公共卫生舆情监测与风险研究综述

舆情监测是国家传染病和突发公共卫生事件监测系统的有力补充。在互联网时代,人们通过网络媒体、社交媒体等方式极其迅速的传播舆情信息,这些信息是非常重要的情报来源,成为预测疾病暴发、早期预警和防灾应急的重要手段。本文系统总结了国内外舆情监测在公共卫生事件监测预警中的研究进展及应用案例,归纳了三种舆情监测的方式,分析了舆情监测相对于传统监测系统的优势,能够有效地在事件发生前预警预测或事件发生后反馈应对。但是,目前舆情监测受到网络使用、媒体引导及搜索习惯的限制,建议未来将基于网络的舆情监测系统整合入已有的监测系统,探索适应我国的舆情监测平台和手段。     

Physical Activity,Sedentary Behavior,and Sleep on Twitter: Multicountry and Fully Labeled Public Data Set for Digital Public Health Surveillance Research

BackgroundAdvances in automated data processing and machine learning (ML) models, together with the unprecedented growth in the number of social media users who publicly share and discuss health-related information, have made public health surveillance (PHS) one of the long-lasting social media applications. However, the existing PHS systems feeding on social media data have not been widely deployed in national surveillance systems, which appears to stem from the lack of practitioners and the public’s trust in social media data. More robust and reliable data sets over which supervised ML models can be trained and tested reliably is a significant step toward overcoming this hurdle. The health implications of daily behaviors (physical activity, sedentary behavior, and sleep [PASS]), as an evergreen topic in PHS, are widely studied through traditional data sources such as surveillance surveys and administrative databases, which are often several months out-of-date by the time they are used, costly to collect, and thus limited in quantity and coverage.ObjectiveThe main objective of this study is to present a large-scale, multicountry, longitudinal, and fully labeled data set to enable and support digital PASS surveillance research in PHS. To support high-quality surveillance research using our data set, we have conducted further analysis on the data set to supplement it with additional PHS-related metadata.MethodsWe collected the data of this study from Twitter using the Twitter livestream application programming interface between November 28, 2018, and June 19, 2020. To obtain PASS-related tweets for manual annotation, we iteratively used regular expressions, unsupervised natural language processing, domain-specific ontologies, and linguistic analysis. We used Amazon Mechanical Turk to label the collected data to self-reported PASS categories and implemented a quality control pipeline to monitor and manage the validity of crowd-generated labels. Moreover, we used ML, latent semantic analysis, linguistic analysis, and label inference analysis to validate the different components of the data set.ResultsLPHEADA (Labelled Digital Public Health Dataset) contains 366,405 crowd-generated labels (3 labels per tweet) for 122,135 PASS-related tweets that originated in Australia, Canada, the United Kingdom, or the United States, labeled by 708 unique annotators on Amazon Mechanical Turk. In addition to crowd-generated labels, LPHEADA provides details about the three critical components of any PHS system: place, time, and demographics (ie, gender and age range) associated with each tweet.ConclusionsPublicly available data sets for digital PASS surveillance are usually isolated and only provide labels for small subsets of the data. We believe that the novelty and comprehensiveness of the data set provided in this study will help develop, evaluate, and deploy digital PASS surveillance systems. LPHEADA will be an invaluable resource for both public health researchers and practitioners.     

Medication sales and syndromic surveillance, France

Although syndromic surveillance systems using nonclinical data have been implemented in the United States, the approach has yet to be tested in France. We present the results of the first model based on drug sales that detects the onset of influenza season and forecasts its trend. Using weekly lagged sales of a selected set of medications, we forecast influenzalike illness (ILI) incidence at the national and regional level for 3 epidemic seasons (2000-01, 2001-02, and 2002-03) and validate the model with real-time updating on the fourth (2003-04). For national forecasts 1-3 weeks ahead, the correlation between observed ILI incidence and forecast was 0.85-0.96, an improvement over the current surveillance method in France. Our findings indicate that drug sales are a useful additional tool to syndromic surveillance, a complementary and independent source of information, and a potential improvement for early warning systems for both epidemic and pandemic planning.     

Use of Commercial Claims Data for Evaluating Trends in Lyme Disease Diagnoses,United States, 2010–2018

We evaluated MarketScan, a large commercial insurance claims database, for its potential use as a stable and consistent source of information on Lyme disease diagnoses in the United States. The age, sex, and geographic composition of the enrolled population during 2010–2018 remained proportionally stable, despite fluctuations in the number of enrollees. Annual incidence of Lyme disease diagnoses per 100,000 enrollees ranged from 49 to 88, ≈6–8 times higher than that observed for cases reported through notifiable disease surveillance. Age and sex distributions among Lyme disease diagnoses in MarketScan were similar to those of cases reported through surveillance, but proportionally more diagnoses occurred outside of peak summer months, among female enrollees, and outside high-incidence states. Misdiagnoses, particularly in low-incidence states, may account for some of the observed epidemiologic differences. Commercial claims provide a stable data source to monitor trends in Lyme disease diagnoses, but certain important characteristics warrant further investigation.     

Trends in Health Information Technology Use Among the US Population With and Without Cardiovascular Risk Factors, 2012-2018: Evidence From the National Health Interview Survey

BackgroundThe COVID-19 pandemic has required clinicians to pivot to offering services via telehealth; however, it is unclear which patients (users of care) are equipped to use digital health. This is especially pertinent for adults managing chronic diseases, such as obesity, hypertension, and diabetes, which require regular follow-up, medication management, and self-monitoring.ObjectiveThe aim of this study is to measure the trends and assess factors affecting health information technology (HIT) use among members of the US population with and without cardiovascular risk factors.MethodsWe used serial cross-sectional data from the National Health Interview Survey for the years 2012-2018 to assess trends in HIT use among adults, stratified by age and cardiovascular risk factor status. We developed multivariate logistic regression models adjusted for age, sex, race, insurance status, marital status, geographic region, and perceived health status to assess the likelihood of HIT use among patients with and without cardiovascular disease risk factors.ResultsA total of 14,304 (44.6%) and 14,644 (58.7%) participants reported using HIT in 2012 and 2018, respectively. When comparing the rates of HIT use for the years 2012 and 2018, among participants without cardiovascular risk factors, the HIT use proportion increased from 51.1% to 65.8%; among those with one risk factor, it increased from 43.9% to 59%; and among those with more than one risk factor, it increased from 41.3% to 54.7%. Increasing trends in HIT use were highest among adults aged >65 years (annual percentage change [APC] 8.3%), who had more than one cardiovascular risk factor (APC 5%) and among those who did not graduate from high school (APC 8.8%). Likelihood of HIT use was significantly higher in individuals who were younger, female, and non-Hispanic White; had higher education and income; were married; and reported very good or excellent health status. In 2018, college graduates were 7.18 (95% CI 5.86-8.79), 6.25 (95% CI 5.02-7.78), or 7.80 (95% CI 5.87-10.36) times more likely to use HIT compared to adults without high school education among people with multiple cardiovascular risk factors, one cardiovascular risk factor, or no cardiovascular risk factors, respectively.ConclusionsOver 2012-2018, HIT use increased nationally, with greater use noted among younger and higher educated US adults. Targeted strategies are needed to engage wider age, racial, education, and socioeconomic groups by lowering barriers to HIT access and use.     

Estimating Cardiorespiratory Fitness Without Exercise Testing or Physical Activity Status in Healthy Adults: Regression Model Development and Validation

BackgroundLow cardiorespiratory fitness (CRF) is an independent predictor of morbidity and mortality. Most health care settings use some type of electronic health record (EHR) system. However, many EHRs do not have CRF or physical activity data collected, thereby limiting the types of investigations and analyses that can be done.ObjectiveThis study aims to develop a nonexercise equation to estimate and classify CRF (in metabolic equivalent tasks) using variables commonly available in EHRs.MethodsParticipants were 42,676 healthy adults (female participants: n=9146, 21.4%) from the Aerobics Center Longitudinal Study examined from 1974 to 2005. The nonexercise estimated CRF was based on sex, age, measured BMI, measured resting heart rate, measured resting blood pressure, and smoking status. A maximal treadmill test measured CRF.ResultsAfter conducting nonlinear feature augmentation, separate linear regression models were used for male and female participants to calculate correlation and regression coefficients. Cross-classification of actual and estimated CRF was performed using low CRF categories (lowest quintile, lowest quartile, and lowest tertile). The multiple correlation coefficient (R) was 0.70 (mean deviation 1.33) for male participants and 0.65 (mean deviation 1.23) for female participants. The models explained 48.4% (SE estimate 1.70) and 41.9% (SE estimate 1.56) of the variance in CRF for male and female participants, respectively. Correct category classification for low CRF (lowest tertile) was found in 77.2% (n=25,885) of male participants and 74.9% (n=6,850) of female participants.ConclusionsThe regression models developed in this study provided useful estimation and classification of CRF in a large population of male and female participants. The models may provide a practical method for estimating CRF derived from EHRs for population health research.     

Incidence Trends for SARS-CoV-2 Alpha and Beta Variants,Finland, Spring 2021

Severe acute respiratory syndrome coronavirus 2 Alpha and Beta variants became dominant in Finland in spring 2021 but had diminished by summer. We used phylogenetic clustering to identify sources of spreading. We found that outbreaks were mostly seeded by a few introductions, highlighting the importance of surveillance and prevention policies.     

Characteristics,Comorbidities, and Data Gaps for Coronavirus Disease Deaths,Tennessee, USA

As of March 2021, coronavirus disease (COVID-19) had led to >500,000 deaths in the United States, and the state of Tennessee had the fifth highest number of cases per capita. We reviewed the Tennessee Department of Health COVID-19 surveillance and chart-abstraction data during March 15‒August 15, 2020. Patients who died from COVID-19 were more likely to be older, male, and Black and to have underlying conditions (hereafter comorbidities) than case-patients who survived. We found 30.4% of surviving case-patients and 20.3% of deceased patients had no comorbidity information recorded. Chart-abstraction captured a higher proportion of deceased case-patients with >1 comorbidity (96.3%) compared with standard surveillance deaths (79.0%). Chart-abstraction detected higher rates of each comorbidity except for diabetes, which had similar rates among standard surveillance and chart-abstraction. Investing in public health data collection infrastructure will be beneficial for the COVID-19 pandemic and future disease outbreaks.