Corrao Giovanni, Bonaugurio Andrea Stella, Bagarella Giorgio, Maistrello Mauro, Leoni Olivia, Cereda Danilo, Gori Andrea
University of Milano-Bicocca, Milan, Italy; National Centre for Healthcare Research and Pharmacoepidemiology, University of Milan-Bicocca, Milan, Italy; Welfare Department, Operative Centre for Health Data, Lombardy Region, Milan, Italy.
National Centre for Healthcare Research and Pharmacoepidemiology, University of Milan-Bicocca, Milan, Italy; Biostatistics, Epidemiology and Public Health Unit, Department of Statistics and Quantitative Methods, University of Milano-Bicocca, Milan, Italy.
J Infect Public Health. 2025 Feb;18(2):102621. doi: 10.1016/j.jiph.2024.102621. Epub 2024 Dec 16.
Large-scale diagnostic testing has been proven ineffective for prompt monitoring of the spread of COVID-19. Electronic resources may facilitate enhanced early detection of epidemics. Here, we aimed to retrospectively explore whether examining trends in the use of emergency and healthcare services and the Google search engine is useful in detecting Severe Acute Respiratory Syndrome Coronavirus outbreaks early compared with the currently used swab-based surveillance system.
Healthcare Utilization databases of the Italian region of Lombardy and the Google Trends website were used to measure the weekly utilization of emergency and healthcare services and determining the volume of Google searches from 2020 to 2022. Improved Farrington algorithm (IMPF) and Exponentially Weighted Moving Average (EWMA) control chart were both fitted to detect outliers in weekly searches of nine syndromic tracers. AND/OR Boolean operators were tested aimed for joint using tracers and models. Signals that occurred during periods labelled as free from epidemics were used to measure positive predictive values (PPV) and false negative values (FNV) in anticipating the epidemic wave.
Out of the 156 weeks of interest, 70 (45 %) were affected by epidemic waves. Overall, 54 syndromic signals were obtained from any one of the 7 healthcare or Google tracers, generating an outlier from both the EWMA and IMPF models. PPV values of 0.95, 1.00, 0.96 admitting a delay of 0, 1, and 2 weeks, respectively, between signal and epidemic wave. The values of FNP ranged from 0.19 to 0.21.
High predictive power for anticipating COVID-19 epidemic waves, even two weeks ahead of the official reports, was obtained from electronic syndromic tracers of healthcare-seeking trends and Google search engine use. Following verification via a prospective approach, public health organizations are encouraged to take advantage of this free forecasting system to anticipate and effectively manage respiratory outbreaks.
大规模诊断检测已被证明对迅速监测新冠病毒病（COVID-19）的传播无效。电子资源可能有助于加强对疫情的早期发现。在此，我们旨在回顾性探索，与目前使用的基于拭子的监测系统相比，检查急诊和医疗服务的使用趋势以及谷歌搜索引擎是否有助于早期发现严重急性呼吸综合征冠状病毒疫情。
利用意大利伦巴第地区的医疗保健利用数据库和谷歌趋势网站，来测量2020年至2022年期间急诊和医疗服务的每周使用情况，并确定谷歌搜索量。采用改进的法林顿算法（IMPF）和指数加权移动平均（EWMA）控制图，以检测九种症状追踪指标每周搜索中的异常值。测试了AND/OR布尔运算符，旨在联合使用追踪指标和模型。在标记为无疫情的时间段内出现的信号，用于测量预测疫情波时的阳性预测值（PPV）和假阴性值（FNV）。
在156周的研究期内，70周（45%）受到疫情波影响。总体而言，从7种医疗保健或谷歌追踪指标中的任何一种获得了54个症状信号，EWMA和IMPF模型均产生了异常值。信号与疫情波之间分别延迟0、1和2周时，PPV值分别为0.95、1.00、0.96。FNP值范围为0.19至0.21。
通过对就医趋势的电子症状追踪指标和谷歌搜索引擎使用情况，可获得预测COVID-19疫情波的高预测能力，甚至比官方报告提前两周。经前瞻性方法验证后，鼓励公共卫生组织利用这一免费预测系统来预测并有效管理呼吸道疫情。
