KerrieWrite a message
- What is my age:
W hen Caitie Bossart returned to the U. A part-time nanny looking for full-time work, she found her inbox filled with messages from companies that had instituted hiring freezes and from families who no longer wanted to bring a babysitter into their homes in response to the spread of COVID When their state issued stay-at-home orders, they decided to hole up together. They ordered takeout and watched movies.
OLE simply takes into intervals between occurrences of a phenomenon and their distribution, irrespective of the type of phenomenon studied. It is important to note that case days represent the time when cases were reported, and not the time of transmission.
It is important to note that underlying assumptions of the OLE are not specific to biological organisms and the species extinction process, and that the method does not contain any biologically specific parameters. Our support the existing evidence and suggest that the first case of COVID would have been sometime between early October and mid-November.
S2 Table. This uncertainty arises due to both the presence of earlier potential COVID cases, and the fact that most phylogenetic analyses put the most recent ancestor at between mid-November and early December, [ 5 ]. As such we specifically date the origin of cases that resulted in the virus taking hold in each country. As the dataset of cases in China does not extend to the first verified cases, we used the dataset presented by Huang et al. For this to be meaningful, it requires sufficient sampling and diversity. Origination dates are discussed for the first five countries outside China and each continent.
For example, in France the first case of COVID was recorded as January 25,however a recent retrospective review of medical records from patients in intensive care unit ICU with both influenza-like illness ILI symptoms and pulmonary ground-glass opacity admitted between December 2,and January 16,14 patients of 58 identified one patient as having COVID who had been presented to the emergency ward on December 27 [ 6 ]. OLE has been widely used in diverse scientific fields, and it is recognized as the most robust approach within that family of methods [ 182428 ]. Uncertainty around origination dates extends beyond the suggested zoonotic overspill in China to all countries where SARS-CoV-2 has spread.
However, it should be noted that Italy was one of the six countries with exceptionally early cases and therefore the result for Italy was affected by the removal of this early case i. Finally, OLE is based on extreme value theory, which shows that the distribution of the maximum is well approximated by the generalised extreme value distribution, regardless of the actual distribution of records [ 192526 ].
Here we repurpose extinction models from conservation science to estimate the potential for earlier cases than has been reported of COVID in countries and territories. Without such analyses it is not possible to determine if our have in fact identified early isolated cases or simply reflect poor surveillance and pre-symptomatic transmission. Evidence is building that its origin as a zoonotic spillover occurred prior to the officially accepted timing of early December, We show that six countries had exceptionally early cases, unlikely to represent part of their main case series.
Download: PPT. Fig 1. From these datasets we created time series of new cases for each country. However, as there is no specific start date as it varies depending on the arrival time of COVID in each country, the 10th case date is used as the end of the period. Further analysis of retrospective testing studies will help validate the application of OLE and associated methods.
Such case dates were removed from the dataset, while maintaining k at between 5 and 10 depending on the of available case dates.
'why would anyone do this?'
Here we use this method to identify cases of COVID that appear not to have taken hold within a country. We applied this test using the first 5 to 10 k earliest case dates of COVID depending on the length of the case record for each country. This suggests an earlier and more rapid timeline of spread.
However, currently only the of retrospective testing have been published for Italy as described above. Finally, dates of records used in the analysis represent mostly the time when a case was reported, not the actual time of transmission and infection, and the of the analysis consequently also represent a hypothetical date when the symptoms of the first, unreported case became manifested.
Using the OLE, origination dates were 4. In the same way the extinction events are rarely observed, so too are origination events such as those of COVID Without rigorous tracing systems, dating the first cases has to be inferred.
If it is assumed the species is extinct, interest then lies in determining when extinction occurred. OLE uses the time series of last known chronological occurrences of the studied phenomenon to estimate the time after the last known occurrence when the process that was generating them has stopped, and the phenomenon will consequently no longer be observable. There are a of exceptionally early cases in specific countries that may have arisen for a of reasons e. It has demonstrated high levels of accuracy of its predictions in the majority of scenarios, especially in case of declining and low record frequency, while its flexibility and non-parametric nature allow its wide application for various data types and conditions [ 1824252829 ].
While an origin as a zoonotic spillover in the Huanan Seafood Market, Wuhan, sometime during early December,has been proposed [ 1 ], this has been called into question [ 2 — 4 ]. The origination date was calculated using the R software package sExtinct [ 27 ]. These exceptionally early cases propagate uncertainty in origination estimators and therefore we applied a dating a chinese Corona guy proposed by Solow and Smith [ 14 ] to identify such cases. Competing interests: The authors have declared that no competing interests exist. The recent t WHO-China study on the global origin of SARS-CoV-2 found that, based on a review of molecular evidence, most point estimates place the most recent ancestor at between mid-November and early December, with a range from late September to early December [ 5 ].
In addition, inferences of the origination date may be potentially affected by multiple COVID introduction events within a country. However, potential multiple introduction events temporally situated within the main cluster of records will not be distinguished by such methods. Moreover, some of the approaches from this group of methods can be applied even to records with just two [ 20 ] or even a single record [ 21 ]. These may represent isolated cases, infections that did not contribute to the eventual spread of COVID through the country or territory.
As the method can be effectively applied to very sparse datasets, with as few as 4—5 records [ 1819 ], it illustrates the potential to rapidly gain an understanding of the origination timings of novel zoonotic diseases when they are poorly known.
This makes it readily applicable to diverse types of phenomena, as long as they are characterized by sporadic records made before the phenomenon or the process ceased [ 10 ]. This method has been ly used in conservation science to determine whether new sightings of the European polecat Mustela putorius in Scotland arose from the native population that was thought to be extirpated or arose from surreptitious reintroduction [ 20 ].
Such a knowledge is critical for our understanding of the spread of this disease.
Furthermore, the differences among countries in implemented control measures, such as travel bans, could also affect inferred spatiotemporal dynamics of COVID spread. Our study provides new approaches for estimating dates of the arrival of infectious diseases based on small samples that can be applied to many epidemiological situations. Map layers were created using the R package rworldmap, Version 1.
In the case of emerging infectious diseases, this is most frequently based on phylogenetic analysis.
It also does not require a complete record, but it s for records being generated based on some unknown probability. We use a method that was originally developed to date the timing of extinction, and turn it to date the timing of origination using case dates rather than sighting events. OLE has been shown to perform well under different rates and trends in sighting effort [ 1824 ], which in this case corresponds to reporting probability. Furthermore, OLE is a non-parametric method and it does not make any assumptions about the sighting rates or data distribution, making it more flexible compared to other methods [ 1925 ].
Suppose that an earlier case of COVID is recorded at time yinterest centres on assessing the exceptionality of the earlier record.
Consequently, if interest lies in inferring the time of first transmission, origin date should be pushed further in the past for the duration of the incubation period. The OLE method has proved to be robust in the inference of extinction under a variety of scenarios, reporting probabilities and trends [ 1824 ].
This suggests a much earlier and more rapid spread than is evident from confirmed cases. Nsoesie et al. PLoS Pathog 17 6 : e This is an open access article distributed under the terms of the Creative Commons Attributionwhich permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Also, is an estimate of the shape parameter of the t Weibull distribution of the k earliest case date times. Funding: I. The funders had no role in study de, data collection and analysis, decision to publish, or preparation of the manuscript.
Using the method of Solow and Smith [ 14 ], we identified six countries with exceptionally early cases of COVID compared with the rest of the case time series for those countries. Within the discipline of conservation science, a of models have been developed to infer or date extinction events based on a series of sightings of a species.
Similarly, from an analysis of 40 composite influent wastewater samples from northern Italy, La Rosa et al. Vertical dashed lines represent mean values. Interest lies in determining whether a species still persists, having not been sighted for a period of time. Supporting information. The method also does not for individual differences among characteristics among records in quality or reliability [ 31 ].
By removing the earliest case date from the record for Yemen the of case dates fell below five and therefore Yemen was not analysed. Testing for exceptionally early cases that failed to take hold There are a of exceptionally early cases in specific countries that may have arisen for a of reasons e. Further, our suggest the most likely timing of the first case to be November 17, This is only 1 day after a case identified in a traveller to Thailand from Wuhan on January 8, [ 1516 ].
China’s errors, dating back to the very first patients, were compounded by political leaders who dragged their feet to inform the public of the risks and to take decisive control measures
While one solution for this issue would be to apply weighted resampling method [ 32 ], which allows OLE to effectively incorporate individual reliability of records, detailed analysis of official covid records necessary to develop such reliability scoring is beyond the scope of this study. In the context of COVID, this method asks the question, given an early case, what is the probability it belongs to the main body of cases? Our suggest that the virus emerged in China in early October to mid-November, the most likely date being November 17and by January,had spread globally.
Additional notable are the estimated dates within Europe. In addition, our study provides a new approach for estimating dates of the arrival of infectious diseases in new areas that can be applied to many different situations in the future. It is important, however, to acknowledge some limitations of the presented approach, related to the input data quality and reliability. While the datasets present the of cases per day, it is not possible to determine whether these cases are independent or related.
Here we applied a well-established extinction estimator i. There are considerable differences among countries and regions in the COVID testing rates, surveillance effort, indicators, reporting systems and criteria, and data quality [ 30 ]. Evidence is building that its origin as a zoonotic spillover occurred before the officially accepted timing of early December, We date the origin of COVID cases from countries and territories using a model from conservation science.
Using methods borrowed from conservation science, we are able to estimate a range of likely dates for the zoonotic spillover of COVID into humans in China and the subsequent spread to countries around the world.
The application of these models has been proposed in a of areas beyond extinction modelling to determine end points, particularly the Optimal Linear Estimation OLE method developed by Roberts and Solow [ 8 ], including geological stratigraphy [ 9 ], archaeology [ 10 ], phenological studies [ 11 ], and phylogenetics [ 12 ].
We therefore used the case days rather than individual cases i. However, in our case we are interested in the timing of origination rather than extinction, so we apply it here with the reverse temporal direction [ 10 ]. Bearing in mind all of the limitations of the COVID records data and the method used in the study, these potential issues need to be taken into consideration when interpreting the .
The basic assumption of Solow and Smith [ 14 ] method is that these represent the k largest values of a larger collection of values generated from a distribution from the Gumbel domain of attraction. Having excluded exceptionally early cases using the method of Solow and Smith [ 14 ], as they likely represent cases where COVID has failed to take hold, we used the first 5 to 10 k earliest confirmed case dates for each country as suggested by Solow [ 19 ] and Rivadeneira et al.
While OLE was demonstrated to be robust to limited data [ 1824 ], as with any methods, here the predictions of the method are only as good as the data used, and the diligence and quality of COVID testing and reporting within different countries is likely to affect our .