key: cord-0800600-b3ax52gu
authors: Raberahona, Mihaja; Rakotomalala, Rado; Rakotomijoro, Etienne; Rahaingoalidera, Tokinandrianina; Andry, Christophe Elody; Mamilaza, Natacha; Razafindrabekoto, Lova Dany Ella; Rafanomezantsoa, Efrasie; Andriananja, Volatiana; Andrianasolo, Radonirina Lazasoa; Razafimahefa, Soloniaina Hélio; Rakotoarivelo, Rivonirina Andy; de Dieu Randria, Mamy Jean
title: Clinical and epidemiological features discriminating confirmed COVID-19 patients from SARS-CoV-2 negative patients at screening centres in Madagascar
date: 2020-11-18
journal: Int J Infect Dis
DOI: 10.1016/j.ijid.2020.11.151
sha: 0b3dc1d51ca713315d8f9ca3765d32e5f6482af5
doc_id: 800600
cord_uid: b3ax52gu

Early and fast detection of COVID-19 patients help limit the transmission and the widespread of the virus in the community and will have impact on mortality by reducing the incidence of infection among vulnerable people. Therefore, community-based screening is critical. We aimed to identify clinical signs and symptoms and epidemiological features that could help discriminate confirmed cases of COVID-19 from SARS-CoV-2 negative patients. We found that age (aOR:1.02, 95%CI:1.02-1.03, p < 0.001), symptoms onset between 3 and 14 days (aOR:1.35, 95%CI:1.09)1.68, p = 0.006), fever or history of fever (aOR:1.75, 95%CI:1.42-2.14, p < 0.001), cough (aOR:1.68, 95%CI:1.31-2.04), sore throat (aOR:0.65, 95%CI:0.49-0.85, p = 0.002), ageusia (aOR:2.24, 95%CI:1.42-3.54, p = 0.001), anosmia (aOR:6.04, 95%CI:4.19-8.69, p < 0.001), chest pain (aOR:0.63, 95%CI:0.47-0.85, p = 0.003), myalgia and/or arthralgia (aOR:1.64, 95%CI:1.31-2.04, p < 0.001), household cluster (aOR:1.49, 95%CI:1.17-1.91, p = 0.001) and evidence of confirmed cases in the neighbourhood (aOR:1.92, 95%CI:1.56-2.37, p < 0.001) could help discriminate COVID-19 patients from SARS-CoV-2 negative. A screening score derived from multivariate logistic regression was developed to assess the probability of COVID-19 in patients. We suggest that a patient with a score ≥ 14 should undergo SARS-CoV-2 PCR testing. A patient with a score ≥ 30 should be considered at high risk of COVID-19 and should undergo testing but also need prompt isolation and contact tracing.

epidemiological features that could help discriminate confirmed cases of COVID-19 from SARS-CoV-2 negative patients. We found that age (aOR:1.02, 95%CI:1.02-1.03, p<0.001), symptoms onset between 3 and 14 days (aOR:1.35, 95%CI:1.09)1.68, p=0.006), fever or history of fever (aOR:1.75, 95%CI:1.42-2.14, p<0.001), cough (aOR:1.68, 95%CI:1.31-2.04), sore throat (aOR:0.65, 95%CI:0.49-0.85, p=0.002), ageusia (aOR:2.24, 95%CI:1.42-3.54, p=0.001), anosmia (aOR:6.04, p<0 .001), chest pain (aOR:0.63, 95%CI:0.47-0.85, p=0.003), myalgia and/or arthralgia (aOR:1.64, 95%CI:1.31-2.04, p<0.001), household cluster (aOR:1.49, 95%CI:1.17-1.91, p=0.001) and evidence of confirmed cases in the J o u r n a l P r e -p r o o f neighbourhood (aOR:1.92, 95%CI:1.56-2.37, p<0.001) could help discriminate COVID-19 patients from SARS-CoV-2 negative. A screening score derived from multivariate logistic regression was developed to assess the probability of COVID-19 in patients. We suggest that a patient with a score ≥ 14 should undergo SARS-CoV-2 PCR testing. A patient with a score ≥ 30 should be considered at high risk of COVID-19 and should undergo testing but also need prompt isolation and contact tracing.

COVID-19; SARS-CoV-2; clinical findings; screening; score; prediction Prompt detection, contact tracing and quarantine of cases are estimated to be highly effective in controlling the transmission and reducing mortality from COVID-19 (Kretzschmar et al. 2020; Nussbaumer-Streit et al. 2020) . Therefore, screening based on clinical features is critical at community level especially in a context of local transmission of the virus. We aimed to assess whether some symptoms and combination of several of them could help discriminate COVID-19 among patients attending 2 screening centres.

We included in this analysis routinely collected data on patients attending the screening centre at the Centre Hospitalier Universitaire Joseph Raseta Befelatanana J o u r n a l P r e -p r o o f (CHUJRB), Antananarivo from May, 6 to July, 1 and from July, 4 to August, 14 for those attending the screening centre at the Centre Hospitalier Universitaire Tambohobe (CHUT), Fianarantsoa. We excluded patients with unknown or inconclusive PCR results. We have also investigated whether the patient lives in a neighbourhood or an area where COVID-19 patients were previously confirmed (neighbourhood) and whether other people living in the same dwelling are symptomatic (household cluster).

We compared clinical and epidemiological features of confirmed cases with those with negative test for SARS-CoV-2 by univariate and multivariate analysis by logistic regression model. We used β-coefficient multiplied by 10 and rounded to the nearest multiple of 2 derived from the logistic regression model to generate a screening score to ascertain the probability of COVID-19 in patients aged≥15 years considering a combination of clinical signs and epidemiological features. The performance of the model was assessed by ROC curve. The sensitivity (Se), specificity (Sp), positive likelihood ratio (LR+) and negative likelihood ratio (LR-) were estimated for each cutoff. Statistical analysis was performed with Stata 14.0 (StataCorp, LP). We collected data on 3154 patients. Overall characteristics and comparison between patients with negative and positive PCR results among those symptomatic were detailed in can be obtained with a cut-off point of 30 but with a Se of 34.4%. A cut-off of 34 will provide a LR+ of at least 5 which can be considered red flags according to J o u r n a l P r e -p r o o f commonly arbitrary definition (Struyf et al. 2020) . The area under the ROC curve was 0.7723 (95%CI: 0.75-0.79). The internal validation using 1000 bootstrap samples from the original dataset found a ROC curve area of 0.7614.

We suggest that a patient with a score ≥ 14 should undergo SARS-CoV-2 PCR testing. A patient with a score ≥ 30 should be considered at high risk of COVID-19 and should undergo testing but also need prompt isolation and contact tracing.

Previous study has shown that prediction models that include routine blood tests in addition to clinical findings are efficient to assess the probability for COVID-19 (Sun et al. 2020 ). However, availability, access and affordability of blood tests limit their use in resource-limited settings. A trade-off between sensitivity and specificity is challenging when considering screening tool for suspect cases. Nevertheless, more sensitive tool is often needed and preferred in an ongoing outbreak. A recent systematic review of signs and symptoms in COVID-19 showed low sensitivity of these signs when taken separately (Struyf et al. 2020) . A recent study in Somalia showed that the current WHO case definition for COVID-19 had only 32.7% (95%CI: 20-48) sensitivity that could be slightly improved when integrating anosmia in the case definition (Ahmed et al. 2020) . Anosmia and ageusia are highly specific of COVID-19 and have the highest scores in the model even if they were present in only 27.3% and 16.2% of patients (La Torre et al., 2020; Liou et al., 2020) . Surprisingly, dyspnoea was not associated with positive SARS-CoV-2 test and was associated with negative test in univariate analysis. Similarly, self-reported contact with confirmed case did not help discriminate SARS-CoV-2 positive patients.

Patients may have exaggerated when reporting symptoms like dyspnoea and other subjective signs or contact with confirmed cases because of panic and fear. The same situation has been observed in previous outbreak in Madagascar (Salam et al., J o u r n a l P r e -p r o o f 2020). In addition, epidemiological link is difficult to identify when community transmission occurs. More objective signs like respiratory rate or measure of Sp02 by simple pulse oximetry that may be helpful in detecting silent hypoxia in COVID-19 were more reliable (Dhont et al. 2020; Jouffroy et al. 2020) .

It is also anticipated that considering neighbourhood as a criterion for screening will be less relevant as the epidemic progresses in the community.

This study had several limitations. We could not assess other type of cluster that may be relevant like occupational clusters. Using level of transmission for each neighbourhood or area by considering attack rate would have been more accurate.

Finally, a prospective external validation of the score is needed.

A screening score based on combination of clinical and epidemiological features could help front-line healthcare workers to classify patients according to their probability of COVID-19.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

https://doi.org/10.1093/cid/ciaa322. 

Evidence for significant COVID-19 community transmission in Somalia using a clinical case definition

The pathophysiology of "happy" hypoxemia in COVID-19

Prehospital pulse oximetry: a red flag for early detection of silent hypoxemia in COVID-19 patients

Impact of delays on effectiveness of contact tracing strategies for COVID-19: a modelling study

Anosmia and Ageusia as Predictive Signs of COVID-19 in Healthcare Workers in Italy: A Prospective Case-Control Study

Alteration of taste or smell as a predictor of COVID-19

Quarantine alone or in combination with other public health measures to control COVID-19: a rapid review

Factors Influencing Atypical Clinical Presentations during the 2017 Madagascar Pneumonic Plague Outbreak: A Prospective Cohort Study

Signs and symptoms to determine if a patient presenting in primary care or hospital outpatient settings has COVID-19 disease

Epidemiological and Clinical Predictors of COVID-19

The authors would like to thank the Ministry of Public Health and the medical staff that has been involved in the screening of COVID-19 patients at the Centre Hospitalier Universitaire Joseph Raseta Befelatanana, Antananarivo, the Centre Hospitalier Universitaire Tambohobe, Fianarantsoa and the Centre Hospitalier Universitaire Andrainjato. We also thank the laboratories that have performed PCR