BACKGROUND
Silent carriers were relevant in spreading infection during the coronavirus disease (COVID-19) pandemic. Estimating the prevalence of undocumented cases of COVID-19 has been a significant public health issue since the beginning of the pandemic.
OBJECTIVE
In this work, we propose to modify a commonly used indirect estimation method by assuming undetected COVID-19 cases as a hidden population. The proposed method is based on Bernard’s Network Scale-Up Method (NSUM) and its subsequent modification in a Bayesian framework. The primary aim of this study is to estimate the proportion of undocumented COVID-19 cases in three Italian regions, Veneto, Piemonte, and Lombardia, using a modified NSUM method. The secondary endpoints included estimating COVID-19 cases, people in quarantine, and those who moved between regions after the Italian law act during the earliest pandemic wave.
METHODS
For this purpose, a cross-sectional survey with social networks sampling between 15 April 2020 and 6 May 2020 involving three Italian regions (Lombardia, Piemonte, and Veneto) was conducted. The prevalence of documented and undocumented COVID-19 cases, people in quarantine, and those who moved between regions were estimated. The three methods proposed by Maltiel and based on Network Scale-Up Method: the random degree model (RDM), the barrier effects model (BEM), and the transmission bias model (TBM), were applied. The analysis assumed several scenarios on the average network degree size on the log-normal scale.
RESULTS
The respondents were 1484: 895 (60%) were women with a median age of 39. For all the regions considered, RDM estimates of COVID-19 cases were closer to the official data than those obtained using the other two models. According to the RDM, estimated undocumented cases were higher in Lombardia compared to Piemonte and Veneto (2.78%, 0.44%, and 0.24%, respectively).
CONCLUSIONS
Although there are gold standard methods for detecting the size of undocumented cases, such as mass testing, using an indirect method could still help define the prevalence of a hard-to-reach phenomenon. This modification of NSUM is useful, especially in the early phase of the pandemic when the mass testing procedure was not widespread.
INTERNATIONAL REGISTERED REPORT
RR2-10.3390/ijerph18115713