Table 1. Surveillance System Attributes for Traditional Sources of Surveillance Information and Electronic Health Records (EHRs).

Surveillance system attributes	Traditional national surveillance surveysa		EHRsb
	Strengths	Weaknesses	Strengths	Weaknesses
Timeliness	NA	Can take years between data collection and availability	Available soon after collected	NA
Content and scope	In-depth availability of patient-reported data on behaviors; extensive collection of social determinants of health data	Limited sample sizes, especially for less common sociodemographic groups	Data on millions of patients provides ability to estimate disease prevalence for rare diseases, less common subgroups (Native Hawaiian/Pacific Islander, American Indian/Alaska Native), and small area geographic units and population-based cohorts	Limited availability of patient-reported data; social determinants data availability increasing but limited to insurance type and linked Census data for many EHRs
Structured data; data subjectivity; longitudinal data	Objectively measured health outcomes (vitals, laboratory values) according to study protocol	Cross-sectional or panel designs limit longitudinal follow-up	Longitudinal follow-up on patients allows tracking changes over time; data available on disease control over time	Many data are unstructured (eg, patient notes) and less available for use; structured data standardization is variable; identification of diseases often depends on use of nonspecific diagnostic codes; prescription data typically available but pharmacy dispensing may not be
Representativeness	Nationally representative by design; typically covers entire US population with probability-based sampling strategies	Certain populations can be under-represented (eg, people without a landline telephone, the institutionalized population); characteristics of respondents may differ from nonrespondents in measured or unmeasured ways	Some research networks have data available on people in all US states and territories; patients with multiple types of insurance (commercial and government insurance) are typically available	Representative of care-seeking population, which may limit broad surveillance questions at the population level; representativeness of urban versus rural populations dependent on institutions contributing data
Data quality, completeness	Data collected according to study protocol; robust data completeness and curation	Telephone surveys used in some programs reliant on self-report; all surveys subject to nonresponse	Objective measures of some disease (eg, diabetes, obesity) and robust computable phenotypes of others	Missing data are common; data not collected according to a standardized protocol
Resources required	Infrastructure established by federal agencies to collect data; sampling and weighting strategies well validated and centrally applied by data collectors; some flexibility on adding new questions and data elements	Requires substantial resources and staff to facilitate	Data collected for routine clinical activities and only additional resources for collection required for new data elements	Data processing requires substantial resources, especially to address data quality issues that can arise; adding new data elements challenging

Abbreviation: NA, not applicable.

^a Examples: National Health and Nutrition Examination Survey (NHANES, www.cdc.gov/nchs/nhanes), Behavioral Risk Factor Surveillance System (BRFSS, www.cdc.gov/brfss).

^b Example: National Patient-Centered Clinical Research Network (PCORnet).