as many as four reasons: the prevalence of incomplete or contradictory knowledge, the number of people and opinions involved, the impact of a large economic burden, and the interconnected nature of this problem with other social problems. Understanding and managing a once-in-a-century global pandemic is a wicked problem.
Tackling wicked problems requires out-of-the-box, non-traditional thinking that involves interdisciplinary approaches among experts from a diverse scholarly background.
Last month, the UST-CoV2 scientists entered into a collaboration with our colleagues from the UP OCTA research group to create an inter-disciplinary team of experts from the natural, medical, and social sciences to better understand the current COVID-19 pandemic in the Philippines.
Our goal is to provide models and projections to help government and our people mitigate the morbidity and mortality associated with SARS-CoV2.
The most recent report of our collaborative effort is described here. Please note that the findings and recommendations are those of the authors and do not reflect the official position of the University of Santo Tomas, the University of the Philippines, Providence College, or any of their units.
-Rev. fr. Nicanor Austriaco, O.P., PhD, SThD
Report #2 - June 15, 2020 - Evaluating the Impact of the GCQ on the COVID-19 Pandemic in Metro Manila
Evaluating the Impact of the GCQ on the COVID-19 Pandemic in Metro Manila
On June 1, 2020, the National Capital Region was moved from a MECQ to a GCQ. Since it has been two weeks since the NCR moved to this new level of quarantine – which is equivalent to a single incubation period for COVID-19 – we believe that this is an appropriate time to evaluate its impact on the dynamics of the local pandemic in Metro Manila.
Key Finding #1
• The GCQ increased mobility in the NCR by about 10-15% from levels observed during the MECQ. Overall mobility in Metro Manila remains below 50% of levels recorded before the COVID-19 pandemic, as measured by Apple and Google.
As shown in the left panel of Figure 1, the shift from the MECQ to the GCQ increased mobility in Metro Manila by about 10-15% from levels observed during the MECQ. Nonetheless, mobility levels, especially mobility on public transit, remain relatively low. They are more than 50% below baseline as measured by Apple during the first two months of the year. It also appears that the mobility levels of the GCQ have stabilized at driving (-53%), walking (-58%), and transit (-73%).
These low mobility levels at more than 50% below baseline are expected since the GCQ prohibits movement of those who are younger than 21 and those who are older than 60 years of age. Together, these two groups of persons constitute about 50% of the population of the Philippines.
To avoid selection bias, since it is not clear if Apple users are representative of the population in Metro Manila, we confirmed the Apple mobility trends using mobility data provided by Google. As shown in the right panel of Figure 1, the trends appear to be comparable. The average mobility as determined by Google for the five categories of locations it tracks in Metro Manila was -56% on June 7, 2020. Once again, the baseline is population mobility in the NCR prior to the pandemic.
Key Finding #2
• Though there was an apparent spike in reported cases of COVID-19 earlier in June, we do not believe that this represents a surge in community transmission of the disease. The positivity rate in the NCR remains low and COVID-19 hospitalizations in Metro Manila are decreasing.
As shown in Figure 2, on the left panel, it appears that the number of daily cases of COVID-19 reported by the Department of Health for the NCR spiked earlier in June. However, it is not easy to interpret this spike because the DOH explained that their reported positive cases at the end of May and at the beginning of June included both “fresh” and “late” cases. Fresh cases were patients who received their positive test results three or less days prior to the reporting date and who were only recently validated. Late cases were validated tests for tests administered more than four days prior to the reporting date.
Furthermore, as shown in the right panel of Figure 2, we do not see a spike in the number of daily deaths in the NCR, even two weeks after the spike in daily cases. Deaths from COVID-19 lag behind reported new cases by about 10-14 days because of the incubation period of the disease. Indeed, it is hopeful that the number of daily deaths in Metro Manila continues to decrease from a peak in early May. The numbers of daily cases and daily deaths used for these graphs were those provided by the DOH on its COVID-19 tracker on June 15, 2020.
Therefore, to see if the spike in reported cases reflected a true surge of COVID-19 infections in the community, we calculated the positivity rate for Metro Manila for the past two months. The positivity rate is the percentage of total COVID-19 tests that are positive. If the pandemic is spreading silently within the population, we would expect the positivity rate to increase as more and more people become infected and some are then tested positive.
To calculate the positivity rate, we counted the total number of tests, positive tests, and negative tests reported by the DOH-licensed RT-PCR laboratories in the NCR as listed in the COVID-19 Philippine Situationer #044 published by the DOH. The numbers of tests were those provided by the DOH in its Data Drop for June 15, 2020. Tests with no clear result were ignored.
First, as shown in the left panel of Figure 3, it is clear that the testing capacity for the NCR has increased eight-fold during the past two months in Metro Manila. For the past two weeks, there have been around 8,000 tests performed daily at DOH approved laboratories.
Guidelines from the WHO published on May 12, 2020, indicate that countries should aim to test one out of every 1,000 residents every week and should see less than 5% of those tests come back positive for at least two consecutive weeks before they begin to relax social distancing and other public health measures. A low positivity rate indicates low community transmission.
As shown in the right panel of Figure 3, the positivity rate for Metro Manila has fallen from a high of around 30% at the beginning of April to just around 5% in early June. We believe that this indicates that community spread remains low despite the apparent spike in cases reported at the beginning of June.
To see if the spike in reported cases reflected a true surge of COVID-19 infections in the community, we also determined the COVID-19 hospitalization rate for Metro Manila for the past month. If the pandemic is spreading silently within the population, then we would expect the hospitalization rate to increase as more and more people become infected and some are then hospitalized.
To determine the hospitalization rate, we counted the number of occupied and vacant COVID-19 beds and occupied and vacant COVID-19 ICU beds as reported by the hospitals in the NCR to the DOH, as these numbers were recorded in the DOH Data Drop for June 15, 2020.
As shown in the left panel of Figure 4, the numbers of total beds and ICU beds available for COVID-19 patients have remained steady in the NCR. However, the percentage of both occupied total beds and occupied ICU beds for COVID-19 patients has gradually decreased over the past month. Once again, we believe that this indicates that community spread remains low in Metro Manila despite the apparent spike in cases reported at the beginning of June.
In sum, we conclude that the local pandemic in the NCR remains relatively subdued and controlled despite the move from the MECQ to the GCQ on June 1, 2020. The apparent spike in cases reported at the beginning of June appears to be explained best by changes in the DOH validation process of past cases, rather than by a surge in the number of infected patients.
Key Finding #3
• Though it is difficult to properly forecast the progression of the local pandemic because of the uncertainty in the dating of the DOH data after June 1, 2020, we used our machine-learning UST CoV-2 model to determine future trends of the disease in the NCR.
As shown in Figure 5, on the left panel, with the total daily case numbers provided by the DOH, we forecast that there will be 16,547 cases in the NCR by June 30, 2020, with 2,756 deaths. With total cases included, the modeled Reff = 1.1. However, as shown in Figure 5 on the right panel, when we repeated the modeling using only the “fresh” cases reported by the DOH, these estimates dropped to 14,032 total cases and 1,853 deaths. With only fresh cases included, the modeled Reff = 1.0.
Once again, because of the uncertainty of the DOH data, these numbers remain estimates that can give us a sense of the possible progression of the pandemic. Given our earlier analysis, we favor the second model that only includes the fresh cases since this forecast would follow from a controlled local pandemic.
Furthermore, we want to stress that these numbers are not inevitable. They can be reduced with robust testing and contact tracing programs. Consequently, we urge public health authorities to continue to increase and to strengthen both of these capacities in the National Capital Region.
• Before Metro Manila is moved from a GCQ to an MGCQ, we recommend that the positivity rate should remain below 5% for at least another two weeks, and that levels of hospitalization in the NCR should continue to decrease, again for another two weeks, to indicate that the pandemic is not surging. This will also give local authorities additional time to increase both testing capacity and contact tracing ability to counter future outbreaks of this viral disease.
As we explained in our earlier report published online on this website on May 24, 2020, at this stage of the pandemic, according to the recommendations of a team from Harvard University, the NCR should have a testing capacity of 15,000 tests per day and 1,800 contact tracers working in call centers scattered throughout the region, to control its local pandemic. We do not believe that these recommended numbers need to be revised since they are based on deaths per day, and the number of COVID-19 deaths per day have been gradually decreasing for the past several weeks in Metro Manila. Deo gratias.
Finally, we encourage local authorities to continue to impose targeted lockdowns by zones to quarantine affected buildings, streets, and/or barangays that have elevated numbers of COVID-19 positive patients. This strategy of focused lockdowns should be coupled with extensive testing and tracking of contacts within these “high-risk” red zones to break chains of viral transmission. Recent studies have suggested that young persons who have COVID-19 – the population of the Philippines is young with a median age of 24 years of age – are often asymptomatic. Only testing for the virus using the RT-PCR approach will be able to identify these silent spreaders within the cities of Metro Manila.
Report #1 - May 24, 2020: Introducing the UST CoV-2 Model, Evaluating the Impact of ECQ on Metro Manila
Impact of ECQ Implementation in NCR
As shown in Figure 1, the government-imposed enhanced community quarantine (ECQ) appears to have limited the impact of the pandemic in Metro Manila by significantly lowering the total number of COVID-19 cases and the total number of deaths.
As a point of comparison, Metro Manila and New York City, two metropolitan areas with comparable populations, recorded their first deaths from community spread of COVID-19 on March 11, 2020, and March 13, 2020, respectively. However, Metro Manila entered lockdown on March 15, 2020, while New York City waited one more week to enter its lockdown on March 20, 2020. As of May 21, 2020, Metro Manila reported 621 total deaths from COVID-19 while New York City confirmed 16,232 total deaths. This suggests that the early implementation of the ECQ was able to save thousands of lives in the NCR. We commend the public health authorities for their efforts.
Figure 1. Modeling the Impact of the Enhanced Community Quarantine (ECQ) on the COVID-19 Pandemic in Metro Manila. Forecasted estimates of total COVID-19 cases and total deaths for the National Capital Region without a lockdown and with the Enhanced Community Quarantine compared to the data provided by the Department of Health of the Philippines.
On flattening the first wave of COVID-19
As shown in Figure 2, the first wave of the pandemic appears to be receding. The curve seems to have been flattened: The number of forecasted active cases peaked in early May and is gradually declining. Currently, there is no evidence for a second wave of the pandemic.
However, the flattening of the curve is not dramatic. In fact, it is clear that the ECQ is struggling to suppress the pandemic, i.e., to drive the number of infected cases down to zero. At this rate, the model predicts that the total number of cases in Metro Manila will continue to increase gradually and will not plateau for many months, though the forecasted number of active cases will also be decreasing. If the current quarantine measures are maintained, the forecasted number of active cases of COVID-19 will not fall below 1,000 cases until early September this year.
ECQ Effectiveness in Some NCR Cities
As shown in Figure 3, our modeling reveals that the effectiveness of the ECQ has varied widely throughout the NCR. Cities like Makati, Mandaluyong, and Parañaque have been more successful at suppressing the community spread of COVID-19 than Manila or Quezon City.
If the current quarantine measures are maintained, the total number of cases is forecasted to plateau by mid-July in Mandaluyong, by early August in Makati, and by September in Parañaque. However, the same cannot be said for the latter two cities. Despite the laudable effort of the local authorities, the total number of cases is forecasted to gradually keep increasing for many months, though the forecasted number of active cases will also be decreasing.
It is not clear why the ECQ has not been as effective in both the City of Manila and Quezon City as it has been in their neighboring municipalities, especially since the local authorities have worked hard to suppress their infection rates. They are not the top two most dense cities in the NCR so it is unlikely to be explained by appealing to population density alone. Nonetheless, both these urban areas have experienced a several-fold higher number of infections than the other component cities in the capital region. Thus, it is not unreasonable to think that the dynamics of the COVID-19 pandemic in Metro Manila are being driven by both local pandemics. It suggests that public health authorities should increase their efforts to contain the pandemic in the NCR in these two cities.
Forecasted Impact of Metro Manila’s Transition to GCQ
As shown in Figure 4, it is clear that replacing the ECQ with a GCQ will increase the forecasted number of infected cases and deaths in Metro Manila. However, it is important to emphasize that these increases are not inevitable. They can be offset with a rigorous tracking, testing, and tracing program that seeks to limit community spread by breaking chains of viral transmission. We acknowledge that national and local authorities have worked hard to increase the testing capacity in the Philippines in the past two months, and we commend them for their efforts.
Given the observation that typically, only 20% of a population contributes to at least 80% of the potential to transmit infectious disease, we recommend that tracking, testing, and tracing programs in Metro Manila focus their efforts on limiting the impact of superspreading events (SSEs) which are associated with both explosive growth early in an outbreak and sustained transmission in later stages.
How many tests do we need? What about contact tracing?
How much testing and tracing capacity will be needed in Metro Manila to keep the pandemic at bay? A team of scholars from Harvard University has recently published recommendations for communities of different sizes and at different stages in controlling their pandemics. They propose that every municipality with a moderate infection rate, i.e., with less than 1% prevalence of active virus in its population – for the most part, the NCR appears to fulfill this criterion – should hire sixty teams of five tracers for every death per day that they observe in their community, and maintain a testing capacity of 2,500 tests for every death per day.
On May 22, 2020, the seven-day running average of deaths per day recorded in Metro Manila was six deaths per day. This means that at this stage of the pandemic, according to the recommendations of the Harvard team, the NCR would need a testing capacity of 15,000 tests per day and 1,800 contact tracers working in call centers scattered throughout the region, to control its local pandemic. Geographic distribution of this testing and contact tracing capacity should correspond to the severity of the local pandemics in each of the cities of Metro Manila, with a particular focus on the City of Manila and Quezon City.
Notably, the College of Science and the Faculty of Engineering at the University of Santo Tomas is developing a design proposal to create mobile COVID-19 testing labs that can be transported on flatbed trucks using repurposed container vans. We recommend that LGU’s in Metro Manila consider adopting this design proposal for mobile testing in their municipalities.
For more detail on the COVID-19 model and citations to the primary literature, see: Bernhard Egwolf and Nicanor Austriaco, O.P., “Mobility-Guided Modeling of the COVID-19 Pandemic in Metro Manila” MedRxiv ID#: MEDRXIV/2020/111617. Uploaded on May 26, 2020.
Rev. Nicanor Pier Giorgio Austriaco, O.P., Ph.D., S.Th.D.
Fr. Nicanor Austriaco, O.P., is a Catholic priest in the Order of Friars Preachers, also known as the Dominican Order. Born in the Philippines, he completed his Bachelor’s Degree in Bioengineering, summa cum laude, at the University of Pennsylvania, and then earned his Ph.D. in Biology from M.I.T. where he was a fellow of the Howard Hughes Medical Institute. After completing his doctoral studies, Fr. Austriaco was a fellow of the International Human Frontier Science Program at the Ludwig Institute for Cancer Research at the University College London.
In 1997, following a deeper conversion to the Lord, Fr. Austriaco entered the novitiate of the Order of Friars Preachers. He completed both his pontifical bachelor’s degree in sacred theology and his Master’s of Divinity degree at the Pontifical Faculty of the Immaculate Conception at the Dominican House of Studies in Washington, D.C., in 2003. He was ordained to the priesthood on May 21, 2004. He completed his Pontifical License in Sacred Theology (S.T.L.) in Moral Theology at the Dominican House of Studies in Washington, DC, and a Pontifical Doctorate in Sacred Theology (S.T.D.) at the University of Fribourg in Switzerland. Fr. Austriaco earned his M.B.A. from Providence College in 2020.
Fr. Austriaco currently serves as Professor of Biology and of Theology at Providence College in Providence, Rhode Island. He is also a research fellow at the Center for Religious Studies and Ethics at the University of Santo Tomas in Manila, Philippines, and director of ThomisticEvolution.org. His NIH-funded laboratory at Providence College is investigating the genetics of programmed cell death using the yeasts, Saccharomyces cerevisiae and Candida albicans, as model organisms. Fr. Austriaco is a bioethics consultant for the Catholic bishops’ conferences of the United States and of the Philippines.
Fr. Austriaco has published over fifty peer-reviewed papers in biology, philosophy, and theology, and two books, including. Biomedicine and Beatitude: An Introduction to Catholic Bioethics, which was recognized as a 2012 Choice outstanding academic title by the Association of College and Research Libraries. A second edition of the bestselling book is expected in 2021.
Assoc. Prof. Bernhard Egwolf of the Department of Mathematics and Physics holds a Doctor of Natural Sciences in Physics degree, magna cum laude, from the Ludwig Maximilian University of Munich, Germany, where he also earned his University Diploma in Physics.
As a biophysicist, his research interests are geared toward studies on theoretical and computational biophysics, computer simulations of ion channel proteins, calculaton of entropies from high-dimensional simulation data, and molecular docking of drug molecules. His studies have been published in reputable journals, like the Journal of Computational Chemistry, Journal of Biological Chemistry, and the Journal of Molecular Biology, among others.
Dr. Egwolf took part in a two-year post-doctoral research fellowship in the University of Chicago, United States of America, in 2007 upon receiving a grant from the German Research Foundation. Currently, he is a researcher at the Research Center for Natural and Applied Sciences of the University of Santo Tomas.
Angus White is a senior at Providence College, Rhode Island, where he majors in Biology. Originally from Oakura, New Zealand, White is now assisting Fr. Austriaco and Dr. Egwolf full-time in this research.