scholarly journals Need of care in interpreting Google Trends-based COVID-19 infodemiological study results: potential risk of false-positivity

Author(s):  
Kenichiro Sato ◽  
Tatsuo Mano ◽  
Atsushi Iwata ◽  
Tatsushi Toda

Introduction: Google Trends (GT) is being used as an epidemiological tool to study coronavirus disease (COVID-19) by identifying keywords in search trends that are predictive for the COVID-19 epidemiological burden. However, many of the earlier GT-based studies include potential statistical fallacies by measuring the correlation between non-stationary time sequences without adjusting for multiple comparisons or the confounding of media coverage, leading to concerns about the increased risk of obtaining false-positive results. In this study, we aimed to apply statistically more favorable methods to validate the earlier GT-based COVID-19 study results. Methods: We extracted the relative GT search volume for keywords associated with COVID-19 symptoms, and evaluated their Granger-causality to weekly COVID-19 positivity in eight English-speaking countries and Japan. In addition, the impact of media coverage on keywords with significant Granger-causality was further evaluated using Japanese regional data. Results: Our Granger causality-based approach largely decreased (by up to approximately one-third) the number of keywords identified as having a significant temporal relationship with the COVID-19 trend when compared to those identified by the Pearson correlation-based approach. 'Sense of smell' and 'loss of smell' were the most reliable GT keywords across all the evaluated countries; however, when adjusted with their media coverage, these keyword trends did not Granger-cause the COVID-19 positivity trends (in Japan). Conclusions: Our results suggest that some of the search keywords reported as candidate predictive measures in earlier GT-based COVID-19 studies may potentially be unreliable; therefore, caution is necessary when interpreting published GT-based study results.

2021 ◽  
Author(s):  
Kenichiro Sato ◽  
Tatsuo Mano ◽  
Tatsushi Toda ◽  
Atsushi Iwata

Abstract Background Google Trends (GT) is being used as an epidemiological tool to study coronavirus disease (COVID-19) by identifying keywords in search trends that are predictive for the COVID-19 epidemiological burden. However, many of the earlier GT-based studies include potential statistical fallacies by measuring the correlation between non-stationary time sequences without adjusting for multiple comparisons or the confounding of media coverage, leading to concerns about the increased risk of obtaining false-positive results. In this study, we aimed to apply statistically more favorable methods to validate the earlier GT-based COVID-19 study results. Methods We extracted the relative GT search volume for keywords associated with COVID-19 symptoms, and evaluated their Granger-causality to weekly COVID-19 positivity in eight English-speaking countries and Japan. In addition, the impact of media coverage on keywords with significant Granger-causality was further evaluated using Japanese regional data. Results Our Granger causality-based approach largely decreased (by up to approximately one-third) the number of keywords identified as having a significant temporal relationship with the COVID-19 trend when compared to those identified by the Pearson correlation-based approach. “Sense of smell” and “loss of smell” were the most reliable GT keywords across all the evaluated countries; however, when adjusted with their media coverage, these keyword trends did not Granger-cause the COVID-19 positivity trends (in Japan). Conclusions Our results suggest that some of the search keywords reported as candidate predictive measures in earlier GT-based COVID-19 studies may potentially be unreliable; therefore, caution is necessary when interpreting published GT-based study results.


2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Kenichiro Sato ◽  
Tatsuo Mano ◽  
Atsushi Iwata ◽  
Tatsushi Toda

Abstract Background Google Trends (GT) is being used as an epidemiological tool to study coronavirus disease (COVID-19) by identifying keywords in search trends that are predictive for the COVID-19 epidemiological burden. However, many of the earlier GT-based studies include potential statistical fallacies by measuring the correlation between non-stationary time sequences without adjusting for multiple comparisons or the confounding of media coverage, leading to concerns about the increased risk of obtaining false-positive results. In this study, we aimed to apply statistically more favorable methods to validate the earlier GT-based COVID-19 study results. Methods We extracted the relative GT search volume for keywords associated with COVID-19 symptoms, and evaluated their Granger-causality to weekly COVID-19 positivity in eight English-speaking countries and Japan. In addition, the impact of media coverage on keywords with significant Granger-causality was further evaluated using Japanese regional data. Results Our Granger causality-based approach largely decreased (by up to approximately one-third) the number of keywords identified as having a significant temporal relationship with the COVID-19 trend when compared to those identified by Pearson or Spearman’s rank correlation-based approach. “Sense of smell” and “loss of smell” were the most reliable GT keywords across all the evaluated countries; however, when adjusted with their media coverage, these keyword trends did not Granger-cause the COVID-19 positivity trends (in Japan). Conclusions Our results suggest that some of the search keywords reported as candidate predictive measures in earlier GT-based COVID-19 studies may potentially be unreliable; therefore, caution is necessary when interpreting published GT-based study results.


2019 ◽  
Vol 25 (6) ◽  
pp. 476-486 ◽  
Author(s):  
Wendy L. Sarver ◽  
Rosanne Radziewicz ◽  
Georgean Coyne ◽  
Kelly Colon ◽  
Lisa Mantz

BACKGROUND: Violence on inpatient psychiatric settings has significant consequences for patients and staff. Research is needed to determine if Brøset Violence Checklist (BVC) is an accurate predictor of violence. AIMS: The study aims were to determine the relationship between BVC scores and incidence of violent behavior within 24 hours, to compare scores among those requiring high-level nursing interventions for violence, and to investigate the impact of scores on length of stay (LOS) and 30 day-readmission rates. METHOD: Retrospective cohort study. RESULTS: Logistic regression indicates 3.4 times greater risk of violence for every additional point on admission BVC (odds ratio = 3.4, 95% confidence interval = [2.29, 5.08], p < .0001). Patients requiring high-level interventions for violence had higher mean BVC scores on both Day 1 and 2 of admission. Pearson correlation was significant for positive association between BVC on admission and LOS ( p < .001). Findings did not establish a link between BVC scores and violence with 30-day readmission rates. CONCLUSIONS: Efforts toward early identification and management of agitation and disruptive behavior is encouraged. Results showed increased risk of violence with every additional point on BVC on admission; further attention should be paid to these patients on admission when using violence screening tools.


2021 ◽  
Author(s):  
Alessandro Rovetta ◽  
Lucia Castaldo

Abstract Alongside the COVID-19 pandemic, the world has had to face a growing infodemic, which has caused severe damage to economic and health systems and has often compromised the effectiveness of infection containment regulations. Although this has spread mainly through social media, there are numerous occasions in which the mass media have shared dangerous information, giving resonance to statements without a scientific basis. For these reasons, infoveillance and infodemiology methods are increasingly exploited to monitor online information traffic. The same tools have also been used to make epidemiological predictions. Among these, Google Trends - a service by GoogleTM that quantifies the web interest of users in the form of relative search volume - has often been adopted by the scientific community. In this regard, the purpose of this paper is to use Google Trends to estimate the impact of Italian mass media on users' web searches in order to understand the role of press and television channels in both the infodemic and the interest of Italian netizens on COVID-19. In conclusion, our results suggest that the Italian mass media have played a decisive role both in the spread of the infodemic and in addressing netizens' web interest, thus favoring the adoption of terms unsuitable for identifying the novel coronavirus (COVID- 19 disease). Therefore, it is highly advisable that the directors of news channels and newspapers be more cautious and government dissemination agencies exert more control over such news.


2020 ◽  
Vol 8 (12) ◽  
pp. 279-288
Author(s):  
Dinsi Stanley Chung ◽  

The Cameroon Development Corporation has been severely affected by the armed conflict in the English speaking (North West and South West regions) part of the country that has been on for close to four years running. How then has the armed conflict in the North West and South West regions of Cameroon affected the Cameroon Development Corporation? How can the growth of the agro-industry be guaranteed? This study looks at the impact of the armed conflict in the North West and South West regions of Cameroon on the Cameroon Development Corporation. Making use of public policy theories, this study establishes a link between government defense strategies/sector development policies and agro-industrial development. The study results show that due to the armed conflict in the English speaking regions, the CDC has incurred major damages including: loss in human capital, drop in production capacity, heavy financial loss and equipment damages. The study results also reveal that, the survival of the CDC depends largely on strategic options to be taken at two separate levels including: political options by conflicting parties - the government and separatist fighters on the one hand, and on the other, options taken by the CDC at both managerial and technical levels. The study concludes that for the CDC to attain structural growth and development that will significantly contribute to the national economy, conflicting.


2020 ◽  
Author(s):  
Solene Huynh ◽  
Tu Anh DUONG ◽  
Guillaume Lame ◽  
Thomas Hubiche ◽  
Khaled Ezzedine

BACKGROUND External factors may hinder Google Trends (GT)’s role as an infodemiology tool. For COVID-19, new symptoms and their searches on internet prior to world organization of health validation were previsouly described. In western country, an unexpected outbreak of cutaneous acral lesion e.g chilblains was released by the dermatologists in April. OBJECTIVE The aim of the study was to assess the temporality between media announcement or lockdown and online searches related to cutaneous acral lesion during COVID-19 outbreak. METHODS Two searches on GT including daily relative search volumes related to 1/ “toe” or “chilblains” and 2/ “coronavirus” were made from January 1st to May 16th 2020 for US, UK, France, Italy, Spain, and Germany. The ratio chilblains to coronavirus was plotted. To assess the impact of lockdown and media coverages, interrupted time series analysis were performed for each country. RESULTS During the period the ratio chilblains /coronavirus searches showed a constant upward trend. In France, Italy and UK, the lockdown was associated to a significant slope change of chilblain searches with coefficient value of 1.06  0.42, 1.04  0.28 and, 1.21  0.44 (p<0.01) respectively. After media announcement, a steep statistically significant increase was found in France, Spain, Italy and the US with coefficient values of 18.95 5.77, 31.31  6.31, 14.57  6.33, 11.24  4.93 (p<0.01), followed by with a statistically significant downward trend in France, Spain and Italy (p<0.01). Adjusted R2 values were 0.311, 0.351, 0.325, 0.305 in France, Spain, Italy and US, in favor of an average correlation between time and research volume. The correlation was weak in Germany and UK. CONCLUSIONS To date chilblains link to COVID-19 remains controversial in the scientific community. However their relative search volumes on GT were highly influenced by lockdown measures and media coverage suggesting caution when using GT as pandemic surveillance tool.


2021 ◽  
Vol 12 ◽  
Author(s):  
Markus Wahrmann ◽  
Bernd Döhler ◽  
Marie-Luise Arnold ◽  
Sabine Scherer ◽  
Katharina A. Mayer ◽  
...  

The functional Fc gamma receptor (FcγR) IIIA polymorphism FCGR3A-V/F158 was earlier suggested to determine the potential of donor-specific HLA antibodies to trigger microcirculation inflammation, a key lesion of antibody-mediated renal allograft rejection. Associations with long-term transplant outcomes, however, have not been evaluated to date. To clarify the impact of FCGR3A-V/F158 polymorphism on kidney transplant survival, we genotyped a cohort of 1,940 recipient/donor pairs. Analyzing 10-year death-censored allograft survival, we found no significant differences in relation to FCGR3A-V/F158. There was also no independent survival effect in a multivariable Cox model. Similarly, functional polymorphisms in two other activating FcγR, FCGR2A-H/R131 (FcγRIIA) and FCGR3B-NA1/NA2 (FcγRIIIB), were not associated with outcome. There were also no significant survival differences among patient subgroups at increased risk of rejection-related injury, such as pre-sensitized recipients (&gt; 0% panel reactivity; n = 438) or recipients treated for rejection within the first year after transplantation (n = 229). Our study results suggest that the earlier shown association of FcγR polymorphism with microcirculation inflammation may not be strong enough to exert a meaningful effect on graft survival.


2020 ◽  
Vol 148 ◽  
Author(s):  
M. D. Walker ◽  
M. Sulyok

Abstract The current coronavirus (COVID-19) pandemic offers a unique opportunity to conduct an infodemiological study examining patterns in online searching activity about a specific disease and how this relates to news media within a specific country. Google Trends quantifies volumes of online activity. The relative search volume was obtained for ‘Coronavirus’, ‘handwashing’, ‘face mask’ and symptom related keywords, for the United Kingdom, from the date of the first confirmed case until numbers peaked in April. The relationship between online search traffic and confirmed case numbers was examined. Search volumes varied over time; peaks appear related to events in the progression of the epidemic which were reported in the media. Search activity on ‘Coronavirus’ correlated well against confirmed case number as did ‘face mask’ and symptom-related keywords. User-generated online data sources such as Google Trends may aid disease surveillance, being more responsive to changes in disease occurrence than traditional disease reporting. The relationship between media coverage and online searching activity is rarely examined, but may be driving online behavioural patterns.


2017 ◽  
Vol 12 (2) ◽  
pp. 80 ◽  
Author(s):  
Aneil Malhotra ◽  
Sanjay Sharma ◽  
◽  

Sudden cardiac death (SCD) in a young person is a rare but tragic occurrence. The impact is widespread, particularly in the modern era of media coverage and visibility of social media. Hypertrophic cardiomyopathy (HCM) is reported historically as the most common cause of SCD in athletes younger than 35 years of age. A diagnosis of HCM may be challenging in athletes as pathological hypertrophy of the left ventricle may also mimic physiological left ventricular hypertrophy (LVH) in response to exercise. Differentiation of physiological LVH from HCM requires an array of clinical tools that rely on detecting subtle features of disease in a supposedly healthy person who represents the segment of society with the highest functional capacity. Most studies are based on comparisons of clinical tests between healthy unaffected athletes and sedentary individuals with HCM. However, data are emerging that report the clinical features of athletes with HCM. This article focuses on studies that help shed further light to aid the clinical differentiation of physiological LVH from HCM. This distinction is particularly important in a young person: a diagnosis of HCM has significant ramifications on participation in competitive sport, yet an erroneous diagnosis of physiological adaptation in a young athlete with HCM may subject them to an increased risk of SCD.


2005 ◽  
Vol 37 (5) ◽  
pp. 529-554 ◽  
Author(s):  
mian b. hossain

with a population of over 131 million and a fertility rate of 29·9 per 1000, population growth constitutes a primary threat to continued economic growth and development in bangladesh. one strategy that has been used to cease further increases in fertility in bangladesh involves using family planning outreach workers who travel throughout rural and urban areas educating women regarding contraceptive alternatives. this study uses a longitudinal database to assess the impact of family planning outreach workers’ contact upon contraceptive switching and upon the risk of an unintended pregnancy. using longitudinal data on contraceptive use from the operations research project (orp) of the international centre for diarrhoeal disease research (icddr,b) in bangladesh, multiple decrement life table analysis and multilevel, discrete-time competing risk hazards models were used to estimate the cumulative probabilities of switching to an alternative form of contraceptive use after a woman engaged in a discussion with an outreach worker. after controlling for the effects of socio-demographic and economic characteristics, the analysis revealed that family planning outreach workers’ contact with women significantly decreases the risk of transitioning to the non-use of contraceptives. this contact also reduces the risk of an unintended pregnancy. family planning workers’ contact with women is associated with the increased risk of a woman switching from one modern method to another modern method. the study results indicate that side-effects and other method-related reasons are the two primary reasons for contraceptive discontinuation in rural bangladesh.


Sign in / Sign up

Export Citation Format

Share Document