PCORnet, the National Patient-Centered Clinical Research Network,1 is a research ecosystem of highly representative real-world data from electronic health records (EHRs) and partnerships with patients to ensure that PCORnet-enabled research addresses the needs, priorities, and diversity of patient communities. PCORnet is a national consortium of 79 health systems encompassing over 13,000 individual hospitals and clinics that participate in 8 Clinical Research Networks (CRNs). In addition, the PCORnet Coordinating Center supports data, research, and engagement activities of the network. CRNs, the Coordinating Center, and PCORnet governance include Patient Partners to identify and prioritize research questions, health outcomes, and data needs that are important to patients, their caregivers, and families as well as providers and payers. In addition to health outcomes, cost-related factors are important in patients’ health care experiences. This article reviews opportunities for PCORnet to enable research that considers patient-centered economic outcomes by augmenting its infrastructure with sources of economic data, including administrative claims from Medicare, Medicaid, and commercial health plans; patients’ expenditures such as copayments, coinsurance, deductibles, and other out-of-pocket costs; and area-level and individual-level social and financial factors. These data, often lacking in EHRs, inform outcomes and questions that matter to patients, such as how interventions or payment policies impact their cost of care in addition to their health outcomes. For some patients and families, economic outcomes may be as salient as health outcomes due to financial constraints, which underscores the importance of studying patient-centered economic outcomes in PCORnet-enabled research and generating evidence to inform health care that best meets patients’ needs in terms of both health and economic factors.

PCORnet CRNs use a distributed architecture with sites’ curated data from EHRs stored in the PCORnet Common Data Model (CDM)2 and all datasets residing at the sites. This baseline architecture is illustrated in Figure 1 following the conventions described in the 2021 Food and Drug Administration (FDA) guidance3 regarding the use of Real-World Data. This distributed architecture support local health system concerns regarding access control but has the limitation that investigators do not have direct access to study-specific datasets which may limit analytic approaches and complicate data quality assessment.

FIGURE 1:

PCORnet distributed data architecture. CDM indicates Common Data Model; GPC, Greater Plains Collaborative; PCORnet, Patient-Centered Clinical Research Network.

Driven by research priorities identified in collaboration with patients, the network has formed partnerships with other data sources, such as commercial health plans,4 federal insurance claims,5 and disease registries,6 to provide more comprehensive representation of patients’ health and lived experiences. By expanding the data linkage capacity, addressing regulatory barriers, and building trust for data sharing with partner institutions, we enhance the infrastructure’s ability to incorporate the types of data that patients identify as important to their lived experiences, thereby enabling research that better addresses patients’ needs.

The 2010 Authorizing Law that established the Patient-Centered Outcomes Research Institute (PCORI), the primary funder of PCORnet, did not include economic outcomes within the scope of research that PCORI was authorized to fund. Thus, PCORI’s comparative effectiveness studies led to a better understanding of clinical benefits, harms, and patient-centered outcomes but did not address the economic impact of those choices on overall patient well-being. However, the authorizing law was amended in 2019 to explicitly state that patient-centered outcomes shall include the “potential burdens and economic impacts of medical treatments and services” on different stakeholders and decision-makers as outlined in PCORI’s published “Principles for the Consideration of the Full Range of Outcomes Data in PCORI-Funded Research,”7 thereby allowing PCORI-funded studies to incorporate patient-centered economic outcomes in primary research aims. These legislative changes further motivate PCORnet participants to enhance research capacity by: (1) augmenting infrastructure for data linkage; and (2) engaging patients and pursuing partnerships with relevant data sources to identify and fulfill data needs for studying patient-centered economic outcomes. As PCORnet participants, it is our hope that this commentary can galvanize collaboration with the health economics research community by presenting initial opportunities for the network to incorporate economic data, informed by patient and stakeholder perspectives.8 The first 3 sections describe enhancements that have been incorporated to date, and the remaining sections discuss potentials to capture holistic data that include cost and other economic variables for analysis in patient-centered outcomes research (PCOR).

INDIVIDUAL-LEVEL DATA LINKAGE INFRASTRUCTURE

Linking patients represented in a PCORnet site’s curated CDM with other individual-level sources, such as insurance claims, has 2 main advantages: (1) ascertaining health care received and outcomes that occur outside the PCORnet partner health system; and (2) incorporating data not available from the site’s CDM such as insurance payments to the provider. PCORnet CRNs have established the ability to perform such linkage without sharing personally identifiable information outside of network health systems by implementing privacy-preserving record linkage (PPRL) technology network-wide provided by Datavant.9 Queries of the distributed data network generally do not resolve, or de-duplicate, the identity of a patient who has been treated by 2 or more participating health care systems so that they are counted once instead of 2 or more times in analyses. Instead, the PPRL infrastructure enables matching and linkage of patient records for research purposes, both among multiple health care systems and with commercial health plans to aggregate EHR and claims data. As shown in Figure 2,10 PCORnet partners have also linked with data from the Centers for Medicare and Medicaid Services (CMS) by submitting finder files of patient identifiers (upper left) appended with a unique “Site Coded ID,” which are used by Medicare’s honest broker contractor to match patients with beneficiaries; generating a cross-walk file that enables analyses of the linked EHR data and the financial activity reflected in claims records as a limited data set. Further advancements to PCORnet linkage mechanisms may include: (1) improving matching accuracy through referential patient matching which “uses large collections of demographic records, such as information from credit reporting agencies or address change records, providing a multirecord benchmark to match identities”11; and (2) embedding PPRL within cloud-based data ecosystems using external tokenization12 which enables data linkage when analyses are conducted, thereby reducing the costs and delays associated with generating PPRL before analyses.

FIGURE 2:

PCORnet Clinical Research Network linkage process with Medicare Claims. CMS indicates Centers for Medicare and Medicaid Services; EHR, electronic health record; PCORnet, Patient-Centered Clinical Research Network.

While some CRNs maintain linked databases, such infrastructure may have limited scalability for organizations using an older, on-site data infrastructure because it can become prohibitively expensive to maintain as the volume of data increases. Other CRNs have standing regulatory agreements with various data partners that enable linkage on a project-specific basis. The data remain at the respective institutions until there is a request for a subset of data to be shared and linked for a given research purpose. Standing agreements may simply be amended for each use case, while the main terms of the agreement are already established. These arrangements are relatively inexpensive to maintain and represent trust among organizations that underscores long-standing, sustainable partnerships.13 This reusable regulatory framework enables efficient data sharing and linkage without requiring substantial infrastructural investment and maintenance. The regulatory framework, developed in conjunction with patient advisors, addresses patient privacy and security concerns throughout.

COMMERCIAL HEALTH PLAN PARTNERSHIPS

CRNs conduct research through data linkage with health plans as exemplified by REACHnet’s (one of the PCORnet CRNs) partnerships with 3 commercial health plans—Humana, CVS Health, and Blue Cross Blue Shield of Louisiana (BCBS-LA). Regulatory agreements and PPRL infrastructure are in place to link claims with EHR data on a study-specific basis to enhance the completeness of outcomes and include economic data exemplified by a study of the impacts of health plan benefits on clinical and cost-related type II diabetes14 outcomes. REACHnet linked clinical data from 3 PCORnet participating health systems in Louisiana with claims from BCBS-LA. The primary aim is to evaluate BCBS-LA’s zero-dollar copayment benefit for generic prescription drugs on medication adherence, hemoglobin A1c control, out-of-pocket spending, and overall medical costs. While A1c measures are obtained from EHR data, medication dispensing information is incomplete in EHRs15 so pharmacy claims are used to assess medication adherence. Out-of-pocket and overall medical costs are also evaluated from the health plan data. This study directly addresses patients’ prescription costs as a potential barrier to medication adherence and A1c control in diabetes management. Standing partnerships between PCORnet and health plans enable opportunities to evaluate the impacts of benefit design on patients’ day-to-day clinical outcomes.

MEDICARE AND MEDICAID LINKAGE EXAMPLES

CMS makes data available for research supported by the Research Data Assistance Center (ResDAC).16 Some CRNs also work with state offices to access Medicaid claims more rapidly than through ResDAC.17 Several PCORnet CRNs have conducted projects that link CMS claims with the PCORnet CDM following ResDAC’s processes as outlined in a recent PCORI white paper.18 An example of how Medicare data has been used to study patient-centered economic outcomes is the Louisiana Experiment Addressing Diabetes Outcomes (LEAD study) linking clinical data from 3 PCORnet participant health systems in Louisiana with Medicare claims to evaluate reimbursement for non–face-to-face chronic care management (NFFCCM) services, a CMS policy instituted in 2015.14 The analyses found that use of NFFCCM was not associated with additional copayments for Medicare beneficiaries but was associated with a reduction in total medical expenditures.

The Greater Plains Collaborative (GPC) CRN uses current ResDAC processes to populate a cloud-based data-sharing environment5 for multistate CMS claims, site PCORnet CDMs, tumor registries, and external sources (Fig. 3). The GPC study5 also describes the complementary nature of EHRs and claims in that diagnoses codes in claims underreport obesity in comparison to body mass index consistently collected in EHRs, while common comorbidities such as diabetes are underreported in individual health systems EHRs in comparison with claims. Cloud-based data sharing enables dynamic linkage to CRN data enabling investigators to access study-specific data in a secure environment, reduce the storage and time required to replicate claims at each distributed site thus lowering costs thereby facilitating PCOR to improve care for CMS beneficiaries. To address privacy concerns about linking claims and EHR data, CMS as well as federal funders are increasingly requiring researchers to adopt standardized federal information security practices5 available consistently from cloud service providers. Additional avenues for CMS claims integration include the CMS Virtual Data Research Center (VDRC) and the Data at the Point of Care19 pilot Application Programmer Interface (API) for Medicare Claims.

FIGURE 3:

Greater Plains Collaborative GROUSE architecture for claims integration leveraging CMS claims and Snowflake cloud-based data sharing. ACS indicates American Community Survey; CDM, Common Data Model; CMS, Centers for Medicare and Medicaid Services; GPC, Greater Plains Collaborative; HUD, US Department of Housing and Urban Development; PCORnet, Patient-Centered Clinical Research Network. Blue * indicate is the Snowflake database company's icon.

HEALTH SYSTEM BILLING-BASED BENCHMARKING

Since the 1990s adult20 and pediatric21 health systems have engaged in clinical and financial benchmarking; including consistent length of stay/mortality models, financial comparators, the Joint Commission core measures, and other patient characteristics important to learning health systems. Federally Qualified Health Centers have also studied22,23 Medicaid expansion’s impact upon these practices devoted to underserved communities encompassed by the ADVANCE CRN. Extending benchmarking with other data can further economic analyses. For example, linking the Visient Clinical Database to area-level measures revealed significant disparities in access to chimeric antigen receptor T-cell therapy.24 The underlying billing systems that support Visient benchmarking also record patient copays and out-of-pocket expenses and could provide consistent measures of financial burden instead of patient survey-based estimates.25

AREA-LEVEL MEASURES

Health outcomes are partially associated with the environment in which patients live, and the PCORnet CDM stores patients’ addresses at various levels of resolution (county, 5 and 9 digit postal codes, street address, and detail) and address changes over time to support geographic linkage (ie, geocoding26). Publicly available community-level data provides a way to characterize social and economic factors affecting patients27 (eg, social deprivation indices incorporating Census tract median incomes) as used in a study of the effect of the Affordable Care Act on diabetes care.28 PCORnet participating sites linked with the US Census Bureau’s American Community Survey (ACS) data29,30 and may also leverage more recent investments31 such as the Agency for Healthcare Research and Quality’s (AHRQ) database on Social Determinants of Health (SDOH)32 and environmental exposure databases on water,33 soil,34 air,35 and climate36 released by Environmental Protection Agency (EPA). Economic variables at the zip code or census tract level provide contextual information for studying associations between health outcomes and socioeconomic factors of day-to-day life in communities, like computer and internet access/use, labor force participation, and home ownership. Bidirectional relationships may exist between SDOH and health status, whereby social and economic factors impact health outcomes, but health status may also affect economic outcomes like the ability to participate in the workforce.

INDIVIDUAL-LEVEL SOCIAL DETERMINANTS OF HEALTH AND PATIENT-REPORTED DATA

For decades, database marketing companies have sold individual consumer and household insights37 for targeted marketing but have recently developed social determinants of health packages.38 Personal or household attributes include income, home/car ownership, educational attainment, and health/personal care/diabetes purchasing interest and expenditures in these categories. Credit reports are also linked to understand medical debt.39,40 These insights hold promise for more accurate SDOH than area-level measures and also provide insights regarding family and household characteristics. Family linkages are not consistently recorded and extracted from EHRs although a current effort within PCORnet is resolving mother-baby linkage to support PCORI’s maternal morbidity and mortality research priority. While health systems and payors are integrating individual-level SDOH data into practice, their accuracy across diverse patient populations is largely unreported in peer-reviewed literature warranting research into their suitability for use in PCOR.

Patient-reported data is arguably the most patient-centered of all data sources, particularly if the information collected is identified by the patient community as especially salient to their health care and economic experiences and needs, such as out-of-pocket costs and financial stresses upon the household. However, the health care industry lacks a large-scale, systematic, and standardized way of collecting these data. Recording individual-level social and financial determinants in the EHR may provide relevant, patient-oriented, standardized measures for use in pragmatic comparative effectiveness research.41 In conjunction with AHRQ, the Gravity Project42 is developing standard representations of SDOH data and terminology to support consistency and sharing of measures between EHR systems. This direct integration of standardized and systematically collected patient-focused economic factors into EHR data would be especially advantageous for incorporating into an EHR-based infrastructure like PCORnet as it could support real-time economically informed shared decision-making with patients during their care.

PHARMACY BENEFITS AND RETAIL PHARMACY DATA

Drug exposure is fundamental to PCOR. Dispensing information via claims (CMS, commercial noted above) can also be obtained from Pharmacy Benefits Managers connected to Surescripts (a company that enables e-prescribing) which in turn can directly update EHRs43 and support individual-level linkage. However, which health systems receive Surescripts updates is not consistently characterized across PCORnet sites and if addressed would improve our ability to understand drug exposure, adherence, medication safety, and costs. The National Retail Data Mart (NRDM)44,45 is a complementary area-level resource underutilized for PCOR. National retailers provided geocoded data used by health departments and the Center for Disease Control to confirm the level of disease for ongoing and outbreaks of COVID and influenza. Linking to these open-source repositories46 would enable PCORnet to use data on out-of-pocket costs to assess patient-centered economic outcomes.

LIMITATIONS AND PRIVACY CONSIDERATIONS

Ultimately, comparative effectiveness research using data derived from available EHR data may suffer from uncertainties affecting confidence in results.47,48 The necessary data transformations may censor incomplete records in a systematic way (eg, claims analyses exclude the uninsured/underserved). The data sources may contain gaps in coverage in a systematic way based on funding models and business priorities. Linkages to community-based datasets for social and economic outcomes through the various privacy-preserving methods are often incomplete and may magnify existing biases in the source data. Another area of ongoing concern is the impact of record linkages on the privacy of individuals, groups of individuals, and participating health systems. Ongoing efforts by research teams associated with PCORnet participants are addressing these issues in a systematic way.9,28 It is incumbent on the PCORnet research community to clarify these limitations and invest in engagement efforts to advance trustworthy research that might now holistically evaluate health choices in relation to the financial stability and livelihoods of our patients, their families, and communities.

CONCLUSIONS

PCORnet is designed to attract stakeholders and collaborators that can use its infrastructure to conduct research addressing factors that matter most to patients’ health care decisions and experiences. Initially, PCORnet capitalized on the rapid adoption of EHRs incentivized by the Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009 which had led to nationally available interoperable clinical data.49 In addition to clinical factors, patient-centered economic outcomes may be relevant to the interests of patients, their families, and health care providers. The enhancements discussed in this commentary highlight ways in which PCORnet can enable research that includes patient-centered economic outcomes. We see this as an opportunity for the health economics research community to inform our infrastructure’s evolution and partner with PCORnet in conducting studies that include salient economic factors in PCOR that improves our national health care system.50

ACKNOWLEDGMENTS

The authors acknowledge funding from the Office of the Assistant Secretary for Planning and Evaluation (ASPE) to support travel to the Symposium on Building Data Capacity to Study Economic Outcomes for Patient-Centered Outcomes Research that was held on December 5, 2022, in Washington, DC, and provided an opportunity for the authors to present their work and receive feedback from attendees as well as Erin Holve, Casey Quinn, and Kim Marschhauser from PCORI for review.

REFERENCES 1. Fleurence RL, Curtis LH, Califf RM, et al. Launching PCORnet, a national patient-centered clinical research network. J Am Med Inform Assoc. 2014;21:578–582. 2. Qualls LG, Phillips TA, Hammill BG, et al. Evaluating foundational data quality in the national patient-centered clinical research network (PCORnet®). EGEMS. 2018;6:3. 3. U.S Food and Drug Administration. Real-World Data: Assessing Electronic Health Records and Medical Claims Data To Support Regulatory Decision-Making for Drug and Biological Products, Draft Guidance for Industry; September 2021. Docket Number: FDA-2020-D-2307, Issued by: Center for Drug Evaluation and Research, Center for Biologics Evaluation and Research, Oncology Center of Excellence. Accessed November 1, 2022. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/real-world-data-assessing-electronic-health-records-and-medical-claims-data-support-regulatory 4. Ma Q, Chung H, Shambhu S, et al. Administrative claims data to support pragmatic clinical trial outcome ascertainment on cardiovascular health. Clin Trials. 2019;16:419–430. 5. Waitman LR, Song X, Walpitage DL, et al. Enhancing PCORnet Clinical Research Network data completeness by integrating multistate insurance claims with electronic health records in a cloud environment aligned with CMS security and privacy requirements. J Am Med Inform Assoc. 2022;29:660–670. 6. Carnahan RM, Waitman LR, Charlton ME, et al. Exploration of PCORnet data resources for assessing use of molecular-guided cancer treatment.. JCO Clin Cancer Inform. 2020;4:724–735. 7. Patient-Centered Outcomes Research Institute. Principles for the consideration of the full range of outcomes data in PCORI-funded research; 2021. Accessed September 18, 2023. https://www.pcori.org/research/about-our-research/patient-centered-economic-outcomes/principles-consideration-full-range-outcomes-data-pcori-funded-research. 8. Khavjou O, Bradley C, D’Angelo S, et al. Landscape review and summary of patient and stakeholder perspectives on value in health and health care; August 2022. Prepared by RTI International under RTI Project No. 0218249.005. 9. Kiernan D, Carton T, Toh S, et al. Establishing a framework for privacy-preserving record linkage among electronic health record and administrative claims databases within PCORnet®, the National Patient-Centered Clinical Research Network. BMC Res Notes. 2022;15:337. 10. Dusetzina SB, Huskamp HA, Rothman RL, et al. Many Medicare beneficiaries do not fill high-price specialty drug prescriptions. Health Aff (Millwood). 2022;41:487–496. 11. Grannis SJ, Williams JL, Kasthuri S, et al. Evaluation of real-world referential and probabilistic patient matching to advance patient identification strategy. J Am Med Inform Assoc. 2022;29:1409–1415. 12. Snowflake Documentation. Understanding External Tokenization. 2022. Accessed November 1, 2022. https://docs.snowflake.com/en/user-guide/security-column-ext-token-intro.html 13. Forrest CB, McTigue KM, Hernandez AF, et al. PCORnet® 2020: current state, accomplishments, and future directions. J Clin Epidemiol. 2021;129:60–67. 14. Shao Y, Stoecker C, Hong D, et al. The impact of reimbursement for non-face-to-face chronic care management on health utilization among patients with type 2 diabetes in Louisiana. Value Health. 2022;26:676–684. 15. Lin PD, Daley MF, Boone-Heinonen J, et al; PCORnet Antibiotics and Childhood Growth Study Group. Comparing prescribing and dispensing data of the PCORnet common data model within PCORnet Antibiotics and Childhood Growth Study. EGEMS (Wash DC). 2019;7:11. 16. About Us. Accessed November 1, 2022. https://resdac.org/about-resdac 17. Hogan WR, Shenkman EA, Robinson T, et al. The OneFlorida Data Trust: a centralized, translational research data infrastructure of statewide scope. J Am Med Inform Assoc. 2022;29:686–693. 18. Dullabh P, Heaney-Huls K, Leaphart D, et al. Expanding access to Medicare and Medicaid claims data across PCORnet® Clinical Research Networks. Patient Centered Outcomes Research Institute; 2022. 19. Centers for Medicare and Medicaid Services. Data at the Point of Care. 2022. Accessed November 1, 2022. https://dpc.cms.gov/ 20. Nguyen NT, Silver M, Robinson M, et al. Result of a national audit of bariatric surgery performed at academic centers: a 2004 University HealthSystem Consortium Benchmarking Project. Arch Surg. 2006;141:445–450. 21. Tanaka ST, Grantham JA, Thomas JC, et al. A comparison of open vs laparoscopic pediatric pyeloplasty using the pediatric health information system database—do benefits of laparoscopic approach recede at younger ages? J Urol. 2008;180:1479–1485. 22. Huguet N, Springer R, Marino M, et al. The impact of the Affordable Care Act (ACA) Medicaid expansion on visit rates for diabetes in safety net health centers. J Am Board Fam Med. 2018;31:905–916. 23. Larson AE, Hoopes M, Angier H, et al. Private/marketplace insurance in community health centers 5 years post-affordable care act in medicaid expansion and non-expansion states. Prev Med. 2020;141:106271. 24. Ahmed N, Shahzad M, Shippey E, et al. Socioeconomic and racial disparity in chimeric antigen receptor T cell therapy access. Transplant Cell Ther. 2022;28:358–364. 25. Narang AK, Nicholas LH. Out-of-pocket spending and financial burden among Medicare beneficiaries with cancer. JAMA Oncol. 2017;3:757–765. 26. Krieger N, Chen JT, Waterman PD, et al. Geocoding and monitoring of US socioeconomic inequalities in mortality and cancer incidence: does the choice of area-based measure and geographic level matter? the Public Health Disparities Geocoding Project. Am J Epidemiol. 2002;156:471–482. 27. Using the American Community Survey: Benefits and Challenges. The National Academies Press; 2007. 28. Furmanchuk A, Liu M, Song X, et al. Effect of the Affordable Care Act on diabetes care at major health centers: newly detected diabetes and diabetes medication management. BMJ Open Diabetes Res Care. 2021;9(suppl 1):e002205. 29. Gardner BJ, Pedersen JG, Campbell ME, et al. Incorporating a location-based socioeconomic index into a de-identified i2b2 clinical data warehouse. J Am Med Inform Assoc. 2019;26:286–293. 30. Block RG, Puro J, Cottrell E, et al. Recommendations for improving national clinical datasets for health equity research. J Am Med Inform Assoc. 2020;27:1802–1807. 31. Dullabh P, Hovey L, Leaphart D, et al. Expanding Social Determinants of Health Data across PCORnet® Clinical Research Networks. Patient Centered Outcomes Research Institute; 2022. 32. Agency for Healthcare Research and Quality. Social Determinants of Health Database. 2020. Accessed November 1, 2022. https://www.ahrq.gov/sdoh/data-analytics/sdoh-data.html 33. Environmental Protection Agency. National Water Information System (W). 2015. Accessed November 1, 2022. https://www.epa.gov/waterdata/waters-geospatial-data-downloads 34. United States Department of Agriculture. Soil Survey Geographic Database (S). 2022. Accessed November 1, 2022. https://gdg.sc.egov.usda.gov/GDGHome_DirectDownLoad.aspx 35. Environmental Protection Agency. U. S. E. P. Air Quality System (AQS). 2021. Accessed November 1, 2022. https://aqs.epa.gov/aqsweb/documents/about_aqs_data.html#id2 36. National Centers for Environmental Information (NCEI) Dataset. 2015. Accessed November 1, 2022. https://www.ncei.noaa.gov/access/search/dataset-search 37. Acxiom. Infobase. 2022. Accessed November 1, 2022. https://www.acxiom.com/customer-data/infobase/ 38. LexisNexis Risk Solutions. Social Determinants of Health. 2022. Accessed November 1, 2022. https://risk.lexisnexis.com/healthcare/social-determinants-of-health 39. Kluender R, Mahoney N, Wong F, et al. Medical debt in the US, 2009-2020. JAMA. 2021;326:250–256. 40. Shankaran V, Li L, Fedorenko C, et al. Risk of adverse financial events in patients with cancer: evidence from a novel linkage between cancer registry and credit records. J Clin Oncol. 2022;40:884–891. 41. Chen M, Tan X, Padman R. Social determinants of health in electronic health records and their impact on analysis and risk prediction: a systematic review. J Am Med Inform Assoc. 2020;27:1764–1773. 42. Health Level Seven International (HL7). The Gravity Project. 2018. Accessed January 18, 2023. http://www.hl7.org/gravity/ 43. Blecker S, Adhikari S, Zhang H, et al. Validation of EHR medication fill data obtained through electronic linkage with pharmacies. J Manag Care Spec Pharm. 2021;27:1482–1487. 44. Wagner MM, Robinson JM, Tsui FC, et al. Design of a national retail data monitor for public health surveillance. J Am Med Inform Assoc. 2003;10:409–418. 45. Wagner MM, Tsui FC, Espino JU, et al. The emerging science of very early detection of disease outbreaks. J Public Health Manag Pract. 2001;7:51–59. 46. Wagner M, Tsui F, Cooper G, et al. Probabilistic, decision-theoretic disease surveillance and control. Online J Public Health Inform. 2011;3:ojphi.v3i3.3798. 47. Weber GM, Adams WG, Bernstam EV, et al. Biases introduced by filtering electronic health records for patients with “complete data”. J Am Med Inform Assoc. 2017;24:1134–1141. 48. Klann JG, Joss M, Shirali R, et al. The Ad-Hoc uncertainty principle of patient privacy. AMIA Jt Summits Transl Sci Proc. 2018;2017:132–138. 49. Adler-Milstein J, Jha AK. HITECH Act drove large gains in hospital electronic health record adoption. Health Aff (Millwood). 2017;36:1416–1422. 50. Berwick DM, Nolan TW, Whittington J. The triple aim: care, health, and cost. Health Aff (Millwood). 2008;27:759–769.