Points-Based Models for Predicting Severe Disease related to Covid-19 with the SARS-CoV-2 Omicron Variant

IntroductionThroughout the SARS-CoV-2 pandemic, resources for various aspects of patient care have been limited, necessitating risk-stratification. The need for good risk-stratification tools has been enhanced by the availability of new Covid-19 therapeutics that are effective at preventing severe disease among high-risk patients if given promptly following SARS-CoV-2 infection. We describe the development of two points-based models for predicting the risk of deterioration to severe disease from an Omicron-variant SARS-CoV-2 infection. MethodsWe developed two logistic regression-based models for predicting the risk of severe Covid-19 within a 21-days follow-up period among Clalit Health Services members aged 18 and older, with confirmed SARS-CoV-2 infection from December 25, 2021 to March 16, 2022. In the first model, aimed for the use of healthcare providers, the model coefficients were linearly transformed into integer risk points. In the second model, a simplified version designed for self-assessment by the general public, the risk points were further scaled down to smaller numbers with less variability across risk factors. Results613,513 individuals met the inclusion criteria, of which 1,763 (0.287%) developed the outcome. The AUROC estimates for both models were 0.95, although the full model demonstrated more granular risk-stratification capabilities (77 vs. 27 potential thresholds on the test set). Both models proved effective in identifying small subsets of the population enriched with individuals who ended up deteriorating. For example, prioritizing the top 1%, 5% or 10% individuals in the population for interventions with the full model results in coverage of 36%, 68% or 83% (respectively) of the individuals that actually end up deteriorating. Risk point count increased with age, number of chronic conditions and previous hospitalizations, and decreased with recent vaccination and infection. DiscussionThe models presented, one more expressive and one more accessible, are transparent and explainable models applicable to the general population that can be used in the prioritization of Covid-19-related resources, including therapeutics.

MHC Haplotyping of SARS-CoV-2 patients: HLA subtypes are not associated with the presence and severity of Covid-19 in the Israeli population

HLA haplotypes were found to be associated with increased risk for viral infections or disease severity in various diseases, including SARS. Several genetic variants are associated with Covid-19 severity. However, no clear association between HLA and Covid-19 incidence or severity has been reported. We conducted a large scale HLA analysis of Israeli individuals who tested positive for SARS-CoV-2 infection by PCR. Overall, 72,912 individuals with known HLA haplotypes were included in the study, of whom 6,413 (8.8%) were found to have SARS-CoV-2 by PCR. a Total of 20,937 subjects were of Ashkenazi origin (at least 2/4 grandparents). One hundred eighty-one patients (2.8% of the infected) were hospitalized due to the disease. None of the 66 most common HLA loci (within the five HLA subgroups; A, B, C, DQB1, DRB1) was found to be associated with SARS-CoV-2 infection or hospitalization. Similarly, no association was detected in the Ashkenazi Jewish subset. Moreover, no association was found between heterozygosity in any of the HLA loci and either infection or hospitalization. We conclude that HLA haplotypes are not a major risk/protecting factor among the Israeli population for SARS-CoV-2 infection or severity.

A score-based risk model for predicting severe COVID-19 infection as a key component of lockdown exit strategy

BackgroundAs many countries consider and employ various lockdown exit strategies, health authorities seek tools to provide differential targeted advice for social distancing based on personal risk for severe COVID-19. However, striking a balance between a scientifically precise multivariable risk prediction model, and a model which can easily be used by the general public, remains a challenge. A list of risk criteria, as defined by the CDC for example, provides a simple solution, but may be too inclusive by classifying a substantial portion of the population at high risk. Score-based risk classification tools may provide a good compromise between accuracy and simplicity. ObjectiveTo create a score-based risk classification tool for severe COVID-19. MethodsThe outcome was defined as a composite of being labeled severe during hospitalization or dying due to COVID-19. The risk classification tool was developed using retrospective data from all COVID-19 patients that were diagnosed until April 1st, 2020 in a large healthcare organization ("training set"). The developed tool combines 10 risk factors using simple summation, and defines three risk levels according to the patients age and number of accumulated risk points - basic risk, high risk and very-high risk (the last two levels are also considered together as the elevated risk group). The tools performance in accurately identifying individuals at risk was evaluated using a "temporal test set" of COVID-19 patients diagnosed between April 2nd and April 22nd, 2020, later than those used for model development. The tools performance was also compared to that of the CDCs criteria. The healthcare organizations general population was used to evaluate the proportion of patients that would be classified to each of the models risk levels and as elevated risk by the CDC criteria. ResultsA total of 2,421, 2,624 and 4,631,168 individuals were included in the training, test, and general population cohorts, respectively. The outcome rate in the training and test sets was 5%. Overall, 18% of the general population would be classified at elevated risk by the model, with a resulting sensitivity of 92%, compared to 35% that would be defined as elevated risk by the CDC criteria, with a resulting sensitivity of 96%. Within the models elevated risk groups, the high and very-high risk groups comprised 15% and 3% of the general population, with an incidence rate (PPV) of 15% and 33%, respectively. DiscussionA simple to communicate score-based risk classification tool classifies at elevated risk about half of the population that is considered to have an elevated risk by the CDC risk criteria, with only a 4% reduction in sensitivity. The models ability to further divide the elevated risk population into two markedly different subgroups allows providing more refined recommendations to the general public and limiting the restrictions of social distancing to a smaller and more manageable subset of the population. This model was adopted by the Israeli ministry of health as its risk classification tool for COVID-19 lab tests prioritization and for targeting its instructions on risk management during the lockdown exit strategy.

Who should we test for COVID-19?A triage model built from national symptom surveys

The gold standard for COVID-19 diagnosis is detection of viral RNA in a reverse transcription PCR test. Due to global limitations in testing capacity, effective prioritization of individuals for testing is essential. Here, we devised a model that estimates the probability of an individual to test positive for COVID-19 based on answers to 9 simple questions regarding age, gender, presence of prior medical conditions, general feeling, and the symptoms fever, cough, shortness of breath, sore throat and loss of taste or smell, all of which have been associated with COVID-19 infection. Our model was devised from a subsample of a national symptom survey that was answered over 2 million times in Israel over the past 2 months and a targeted survey distributed to all residents of several cities in Israel. Overall, 43,752 adults were included, from which 498 self-reported as being COVID-19 positive. We successfully validated the model on held-out individuals from Israel where it achieved a positive predictive value (PPV) of 46.3% at a 10% sensitivity and demonstrated its applicability outside of Israel by further validating it on an independently collected symptom survey dataset from the U.K., U.S. and Sweden, where it achieved a PPV of 34.7% at 10% sensitivity. Moreover, evaluating the models performance on this latter independent dataset on entries collected one week prior to the PCR test and up to the day of the test we found the highest performance on the day of the test. As our tool can be used online and without the need of exposure to suspected patients, it may have worldwide utility in combating COVID-19 by better directing the limited testing resources through prioritization of individuals for testing, thereby increasing the rate at which positive individuals can be identified and isolated.

The effect of a national lockdown in response to COVID-19 pandemic on the prevalence of clinical symptoms in the population

The vast and rapid spread of COVID-19 calls for immediate action from policy-makers, and indeed, many countries have implemented lockdown measures to varying degrees. Here, we utilized nationwide surveys that assess COVID-19 associated symptoms to analyse the effect of the lockdown policy in Israel on the prevalence of clinical symptoms in the population. Daily symptom surveys were distributed online and included questions regarding fever, respiratory symptoms, gastrointestinal symptoms, anosmia and ageusia. A total of 2,071,349 survey responses were analysed. We defined a single measure of symptoms, Symptoms Average (SA), as the mean number of symptoms reported by responders. Data were collected between March 15th to June 3rd, 2020. Notably, on the population level, following severe lockdown measures between March 15 th and April 20th, SA sharply declined by 83.8% (p < 0.05), as did every single symptom, including the most common symptoms reported by our responders, cough and rhinorrhea and\or nasal congestion, which decreased by 74.1% (p < 0.05) and 69.6% (p < 0.05), respectively. Similarly, on the individual level, analysis of repeated responses from the same individuals (N = 208,637) over time also showed a decrease in symptoms during this time period. Moreover, the reduction in symptoms was observed in all cities in Israel, and in several stratifications of demographic characteristics. Different symptoms exhibit different reduction dynamics, suggesting differences in the nature of the symptoms or in the underlying medical conditions. Between May 13th and June 3rd, following several subsequent lockdown relief measures, we observed an increase in individual symptoms and in SA, which increased by 31.42%. Overall, these results demonstrate a profound decrease in a variety of clinical symptoms following the implementation of a lockdown in Israel, and an increase in the prevalence of symptoms following the loosening of lockdown restrictions. As our survey symptoms are not specific to COVID-19 infection, this effect likely represents an overall nationwide reduction in the prevalence of infectious diseases, including COVID-19. This quantification may be of major interest for COVID-19 pandemic, as many countries consider implementation of lockdown strategies.

Performing risk stratification for COVID-19 when individual level data is not available, the experience of a large healthcare organization

With the global coronavirus disease 2019 (COVID-19) pandemic, there is an urgent need for risk stratification tools to support prevention and treatment decisions. The Centers for Disease Control and Prevention (CDC) listed several criteria that define high-risk individuals, but multivariable prediction models may allow for a more accurate and granular risk evaluation. In the early days of the pandemic, when individual level data required for training prediction models was not available, a large healthcare organization developed a prediction model for supporting its COVID-19 policy using a hybrid strategy. The model was constructed on a baseline predictor to rank patients according to their risk for severe respiratory infection or sepsis (trained using over one-million patient records) and was then post-processed to calibrate the predictions to reported COVID-19 case fatality rates. Since its deployment in mid-March, this predictor was integrated into many decision-processes in the organization that involved allocating limited resources. With the accumulation of enough COVID-19 patients, the predictor was validated for its accuracy in predicting COVID-19 mortality among all COVID-19 cases in the organization (3,176, 3.1% death rate). The predictor was found to have good discrimination, with an area under the receiver-operating characteristics curve of 0.942. Calibration was also good, with a marked improvement compared to the calibration of the baseline model when evaluated for the COVID-19 mortality outcome. While the CDC criteria identify 41% of the population as high-risk with a resulting sensitivity of 97%, a 5% absolute risk cutoff by the model tags only 14% to be at high-risk while still achieving a sensitivity of 90%. To summarize, we found that even in the midst of a pandemic, shrouded in epidemiologic "fog of war" and with no individual level data, it was possible to provide a useful predictor with good discrimination and calibration.

Building an International Consortium for Tracking Coronavirus Health Status

Information is the most potent protective weapon we have to combat a pandemic, at both the individual and global level. For individuals, information can help us make personal decisions and provide a sense of security. For the global community, information can inform policy decisions and offer critical insights into the epidemic of COVID-19 disease. Fully leveraging the power of information, however, requires large amounts of data and access to it. To achieve this, we are making steps to form an international consortium, Coronavirus Census Collective (CCC, coronaviruscensuscollective.org), that will serve as a hub for integrating information from multiple data sources that can be utilized to understand, monitor, predict, and combat global pandemics. These sources may include self-reported health status through surveys (including mobile apps), results of diagnostic laboratory tests, and other static and real-time geospatial data. This collective effort to track and share information will be invaluable in predicting hotspots of disease outbreak, identifying which factors control the rate of spreading, informing immediate policy decisions, evaluating the effectiveness of measures taken by health organizations on pandemic control, and providing critical insight on the etiology of COVID-19. It will also help individuals stay informed on this rapidly evolving situation and contribute to other global efforts to slow the spread of disease. In the past few weeks, several initiatives across the globe have surfaced to use daily self-reported symptoms as a means to track disease spread, predict outbreak locations, guide population measures and help in the allocation of healthcare resources. The aim of this paper is to put out a call to standardize these efforts and spark a collaborative effort to maximize the global gain while protecting participant privacy.

A framework for identifying regional outbreak and spread of COVID-19 from one-minute population-wide surveys

Coronavirus infection spreads in clusters and therefore early identification of these clusters is critical for slowing down the spread of the virus. Here, we propose that daily population-wide surveys that assess the development of symptoms caused by the virus could serve as a strategic and valuable tool for identifying such clusters to inform epidemiologists, public health officials, and policy makers. We show preliminary results from a survey of over 58,000 Israelis and call for an international consortium to extend this concept in order to develop predictive models. We expect such data to allow: Faster detection of spreading zones and patients; Obtaining a current snapshot of the number of people in each area who have developed symptoms; Predicting future spreading zones several days before an outbreak occurs; Evaluating the effectiveness of the various social distancing measures taken, and their contribution to reduce the number of symptomatic people. Such information can provide a valuable tool for decision makers to decide which areas need strengthening of social distancing measures and which areas can be relieved. Preliminary analysis shows that in neighborhoods with confirmed COVID-19 patient history, more responders report on COVID-19 associated symptoms, demonstrating the potential utility of our approach for detection of outbreaks. Researchers from other countries including the U.S, India, Italy, Spain, Germany, Mexico, Finland, Sweden, Norway and several others have adopted our approach and we are collaborating to further improve it. We call with urgency for other countries to join this international consortium, and to share methods and data collected from these daily, simple, one-minute surveys.