ObjectiveTo assess the feasibility of tracking the prevalence of chronicconditions at the state and community level over time using MDPHnet,a distributed network for querying electronic health record systemsIntroductionPublic health agencies and researchers have traditionally reliedon the Behavioral Risk Factor Surveillance System (BRFSS) andsimilar tools for surveillance of non-reportable conditions. Thesetools are valuable but the data are delayed by more than a year,limited in scope, and based only on participant self-report. Thesecharacteristics limit the utility of traditional surveillance systems forprogram monitoring and impact assessments. Automated surveillanceusing electronic health record (EHR) data has the potential to increasethe efficiency, breadth, accuracy, and timeliness of surveillance. Wesought to assess the feasibility and utility of public health surveillancefor chronic diseases using EHR data using MDPHnet. MDPHnet isa distributed data network that allows the Massachusetts Departmentof Public Health to query participating practices’ EHR data for thepurposes of public health surveillance (www.esphealth.org). Practicesretain the ability to approve queries on a case-by-case basis and thenetwork is updated daily.MethodsWe queried the quarterly prevalence of pediatric asthma, smoking,type 2 diabetes, obesity, overweight, and hypertension statewideand in 9 Massachusetts communities between January 1, 2012 andJuly 1, 2016. We selected these 9 communities because they wereparticipating in a state-funded initiative to decrease the prevalenceof one or more of these conditions. Conditions were defined usingalgorithms based upon vital signs, diagnosis codes, laboratorymeasures, prescriptions, and self-reported smoking status. Eligiblepatients were those with at least 1 encounter of any kind within the2 years preceding the start of each quarter. Results were adjusted forage, sex, and race / ethnicity using the 2010 Massachusetts censusdata.ResultsSurveillance data were available for 1.2 million people overall,approximately 20% of the state population. Coverage varied bycommunity with >28% coverage for 7 of the communities and11% coverage in the eighth. The ninth community had only 2%coverage and was dropped from further analyses. The race / ethnicitydistribution in MDPHnet data was comparable to census datastatewide and in most communities. Queries for all six conditionssuccessfully executed across the network for all time periods ofinterest. The prevalence of asthma among children under 10 yrs rosefrom 12% in January 2012 to 13% in July 2016. Current smoking inadults age≥20 rose from 14% in 2013 to 16% in 2016 (we excludedresults from 2012 due to changes in documentation propelled by theintroduction of meaningful use criteria). This is comparable to the15% rate of smoking per BRFSS in 20141. Obesity among adultsincreased slightly from 22% to 24% during the study period, resultsnearly identical to the most recent BRFSS results for Massachusetts(23% in 2014 and 24% in 2015)2. The prevalence of each conditionvaried widely across the communities under study. For example, forthe third quarter of 2016, the prevalence of asthma among childrenunder 10 ranged from 5% to 23% depending on the community,the prevalence of smoking among adults ranged from 11% to 35%,and the prevalence of type 2 diabetes among adults ranged from7% to 14%. We also examined differences in disease estimates byrace / ethnicity. Substantial racial / ethnic differences were evidentfor type 2 diabetes among adults, with whites having the lowestprevalence at 7% and blacks having the highest at 12% in the thirdquarter of 2016; this trend was consistent over the study period.ConclusionsOur study demonstrates that MDPHnet can provide theMassachusetts Department of Public Health with timely population-level estimates of chronic diseases for numerous conditions at boththe state and community level. MDPHnet surveillance providesprevalence estimates that align well with BRFSS and other traditionalsurveillance sources but is able to make surveillance more timelyand more efficient with more geographical specificity compared totraditional surveillance systems. Our ability to generate real-timetime-series data supports the use of MDPHnet as a source for project/program evaluation.
Characterizing dementia prevalence in the State of Florida: An electronic health record study
