Skip to main content
American Journal of Public Health logoLink to American Journal of Public Health
editorial
. 2017 Sep;107(9):1365–1367. doi: 10.2105/AJPH.2017.303938

Successes and Continued Challenges of Electronic Health Records for Chronic Disease Surveillance

Guthrie S Birkhead 1,
PMCID: PMC5551610  PMID: 28787206

For many years, the Behavioral Risk Factor Surveillance System (BRFSS) has been a mainstay of public health surveillance for chronic diseases. This telephone survey generates national and state-specific estimates of the prevalence of the major chronic diseases and risk factors, including behaviors, that make up many indicators in Healthy People 2020 and state and local public health improvement plans. Recent innovations in BRFSS methods have been responses to evolving surveillance needs: adding mobile telephone numbers to the sample, imputing measures for the 500 largest cities,1 and fielding follow-up surveys to gather clinical care information for conditions such as asthma. However, BRFSS is limited by the number of questions that can be asked, the self-reported nature of the data, the lack of clinical data, declining response rates, and reductions in funding.

CHRONIC DISEASE SURVEILLANCE

Electronic health records (EHRs) are an important potential source of public health surveillance data. EHRs are used by nearly 90% of ambulatory care providers and contain data on large numbers of people. Their development has been driven by the 2009 Health Information Technology for Economic and Clinical Health (HITECH) Act that provides incentive payments for “meaningful use” of EHRs, including public health reporting. However, despite the billions of dollars disbursed through the program, the development of EHR-based chronic disease surveillance systems has been slow, with few published reports.2

Two recent articles in AJPH show significant progress in using EHRs for chronic disease surveillance. In the June 2017 issue, Perlman et al.3 described the New York City Macroscope, a distributed network of more than 700 providers with a single EHR system that contains records on about 1.5 million people, about one in six New York City residents.3 In the current issue, Klompas et al.4 discuss the Massachusetts MDPHnet system, which is installed on multiple EHRs in three large provider groups covering 1.5 million people, or about one in five state residents. The approaches taken by these two groups have many similarities that illustrate not only the steps necessary to build successful EHR-based surveillance systems, but also the challenges to be dealt with going forward.

ALGORITHM DEFINITION

A critical factor in the development of both projects is that they are collaborations among public health departments, academic centers, and clinicians. Both projects started by selecting the EHR systems to be used and developing algorithms based on the available EHR data elements to define the chronic diseases for surveillance, for example, disease-specific diagnosis codes, laboratory values indicating diseases such as diabetes or hypercholesterolemia, and smoking history. Chart reviews were done to validate some of the disease algorithms at the individual patient level. The algorithms and extraction tools were made available by both projects to the clinical sites to enable them to analyze their own data, an important benefit of the cooperative relationship.

VALIDATION

Both groups also took the important step of comparing their EHR-based surveillance results with known standards. The New York City Macroscope citywide prevalence data for selected diseases were compared with the gold standards of the New York City Health and Nutrition Examination Survey, which collects clinical data directly from a statistical sample of city residents, and of the New York City Community Health Survey, which is equivalent to the BRFSS. Several of the Macroscope measures correlated well with those from the standard data sources. Other measures such as depression prevalence and influenza vaccination agreed less well, possibly because the clinical settings did not have standardized depression assessments and vaccinations given in locations such as pharmacies were not documented in the EHR. The MDPHnet data were examined at the state level and for 13 cities for which comparison data were available from the BRFSS 500 Cities Project. They found good agreement for statewide prevalence measures for diabetes, asthma, smoking, hypertension, and obesity but some variation at the city level, particularly in smaller cities with poorer coverage by the EHR data.

LIMITATIONS

The New York City Macroscope and MDPHnet projects demonstrate the great potential for EHR-based chronic disease surveillance. EHR-based surveillance can access large populations with clinical information including provider diagnosis and laboratory and pharmacy data. The data are available in electronic form, facilitating computer analysis in near real time. However, EHR data are limited to people receiving health care, which may differ from those people not receiving health care. Because the proportion of a population receiving care may depend in part on health insurance coverage, EHR-based surveillance may perform less well in areas with lower insurance coverage rates than New York City or Massachusetts. In addition, MDPHnet data suggest that the validity of small-area estimates may vary depending on the proportion of the local population covered by the EHR and the number of participating providers. Small-area estimates may be affected if the participating EHRs cover only a small portion of the population. Ultimately, this concern should be mitigated as more EHRs are added to surveillance systems, increasing the proportion of the population under surveillance. Further work is necessary to determine the impact of these factors and also to validate EHR-based estimates for racial/ethnic subgroups. Finally, EHR-based surveillance will not replace BRFSS, which gathers data on health behaviors and self-perceived health status not typically available in EHRs and on people not in care. BRFSS will continue to serve a complementary function to EHR-based surveillance.

REPLICATION

How adaptable are the New York City and Massachusetts methods to other jurisdictions? Certainly the algorithm definition and validation process will help guide efforts elsewhere. However, jurisdictions implementing EHR-based chronic disease surveillance will likely have to customize their surveillance algorithms to the varying EHRs available for surveillance and analyze their sensitivity and specificity through chart review. This burden can, one hopes, be reduced in the future as broadly applicable algorithms are validated and become available. This goal should be advanced by a provision of the 21st Century Cures Act that calls for the development and use of Application Programming Interfaces (APIs).5 Developing public health surveillance APIs will be a critical step in helping health departments access data from EHRs. Similarly, jurisdictions with regional health information organizations linking together EHRs may be able to use these networks for surveillance purposes.6

Health departments undertaking EHR-based chronic disease surveillance will also need to compare the resulting disease and risk factor prevalence estimates with standards such as BRFSS. This may present a challenge for local jurisdictions outside the 500 Cities data, which encompass only one third of the US population. Expanded county-level BRFSS can provide standards for comparison in some states.

CONTINUED CHALLENGES

The foundational steps described in the two articles mentioned previously are necessary but not sufficient to establish EHR-based surveillance systems at this time when public health resources are under stress. Both author groups note that their EHR-based surveillance systems are cost effective. That may be true after the systems are set up and validated, but many jurisdictions may have difficulty finding the resources to replicate this work. State and local health departments have received only a fraction of the resources that the clinical sector received under the HITECH meaningful use program. More resources will be needed to fully establish EHR-based surveillance systems.7 Continued collaboration between public health and clinical settings, which can access HITECH resources, will also be critical.

EHR-based surveillance is still in its infancy. Perlman et al.3 and Klompas et al.4 show the way forward. However, replicating their models across the country will be challenging. Standardizing methods and algorithms and sharing technological solutions will certainly be helpful. However, for EHR-based surveillance to become a reality, the public health community must stay involved and must advocate for crucial resources.

Footnotes

See also Klompas et al., p. 1406.

REFERENCES

  • 1.Centers for Disease Control and Prevention. 500 cities: local data for better health. Available at: https://www.cdc.gov/500cities. Accessed May 18, 2017.
  • 2.Birkhead GS, Klompas M, Shah NR. Uses of electronic health records for public health surveillance to advance public health. Annu Rev Public Health. 2015;36:345–359. doi: 10.1146/annurev-publhealth-031914-122747. [DOI] [PubMed] [Google Scholar]
  • 3.Perlman SE, McVeigh KH, Thorpe LE, Jacobson L, Greene CM, Gwynn RC. Innovations in population health surveillance: using electronic health records for chronic disease surveillance. Am J Public Health. 2017;107(6):853–857. doi: 10.2105/AJPH.2017.303813. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Klompas M, Cocoros NM, Menchaca JT et al. Can electronic health record based public health surveillance systems provide accurate estimates of state and local chronic disease rates? Am J Public Health. 2017;107(9):1406–1412. doi: 10.2105/AJPH.2017.303874. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Mandl KD, Kohane IS. A 21st-century health IT system—creating a real-world information economy. N Engl J Med. 2017;376(20):1905–1907. doi: 10.1056/NEJMp1700235. [DOI] [PubMed] [Google Scholar]
  • 6.Phillips AB, Wilson RV, Kaushal R, Merrill JA HITEC Investigators. Implementing health information exchange for public health reporting: a comparison of decision and risk management of three regional health information organizations in New York state. J Am Med Inform Assoc. 2014;21(e1):e173–e177. doi: 10.1136/amiajnl-2013-001716. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Leider JP, Shah GH, Williams KS, Gupta A, Castrucci BC. Data, staff, and money: leadership reflections on the future of public health informatics. J Public Health Manag Pract. 2017;23(3):302–310. doi: 10.1097/PHH.0000000000000580. [DOI] [PubMed] [Google Scholar]

Articles from American Journal of Public Health are provided here courtesy of American Public Health Association

RESOURCES