Table 2:

Surveillance system attributes evaluated across the potential national surveillance data sources

Attribute	NHIS	MEPS	NHANES	NAMCS	NEHRS	NHCS	EHRs
Usefulness	Has already provided prevalence estimates and baseline data for surveillance and HP2030.	MEPS is the US standard for estimating nationally representative healthcare expenditure and therefore ideal data source for estimating costs of high-impact pain. Difficult to assess chronic pain without linking to NHIS.	Generic pain questions have not been administered since the 2011–2012 survey. Unable to assess chronic pain or high-impact chronic pain.	Useful for surveillance of pain issues, such as pain management, for those using the healthcare system.	This is a physician survey, so pain questions are not reported at the individual-level. The survey is useful for assessing pain management and prescribing trends in the health care setting.	Useful to assess opioid- and pain-related hospital visits, potentially as a result of differences in pain management practices.	The nonexistence of data standards for electronic health record or medical claims data make it difficult to perform national chronic pain surveillance at this time.
Simplicity	Uses existing data collection infrastructure and standardized survey methodology. Data can be easily obtained from survey website and interactive data tool46.	Uses existing data collection infrastructure and standardized survey methodology, but its complexity, especially when linking to NHIS, is substantial.	Uses existing data collection infrastructure and standardized survey methodology. Data can be easily obtained from survey website.	Uses existing data collection infrastructure and standardized survey methodology. Data can be easily obtained from survey website.	Uses existing data collection infrastructure and standardized survey methodology. Data can be easily obtained from survey website.	Uses existing data collection infrastructure and standardized survey methodology. Data can be obtained from NCHS Research Data Center.	The fragmented nature of the US healthcare system makes the use of EHRs for national surveillance complex.
Flexibility	Moderately flexible. With some lead time the ability exists to add sponsored emerging health topics as needed; many additional pain questions were added for 2019 and later surveys.	Unknown.	Moderately flexible. With lead time the ability exists to add sponsored emerging health topics as needed.	Administers standard questions, survey has not been redesigned recently. Limited ability to add emerging health topics.	Moderately flexible. With lead time the ability exists to add sponsored emerging health topics as needed.	Moderately flexible. Since survey includes a short questionnaire and submission of all diagnosis codes, NHCS can quickly adapt to changing information needs.	Difficult to add questions in EHRs nationally, differences in EHR vendors complicate the process.
Data quality	Chronic pain and highimpact chronic pain questions were developed and cognitively tested prior to being added to the survey. Potential for recall bias.	Pain interference question has been shown to discriminate across different levels of health status and health care utilization. Potential for recall bias.	Prescription medication data obtained by NHANES is high quality based on collection procedures.	Validation of diagnosis codes for assessment of persistent, nonmalignant pain has not been performed.	The frequency with which physicians prescribe controlled substances in this questionnaire has not be validated.	Validation of diagnosis codes for assessment of persistent, nonmalignant pain has not been performed.	Poor data quality, issues with inaccuracies, missingness and no data standardization specific to chronic pain surveillance.
Acceptability	Question refusal rate for CP and HICP are low, only 0.05% and 0.01% respectively, in 2016. Total household response rate was 67.9% in 2016.	Survey response rate for the public use MEPS files was 46.0% in 2016.	Acceptability is declining, in the 2015–2016 survey 61.3% of persons asked to do the interview completed the survey.	Participation rates for eligible physicians who completed at least one patient record form was 39.3% in 2016.	The overall unweighted response rate of the 2015 survey was 51.9%.	~20% of the 581 hospitals sampled during the 2013–2016 survey years participated.	Acceptability of EHRs in health care setting across the U.S. is growing.
Representa tiveness	National; representative of the U.S. civilian, noninstitutionalized population. Does not include individuals in institutional care facilities, therefore may underrepresent older population.	Derives from prior year’s NHIS; National; representative of the U.S. civilian, noninstitutionalized population. Does not include individuals in institutional care facilities, therefore may underrepresent older population.	National; representative of the U.S. civilian noninstitutionalized population. Does not include individuals in institutional care facilities, therefore may underrepresent older population.	National; representative of physician visits in the U.S.	Survey data are nationally representative of physicians practicing in U.S. health care settings.	Response rates are not high enough to produce nationally representative data.	Representative of individuals who seek health care within specific EHR catchment areas; not nationally representative.
Timeliness	Survey conducted annually; survey data are made available ~nine months after survey completion.	Survey conducted annually and begins one year after NHIS completion. Survey data are made available ~nine months after survey completion.	Survey data are released biennially; no generic pain questions administered since the 2011–2012 survey.	Survey conducted annually; survey data are not available for use in a timely manner	Survey conducted annually; survey data are made available ~12–18 months after survey completion.	Survey conducted annually; survey data are made available ~two years after survey completion.	Rapid, EHR data is available in near real-time.
Stability	Established in 1957 and repeated annually since. Well-accepted standard approaches to the survey and data preparation exist.	Established in 1997 and has been repeated annually since. Survey data are highly consistent over time, however there are redesigns that may affect comparability of data.	Established in 1999 and has been repeated biennially since. Pain questions are intermittently administered.	Established in 1973. Survey is now annual and has remained stable since 2006 when a subset of community health centers was added. There was a transition from ICD-9 to ICD-10 codes in 2016.	Established in 2008. Became an independent survey in 2012 and has been administered annually since.	Established in 2013. Survey continues to be administered regularly. There was a transition from ICD-9 to ICD-10 codes around 2015.	EHRs are continually evolving and there is now widespread adoption. There was a transition from ICD-9 to ICD-10 codes around 2015.

^*Positive predictive value and sensitivity were unable to be assessed due to the non-existence of a gold standard for pain measurement. HP2030 = Healthy People 2030; NHIS = National Health Interview Survey; MEPS = Medical Expenditure Panel Survey; NHANES = National Health and Nutrition Examination Survey; NAMCS = National Ambulatory Medical Care Survey; NCHS = National Center for Health Statistics; NEHRS = National Electronic Health Records Survey; NHCS = National Hospital Care Survey; EHRs = electronic health records.