Early epidemiological indicators, outcomes, and interventions of COVID-19 pandemic: A systematic review

Background Coronavirus disease-2019 (COVID-19), a pandemic that brought the whole world to a standstill, has led to financial and health care burden. We aimed to evaluate epidemiological characteristics, needs of resources, outcomes, and global burden of the disease. Methods Systematic review was performed searching PubMed from December 1, 2019, to March 25, 2020, for full-text observational studies that described epidemiological characteristics, following MOOSE protocol. Global data were collected from the JHU-Corona Virus Resource Center, WHO-COVID-2019 situation reports, KFF.org, and Worldometers.info until March 31, 2020. The prevalence percentages were calculated. The global data were plotted in excel to calculate case fatality rate (CFR), predicted CFR, COVID-19 specific mortality rate, and doubling time for cases and deaths. CFR was predicted using Pearson correlation, regression models, and coefficient of determination. Results From 21 studies of 2747 patients, 8.4% of patients died, 20.4% recovered, 15.4% were admitted to ICU and 14.9% required ventilation. COVID-19 was more prevalent in patients with hypertension (19.3%), smoking (11.3%), diabetes mellitus (10%), and cardiovascular diseases (7.4%). Common complications were pneumonia (82%), cardiac complications (26.4%), acute respiratory distress syndrome (15.7%), secondary infection (11.2%), and septic shock (4.3%). Though CFR and COVID-19 specific death rates are dynamic, they were consistently high for Italy, Spain, and Iran. Polynomial growth models were best fit for all countries for predicting CFR. Though many interventions have been implemented, stern measures like nationwide lockdown and school closure occurred after very high infection rates (>10cases per 100 000population) prevailed. Given the trend of government measures and decline of new cases in China and South Korea, most countries will reach the peak between April 1-20, if interventions are followed. Conclusions A collective approach undertaken by a responsible government, wise strategy implementation and a receptive population may help contain the spread of COVID-19 outbreak. Close monitoring of predictive models of such indicators in the highly affected countries would help to evaluate the potential fatality if the second wave of pandemic occurs. The future studies should be focused on identifying accurate indicators to mitigate the effect of underestimation or overestimation of COVID-19 burden.

The origin of severe acute respiratory syndrome coronavirus (SARS-CoV-2) virus was linked to a seafood market in Wuhan from the handling and close contact with animals [4]. In USA, the first case was reported on January 20, 2020, with a recent travel history to Wuhan [5]. According to emerging literature, COVID-19 symptoms can range from mild respiratory illness causing fever, dry cough, dyspnea, myalgia and fatigue to more severe manifestation of pneumonia, cardiac complications requiring intensive care unit (ICU) admission and mechanical ventilation [6]. The median incubation period is around 5 days (range:2-14 days), requiring prolonged monitoring in extreme cases [7,8]. Real-time reverse transcriptase polymerase chain reaction (RT-PCR) of nasopharyngeal and/or oropharyngeal swabs are usually used to confirm the diagnosis [9,10]. Preliminary demographic data of the infected patients suggests that most patients have mild disease, with older adults (≥65 years) appearing to be more susceptible to severe illness requiring hospitalization [11,12]. COVID-19 shows evidence of human to human transmission via respiratory droplets and from contact with contaminated surfaces or objects, with estimated median basic reproduction number (R0) of 2.28 (range: 2.06-2.52) [13], making the spread of the disease tough to contain.
While recently published observational studies have provided insights on the epidemiology of this pandemic, their sample sizes are too limited for any definitive conclusions. Hence, we sought to conduct a systematic review and analysis of all available studies comparing outcomes. Primary aim of the study is to evaluate the epidemiological characteristics, needs of resources, and patients' outcomes. Secondary aim is to evaluate the global burden and interventions.

Endpoints
We evaluated epidemiological characteristics, risk factors, laboratory and imaging findings, complications and treatment utilized. We also calculated the mortality, recovery, and needs of resources like ICU beds and mechanical ventilators.

Eligibility criteria, search strategy and selection criteria
In order to evaluate the primary outcome, we performed a systematic review of these observational studies according to MOOSE guidelines [13,14]. We searched the PubMed database for original observational studies that described any details on epidemiological characteristics on patients with COVID-19. The database was searched from December 1, 2019, to March 25, 2020. The search was conducted using the following keyword/MESH terms: ((COVID-19[Title/Abstract]) OR coronavirus [Title/Abstract]) OR SARS-CoV-2 [Title/Abstract] OR 2019-nCoV [Title/Abstract]. All studies that compared outcomes of interest in COVID-19 patients were included. Any literature other than observational studies was excluded. Non-English literature, non-full text, and animal studies were excluded. Abstracts were reviewed, and articles were retrieved accordingly. Two independent reviewers performed the search and literature screening (UP, PM), with disputes resolved by consensus following discussion with a third author (CP). For the ease of understanding, we used a flow diagram to describe literature search and study selection process in Figure S1 in the Online Supplementary Document.

Data collection
A prespecified data collection Excel sheet was used to collect the data relating to study characteristics and outcomes of interest by two authors (PM and CP), and discrepancies were solved by a discussion with a third author (UP). The following study characteristics were extracted: publication year, country of origin, sample size, age, sex, direct exposure to infection, travel history, signs and symptoms, risk factors and comorbidities, laboratory and radiology findings, treatment utilized, and complications. Data on the following outcomes were extracted: mortality, recovery, need for ICU beds and mechanical ventilators. VIEWPOINTS RESEARCH THEME 1: COVID-19 PANDEMIC

Statistical analysis
All analysis was done in Excel (Microsoft Inc, Seattle WA, USA) and SAS 9.4 (SAS Institute, Cary, NC, USA). The frequencies and percentages of epidemiological characteristics and outcomes were calculated.

Secondary aim evaluation
We evaluated the global burden of COVID-19 including case fatality rates (CFR), strength of association between deaths and cases to predict CFR, case doubling time, COVID-19 specific mortality rates, and control measures by governments to prevent spread among USA, China, Italy, Iran, Spain, Germany, India, and South Korea. For this purpose, data were taken from the Johns Hopkins University Corona-Virus Resource Center [3], KFF.org [14], World Health Organization-COVID-2019 situation reports [15], and Worldometers.info [2] up until March 31, 2020. We evaluated changes in cases and deaths, CFR, created a predictive modeling for CFR, COVID-19 specific mortality rate, and doubling time for cases and deaths.
CFR was defined as the number of cases divided by the number of the diagnosed patients with COVID-19, and COVID-19 specific mortality rate was defined by deaths due to COVID-19 infections divided by total population of the country in 2020, counted per 100 000 population [16] Pearson correlation coefficient (r) was obtained to establish the strength of association between deaths and cases for individual countries. To predict CFR, we modelled the epidemic curves with simple linear regression, exponential growth, and polynomial growth models and used a coefficient of determination (R 2 ) for model selection.
The time of reporting the first death was used as the starting point for that country for all three models.
We utilized government websites, national media, and other standard open sources to evaluate the governments' interventions during COVID-19 pandemic, infection rate [(diagnosed cases/country' s population in 2020) per 100 000 population] [16] at the time of interventions like nationwide school closure and lockdown, and effects of such measures to predict the dates of peak number of cases in each country.

Strength of association between deaths and cases to predict CFR (Predictive modeling)
Several models, including a simple linear regression, exponential and polynomial (quadratic) growth models, were used to determine the type of association between cumulative deaths and cumulative cases to predict CFR ( Table 3). The polynomial growth model had the best fit (higher R 2 ) and indicates that for all countries the death rate increases with the number of cases, and this increase is steeper than a linear relationship. Interestingly, while for the USA, Italy, Iran, Spain, and India this association is always positive, for China, South Korea, and Germany the initial slope is negative but then is reversed as the number of cases continues to increase (Figure 1).   Figure S3A in the Online Supplementary Document). The daily COVID-19 specific death rate is highest in Spain (daily 1.6 deaths per 100 000 population) and Italy (daily 1.38 deaths per 100 000 population) followed by USA (daily 0.27 deaths per 100 000 population) ( Figure S3B in the Online Supplementary Document).

Doubling time
The county-specific timeline of doubling time for cases and deaths is shown in Table 4 and the increment in cases and deaths are plotted in (Figure 4 in the Online Supplementary Document).  20 March Barred entry of foreign nationals who had been to 28 European countries within last 14 days [46] 22 March Nationwide schools closed [47], Lockdown in New York [45] 27 March A US$ 2 trillion coronavirus stimulus bill was passed and signed by the President [48] 30 March More than half of US states underwent lockdown [45] China: 22 January Response to Public Health Emergency launched by Hubei [49] 23 January The Central government of China imposed a lockdown in Wuhan and other cities in Hubei province; Public transport suspended. The Wuhan airport, railway stations and metro were closed, not allowing residents to leave the city without permission [50]; Public Health Emergency response announced by mainland province of Zhejiang [51] 29 January Mainland China has initiated Public Health Emergency response [52]; Quarantined whole Hubei Province [53]; Curfew laws implemented in Huanggang,Wenzhou and other mainland cities [54] South Korea:

February
An unlicensed Covid-19 test authorized by the Korea Centers for Disease Control and Prevention (CDC) [55]; Travel denied to foreign nationals from Hubei Province into South Korea [56] 23 February All kindergartens, elementary schools, middle schools, and high schools were announced to delay the semester start [57] 26 February Entire country opened drive-through testing [58] Italy: 31 January State of emergency declared, flights to and from China suspended [59] 22 February The Council of Ministers announced a new decree-law to quarantining more than 50 000 people from 11 different municipalities in Northern Italy [60] 4 March Nationwide schools and universities closed [61] 10 March Prime Minister imposed Nationwide quarantine lockdown [62] 11 March All commercial activities except pharmacies and supermarkets ordered to shut down [63]; €25billion allocated by the government [64] 1 April Drive-through testing began [65] Iran:

February
All concerts and other cultural events cancelled for one week by Ministry of Islamic Culture and Guidance [66]; Closure of educational institutions in several cities and provinces announced by the Ministry of Health and medical education [67] 5 March Checkpoints placed between cities to limit travel [68] 16 March Fatima Masumeh Shrine, Jamkaran Mosque in Qom city, and Imam Reza Shrine in Mashhad closed [69] VIEWPOINTS RESEARCH THEME 1:

February
New health security measures enacted to regulate air and sea travel that required passengers from China, South Korea, Japan, Italy and Iran to report their health status before entry [70]; Federal police stepped up checks within 30 km of the border [70] 16 March Bavaria declared a state of emergency for 14 days and measures to limit public movement and additional funds for medicine supplies were introduced [71]; All flights from Iran and China stopped by German Ministry of Transport [72]; Travelling in coaches, attending religious meetings, visiting playgrounds or engaging in tourism prohibited [73] 17 Finance minister announced US$24 billion stimulus package [88] Infection rate at the beginning of the major intervention (nationwide closure of school or major   Table 6 mentions the predicted dates of the peak number of cases based on strict interventions. In China and South Korea, it took 16-21 days and 11-14 days respectively in order to achieve the peak of the pandemic before the new number of cases began to decline. We have used a 16-21 days post-interventional model to calculate the peak of the pandemic keeping in mind the effect of China' s model of interventions.

DISCUSSION
COVID-19 has significantly impacted the entire world both socially and economically. The rapid human-to-human transmission has posed a great public health threat. Across 21 studies included in this review, we found 2747 confirmed cases of COVID-19 with the majority of the published studies from China. 47% of the cases had a history of direct exposure or being exposed to the seafood market in Wuhan, 44% were China residents and 24% had a travel history to China. Initially the virus was limited to only Wuhan and despite travel restriction, the virus continued to spread across the world at a rapid rate from China, likely due to asymptomatic transmission in the initial stages of the outbreak with a median incubation period of only 5 days [7,17], before travel restrictions. The COVID-19 cases are increasing exponentially but underestimated due to mild symptoms in a portion of cases, long incubation periods, and shortage of testing kits. In concurrence with other studies [18,29], we found that clinical characteristics of COVID-19 are similar to those of SARS and influenza virus. Fever (91%), cough (68%) and myalgia or fatigue (48%) were the most prominent symptoms. 24% of patients reported dyspnea and sputum production/expectoration. Major comorbidities were hypertension, smoking, diabetes mellitus, and cardiovascular disease. Patients with these comorbidities are at high risk for complications including pneumonia, ARDS and cardiovascular complications. We found that patients had increased inflammatory markers including elevated CRP in 50%, lymphopenia in 36% and elevated ESR in 25% which is similar to other respiratory infections (SARS, influenza). Few studies [18,91], have reported abnormal liver function in COVID-19 patients, and we found 20% of patients had elevated ALT and AST. Additionally, increased LDH (42%), D-dimer(29%) may indicate the severity of the disease [92]. Some studies have also reported elevated neutrophil count and cytokine storm induced by virus leading to coagulation activation and sustained inflammatory response [22] associated with higher mortality [29].
There is no proven therapy available as of now for COVID- 19 [95]. Large scale clinical trials for these drugs are under way. 50% patients received oxygen and antibiotics (69%), antivirals (49%) and steroids (26%) as supportive therapies. The prognosis of patients after receiving these treatments is not yet clear. In people with compromised immune systems such as older age, HIV, malignancy, diabetes, chronic pulmonary disease if treated promptly with antibiotics, convalescent plasma to increase the immune support might reduce the risk of complications and mortality [96].
In our analysis, 15% of the patients required ICU admission, 15% needed mechanical ventilation, 8% died and 20% recovered and were discharged from the hospital. These findings are consistent with Guan et al. and Wang et al that present similar rates [11,22]. Currently in the USA, COVID-19 is in the acceleration phase surpassing China and Italy, and a National emergency was declared by the President, but the VIEWPOINTS RESEARCH THEME 1:

COVID-19 PANDEMIC
duration and severity may vary depending on the virus characteristics and public health response [97]. If confirmed cases continue to grow with this trend, soon the COVID-19 pandemic will cause shortages of ventilators. As per Institute for Health Metrics and Evaluation (IHME) projections, on a peak day in the USA, there would be a shortage of ICU beds by 19 863 and a need of 31 782 ventilators [98]. The growing number of cases will place a burden on the current capacity of hospitals and hence it is essential to develop and implement strategies to mitigate the gap by increasing capacity and fair allocation of available resources.
As of March 31, CFR in Italy was 11.75% and 4.01% in China. According to Onder et al. [99], CFR stratification by age, shows similar rates for 0-60 years (0%-3.6%) but higher in >70 years(8%-20.2%). This difference might be due to high CFR reported in people >90 years in Italy and no data from China for the same age group [99]. Other reasons might be demographics differences between two countries (≥65 years population: Italy-22.8% vs China-10.9%), overwhelming health care system, and shortage of ICU beds and ventilators, which might lead to prioritizing treatment to younger and otherwise healthy patients over older patient [100]. In our analysis CFR in Italy increased from 1.94% on February 23 to 8.57% on March 20, possibly due to the implementation of a strict policy of testing only suspected cases with severe symptoms [99]. Though widespread and drive-through testing is becoming more available in USA, cumulative tests conducted per million population lags behind compared to Germany, Italy, South Korea, and Spain. Our data driven polynomial growth model predicts more deaths in future with an increase in cases in USA [98], Italy, Iran, Spain, and India. As per our model predictions, doubling time of cases in the USA, Germany and India is decreasing suggesting that they are inching towards the peak. Different countries undertook interventions at different points in the timeline of spread of virus. The infection rates in the USA, Italy, Iran, Spain, and Germany were higher when they undertook substantial measures compared to China, South Korea, and India, suggesting a delayed response and failure to undertake timely measures. The aforementioned timelines for peaks look optimistic because multiple other factors may influence the trajectory of spread, ie, population density, economy, demographics, health care, religious beliefs, and legislation. For instance, despite the growing number of cases, Iran continued to keep its shrines open to pilgrims for a long time, but recently closed them, and no stringent curfew laws were imposed. Also, many states in USA have still not implemented strict quarantine measures. Such practices can seriously impede the efforts at containing the spread and skew the projection in many ways. Restrictions have neither been homogeneously imposed nor simultaneously adopted throughout the country, making it difficult to predict the exact model of the spread.
Also, COVID-19 testing capacity of the nations are limited and the true number of the infected people might have been higher than the estimated numbers at the time of our analysis. Hence, an early phase COVID-19 specific death rate would be a better estimate than CFR to compare the severity of the disease. Many factors contribute to the accurate estimation of CFR such as testing capacity, care seeking and lack of understanding of the proportion of asymptomatic and pre symptomatic cases [101,102]. Limited knowledge of these factors in the early COVID-19 phase might have contributed to overestimation of CFR in our study. The use of serological testing for presence of IgM or IgG antibodies against SARS-CoV-2 will provide a better estimate of cumulative prevalence of COVID 19 infection [103]. As recommended by WHO, measuring the seroprevalence of antibodies to COVID-19 is crucial and will contribute to determine accurate CFR and help plan adequate public health response [104].

Limitations, strengths, and future directions
The research on COVID-19 is rapidly evolving and new publications are becoming available daily. The majority of the epidemiologic data are coming from single center with limited sample sizes. To overcome this limitation and provide a global view of the COVID-19 pandemic, we have analyzed data on over 2500 patients from 21 peer-reviewed studies. As a result, we provided more generalizable estimates of laboratory findings, clinical symptoms and complications of COVID-19 patients. We have included data from several countries/regions; however, one limitation is that the majority of cohorts are from China, and as more data from other countries become available, additional meta-analyses would be essential. This is the first study rigorously tracking the timing of government interventions across multiple countries; however, as mentioned earlier, the adherence to those interventions could vary from one country to another, making the projections of the potential effectiveness challenging. We have not evaluated the duration of strict interventions in all these countries. The population prevalence data are based on the symptomatic patients with confirmed RT-PCR testing. Since some patients can be infected and present mild or no symptoms, or have not undergone RT-PCR testing, serological antibody testing in the future may allow a VIEWPOINTS RESEARCH THEME 1: COVID-19 PANDEMIC more accurate understanding of the disease prevalence and death rates. Despite all the limitations, this is the first study in our knowledge, highlighting and explaining epidemiological indicators, testing capacity, interventions, and expected burden of the COVID 19 at early phase.

CONCLUSIONS
We have reviewed the burden of this pandemic and steps taken by the governments of different countries. Though the governments can continue strict lockdowns, it is not a long-term solution. Good hand hygiene, widespread testing, detection and isolation of new cases, rigorous contact tracing in low-prevalence settings, early vaccine development and its quick distribution, strengthening the overburdened health care system, and protecting frontline health care workers may help to gradually relax the strict lockdowns and cope with COVID-19 pandemic. This would only be possible by a collective approach undertaken by responsible governments, wise strategy implementation, and receptive populations. The future studies should be focused on identifying accurate indicators to mitigate the effect of underestimation or overestimation of COVID-19 burden. Close monitoring of such indicators in highly affected countries is very crucial to evaluate the potential fatality if the second wave of pandemic occurs.