Abstract
PURPOSE
Health care systems increasingly incorporate social risk data into electronic health records (EHRs) to address needs like food, housing, and transportation insecurity. This study evaluated whether EHR-integrated social clinical decision support (SCDS) tools improved control of blood pressure (BP) and hemoglobin A1c (HbA1c) and increased social risk–informed care and documentation in community-based clinics.
METHODS
We conducted a cluster randomized trial in a large primary care network. This pragmatic trial was designed to assess tool impact in real-world clinic conditions. Six clinics received SCDS tools embedded in the EHR; 44 clinics served as controls. The tools supported clinic-wide workflows and targeted decision support. A screening alert was triggered for adult patients lacking up-to-date social risk screening. Additional components were activated for patients with uncontrolled hypertension or diabetes, or with a diagnosis of either condition combined with a visit no-show rate of at least 50%. Primary outcomes were BP and HbA1c control. Secondary outcomes included social risk screening and documentation. Generalized linear mixed models accounted for patient clustering. We also examined use patterns of individual tool components.
RESULTS
Blood pressure control improved over 12 months in both arms, with significantly greater improvement in intervention clinics. Control of HbA1c showed no significant differences. Intervention clinics had significantly greater odds of social risk screening and documentation. Use of individual SCDS tool components varied widely across clinics.
CONCLUSION
Access to EHR-integrated SCDS tools was associated with increased documentation of social risks and greater improvements in BP control. These findings support embedding social risk data into clinical workflows to enhance chronic disease management in primary care.
Key words: clinical decision support, disease management, electronic health records, social determinants of heatlh
INTRODUCTION
Health care organizations increasingly recommend documenting adverse social determinants of health (ie, social risks1) such as food, housing, and transportation insecurity in electronic health records (EHRs).2-4 These recommendations are based on consistent and compelling evidence that links social risks and health outcomes,5-8 evidence which also influenced a National Academies of Sciences, Engineering, and Medicine report that outlined potential clinic-led approaches to reducing the impacts of social risks on health.9 These approaches included connecting patients to social services (ie, assistance) and tailoring care plans to accommodate social risks to improve the patient’s ability to follow their plan (ie, adjustment); for example, by prescribing a lower-cost medication alternative when a patient cannot afford their current prescription.9 While many assistance programs have been well studied, less is known about how social risk data supports clinical care adjustments. Clinical decision support (CDS) tools, which have improved care in other areas,10-12 may enable such adjustments when integrated with social risk data–forming social clinical decision support (SCDS) tools.
In the Contextualized Care in Community Health Centers’ Electronic Health Records study,13 we developed and tested SCDS tools designed to alert health care teams to patients’ social risks and facilitate care plan adjustments for patients with uncontrolled blood pressure (BP) or diabetes. The purpose of the tools was to increase care team awareness of social barriers to care plan adherence, then prompt treatment adjustments such as medication selection and follow-up care scheduling. The intention was that such adjustments would enhance patients’ ability to adhere to care plans and thus improve disease management outcomes. Tools were developed through an iterative partner-engagement process14 and refined after pilot testing in 3 community-based clinics.15 The final tools were designed to increase social risk screening, document Z codes related to housing and economic circumstantces, identify social risk–related barriers to medication adherence, and document care plan adjustments in response to social risks. International Classification of Diseases, 10th Revision (ICD-10) diagnosis codes used were the Z59 codes and relevant Z91 codes (ie, Z91.189 and Z91.120).
This study is among the first to evaluate the impact of SCDS tools on social risk–informed care and patient outcomes. This is the first study to examine these effects in safety-net community health centers, which primarily serve populations facing social and economic barriers.
METHODS
Setting
This study was conducted in the OCHIN network which includes more than 2,100 community-based clinics in 36 states. Since 2016, OCHIN clinics have documented social risk data in a shared EHR system.16-20 As of June 2025, more than 3.7 million screenings have been documented for at least 1.9 million patients, with 32% reporting at least 1 social risk. Screenings are typically completed by patients during clinic visits, with staff support as needed.
Design
This pragmatic trial was conducted in OCHIN clinics active for 12 months or more as of July 1, 2021 that provided primary care for adult patients and had 200 or more social risk screenings in the EHR—indicating capacity to collect and record social risk data. Clinics involved in the SCDS pilot were excluded. Clinics were stratified by screening volume and recruited sequentially until 6 intervention clinics were enrolled; 44 eligible clinics that did not implement the tools served as controls (Supplemental Appendix 1).
Due to the grant budget, only 6 clinics implemented the tools. Patients were not randomized; patients from clinics that activated the tools were included in the intervention group. Each clinic activated the tools on their own timeline, with no standard protocol, and integrated tools into existing workflows—reflecting real-world conditions and pragmatic trial principles.
Intervention
The SCDS tools were turned on in the intervention clinics at start dates ranging from February 9, 2023 through September 15, 2023, as determined by clinic preferences and recruitment (Supplemental Table 1). For the 44 control clinics, a random start date from February 3, 2023 through September 15, 2023 was assigned to align with the intervention period. Each clinic was followed for 12 months.
Study Population and Tool Functions
The SCDS tools were designed to support care for adult patients (aged 18 years or older) and appeared in the EHR to assist with both clinic-wide workflows and targeted clinical decision support. Tool alerts were activated for individual patients based on specific patient characteristics.
The Social Determinants of Health Screening Alert was activated for adults with outdated social risk screenings. Screening was considered current if any financial security–related domain (eg, housing, food, transportation, utilities) had been assessed in the past 12 months. This alert appeared for all users, as a pop-up for support staff and as an EHR alert for clinicians.
Other components of the tool were activated for patients meeting specific criteria in the prior year, including uncontrolled hypertension (ie, systolic BP of 140 or more, or diastolic BP of 90 or more), uncontrolled diabetes (ie, HbA1c of 9.0% or greater), or a diagnosis of hypertension or diabetes combined with a visit no-show rate of 50% or more. For these patients the following tool components were activated:
Z code alert recommended adding relevant social risk ICD-10 diagnosis codes to the patient’s medical record based on screening results and enabled 1-click entry for users.
Medication adherence alert notified users when adherence to prescribed medications had not been documented, and allowed for the recording of social risk–related reasons for why medications were not being taken as recommended.
In-line medication alerts appeared during medication ordering for eligible patients with positive social risk screenings and addressed issues such as food insecurity, affordability of branded medications, patient preferences for supply duration, and communication with pharmacists to reduce medication costs.
Documentation checklist supported rapid documentation of care team discussions about care plan adjustments related to social risks by offering a tailored checklist of potential discussion topics based on the patient’s medical and social context.
For additional details on tool functionality, see Supplemental Appendix 2 and previously published tool descriptions.15
Outcomes
Primary outcomes were BP control in patients with hypertension (ie, systolic BP less than 140 mmHg; diastolic BP less than 90 mmHg); and HbA1c control in patients with diabetes (ie, HbA1c less than 9.0%). Binary control measures for BP and HbA1c were selected to align with benchmarks and facilitate interpretation. Outcomes were only assessed if a patient with hypertension had at least 2 clinic visits or a patient with diabetes had HbA1c measured during at least 2 clinic visits within 12 months after SCDS tool activation. The first visit marked the start of the observation period and the second provided follow-up data for assessing changes in BP or HbA1c control. This ensured that patients had exposure to the intervention and enabled a standardized comparison of biomarker changes over time.
Secondary clinical outcomes included mean systolic BP, mean diastolic BP, and mean HbA1c assessed at the patient level using values from the same 2 clinic visits. These continuous outcomes enabled evaluation of biomarker changes across the full distribution. Models were adjusted for the time interval between visits. Additional secondary outcomes included social risk screening and documentation. Screening completion was defined by the presence of documented results during visits for patients due for annual screening. For patients reporting social risks, we examined whether risks were documented in the encounter diagnosis or the problem list, and whether a relevant Z code was added to the EHR or removed if a risk was resolved.
Descriptive Analyses
We assessed SCDS tool use in intervention clinics by tracking alert frequency and related care plan adjustments. For eligible visits, we report how often the alerts appeared in intervention clinics and how often they would have appeared in control clinics based on patient eligibility. Across all clinics, we evaluated documentation of social risk screening, Z code additions or removals, and medication adherence barriers. In intervention clinics, we also tracked checklist use and in-line medication alerts. Table 1 summarizes SCDS tool functions and corresponding outcomes.
Table 1.
Tool Component Logic and Associated Outcome Measures
| Tool | Logic | Outcome measure |
|---|---|---|
| Screening due alerta | For any patient whose last social risk screening was conducted >12 months ago or never conducted | Social risk screening is up to date: patient who was overdue for social risk screening was screened |
| Z code alert to document social risk in the problem lista | Suggests adding or removing Z codes to the problem list or visit diagnosis based on screening results for any patient | Social risk screening documentation: patient with a positive social risk screening but no associated Z code before the visit had Z code added to the problem list or visit diagnosis |
| Medication adherence alerta | For patients with uncontrolled hypertension or diabetes when adherence to prescribed medications is not documented at the visit. Included fields for documenting whether the patient was taking the medication, not taking it (with a reason), or taking it differently (with a reason) | Tool usage frequency in intervention clinics: documentation of patients not taking medications, or taking differently than prescribed, and why |
| In-line medication alertsb | ||
| Alert that medication is not available as a generic | When a newly prescribed outpatient medication is not available as a generic among patients with diabetes and/or hypertension and reported social risk | … |
| Reminder to discuss titrating insulin based on food availability | When a new outpatient insulin order is placed, and the patient is food insecure, among patients with diabetes and/or hypertension and reported social risk | … |
| Prompt for conversation about barriers to taking medication | For a new prescription for diabetes and/or hypertension medication, among patients with diabetes and/or hypertension and reported social risk | … |
| Prompt to consider patient preference for 30- vs 90-day medication | When a new medication has no end date or an end date >30 days in the future, and dispense days are not entered, among patients with diabetes and/or hypertension and reported social risk (does not display for existing medications) | … |
| Note to pharmacy regarding lowering medication costs | For any prescription among patients with diabetes and/or hypertension and reported social risk | … |
| Documentation checklistb | A checklist of adjustments made by the care team to guide clinical discussions and documentation; list of options based on patient-specific factors (ie, BP >140/90, HbA1c ≥9%, missed appointments, and socioeconomic barriers) | … |
BP = blood pressure; EHR = electronic health record; Z code = International Classification of Diseases, 10th Revision diagnosis codes Z59 and relevant Z91 codes related to housing and economic circumstances.
For all clinics, outcome measures were recorded.
For intervention clinics, the in-line medication alerts and the documentation checklist were intended to prompt conversations between patient and clinician. No action is recorded in the EHR, thus no outcome is assessed, but the tool use frequency for the intervention clinics was reported.
Statistical Analysis
We used generalized linear mixed models to evaluate associations between having the SCDS tools and changes in the primary outcomes, secondary clinical outcomes, and other secondary outcomes as feasible. Models accounted for patient clustering within clinics via random intercept and included a fixed effect for the intervention arm. Although this was a randomized controlled trial, statistical adjustment was necessary because randomization does not guarantee perfect balance in baseline characteristics, particularly at the clinic level and with a relatively small sample size. Adjusting for covariates helps reduce residual confounding and improves the precision of estimated intervention effects.21-25 Covariates included in the models were visit type (eg, telemedicine, in-person), age, sex, race, Hispanic ethnicity, language preference, Federal poverty level, insurance status, health care utilization during the study, provider type, and the number of social risk screenings conducted by the clinic before study initiation, to match the clinic sampling frame (Supplemental Appendix 1). For models evaluating BP and HbA1c outcomes, we additionally included a covariate representing the number of weeks between the 2 measurements to account for variation in follow-up timing.
Regulatory and Ethical Oversight
This study was approved by the Kaiser Permanente Interregional Institutional Review Board. All study procedures adhered to ethical standards for research involving human participants. The trial is registered at ClinicalTrials.gov (identification code NCT05022316).
RESULTS
Table 2 presents descriptive patient characteristics by intervention group. The 6 intervention clinics included 9,977 patients; the 44 control clinics included 69,053. Key differences between the groups were more Black/African American patients (28% vs 21%) in the control group, and more Hispanic patients (39% vs 31%) and non-English language preference (37% vs 33%) in the intervention group.
Table 2.
Demographics and Socioeconomics of Patients With an Eligible Visit During the Study Period and Clinic Screening Rates
| Characteristic | Intervention (n = 9,977) No. (%) | Control (n = 69,053) No. (%) |
|---|---|---|
| Age, y | ||
| 18-29 | 490 (4.9) | 4,662 (6.8) |
| 30-49 | 2,871 (28.8) | 19,610 (28.4) |
| 50-64 | 3,885 (38.9) | 25,677 (37.2) |
| ≥ 65 | 2,731 (27.4) | 19,104 (27.7) |
| Sex | ||
| Female | 5,499 (55.1) | 39,146 (56.7) |
| Male | 4,478 (44.9) | 29,850 (43.2) |
| Other | 0 (0.0) | 57 (0.1) |
| Race | ||
| American Indian/Alaska Native | 225 (2.3) | 595 (0.9) |
| Asian | 439 (4.4) | 3,576 (5.2) |
| Black/African American | 2,076 (20.8) | 19,520 (28.3) |
| Multiple Races | 95 (1.0) | 745 (1.1) |
| White | 5,867 (58.8) | 38,600 (55.9) |
| No information | 1,275 (12.8) | 6,017 (8.7) |
| Hispanic Ethnicity | ||
| Yes | 3,857 (38.7) | 21,159 (30.6) |
| No | 5,303 (53.2) | 43,282 (62.7) |
| Unknown | 817 (8.2) | 4,612 (6.7) |
| Language preference | ||
| English | 6,322 (63.4) | 46,468 (67.3) |
| Spanish | 2,600 (26.1) | 17,258 (25.0) |
| Other | 1,055 (10.6) | 5,327 (7.7) |
| Federal poverty level, % | ||
| < 138 | 8,629 (86.5) | 53,151 (77.0) |
| ≥ 138 | 1,140 (11.4) | 12,107 (17.5) |
| No information | 208 (2.1) | 3,795 (5.5) |
| Insurance | ||
| Uninsured | 1,405 (14.1) | 13,400 (19.4) |
| Medicaid | 4,320 (43.3) | 26,706 (38.7) |
| Medicare | 2,716 (27.2) | 17,032 (24.7) |
| Other public | 26 (0.3) | 278 (0.4) |
| Private | 1,510 (15.1) | 11,530 (16.7) |
| No information | 0 (0.0) | 107 (0.2) |
| Clinic screening ratea | ||
| ≥ 600 screens | 6,707 (67.2) | 52,239 (75.7) |
| 400-599 screens | 1,528 (15.3) | 5,253 (7.6) |
| 200-399 screens | 1,742 (17.5) | 11,558 (16.7) |
Clinics were stratified based on their social risk screening rate over the prior 12 months. Although this is a clinici-level variable, it was assigned to each patient to align with patient-level modeling. This section reports the number of patients whose clinics fell into each screening-rate category.
There were 6,233 intervention visits with a social risk screening alert; 32,680 control visits would have triggered this alert based on patient eligibility criteria. Table 3 shows social risk screening was completed in 9% of intervention visits and 11% of control visits. Documentation occurred in 21% of eligible intervention visits involving 2,815 patients compared with 9% of control visits involving 23,468 patients.
Table 3.
Tool Use in Intervention Clinics With Corresponding Comparisons for Control Clinics
| Type of alert | Intervention clinics | Control clinics | ||
|---|---|---|---|---|
| Eligible visits, No.a | Action taken, No. (%)b | Eligible visits, No.a | Action taken, No. (%)b | |
| Social risk screening is due for this patient | 17,732 | 1,512 (8.5) | 74,602 | 7,927 (10.6) |
| Social risk documentation (Z code) is indicated for this patient whose social risk screening results have changed | 6,717 | 1,428 (21.3) | 48,835 | 4,170 (8.5) |
Z code = International Classification of Diseases, 10th Revision diagnosis codes Z59 and relevant Z91 codes related to housing and economic circumstances.
Eligible visits to intervention clinics were those in which the alert fired. In control clinics, eligible visits reflect those where the alert would have fired if the tool had been available.
For social risk screening alerts, an action was considered taken if the clinician completed a social risk screening during the visit in which the tool fired. For of the social risk alert documentation, an action was considered taken if the clinician added or removed a Z code to the problem list or visit diagnosis.
There was no significant difference in BP control between groups at baseline. Control of BP improved over time with a greater improvement for patients in the intervention clinics (Table 4). Unadjusted rates of BP control increased from 58.8% to 61.6% in control clinics and from 56.0% to 60.3% in intervention clinics, aligned with the adjusted model results. For HbA1c control, adjusted models showed no significant differences at baseline, change over time, or intervention effect. Unadjusted rates of HbA1c control increased from 75.8% to 78.8% in control clinics and from 78.1% to 81.7% in intervention clinics, consistent with the adjusted model results. Secondary outcomes showed improvements in mean BP and HbA1c across both groups, with no significant between-group differences (Supplemental Table 2).
Table 4.
Intervention Effects on Blood Pressure and HbA1c Control
| Clinical Outcomes | Fixed effectsa | Clinic random effect, ICC (95% CI)a | |
|---|---|---|---|
| OR (95% CI) | P value | ||
| Blood pressure control | |||
| Intervention group | 0.85 (0.61-1.19) | .35 | 0.04 (0.02-0.06) |
| Follow-up period | 1.14 (1.11-1.17) | <.001 | |
| Intervention group × follow-up period | 1.09 (1.01-1.19) | .04 | |
| Hemoglobin A1c control | |||
| Intervention group | 1.07 (0.69-1.63) | .77 | 0.05 (0.02-0.09) |
| Follow-up period | 1.02 (0.91-1.15) | .68 | |
| Intervention group × follow-up period | 1.06 (0.87-1.29) | .59 | |
ICC = intraclass correlation coefficient; OR = odds ratio.
Estimates from the covariate-adjusted generalized linear mixed models. Each outcome had a separate model.
Table 5 shows intervention clinics had higher odds of completing social risk screenings and documenting social risks. Medication alert activations are summarized in Supplemental Table 3. The documentation checklist was used 215 times in 2 intervention clinics. Medication taken differently than prescribed was documented 31 times, with cost-related reasons recorded only twice using the new tool.
Table 5.
Intervention Effects on Care Process Outcomes
| Care process outcomes | Intervention marginal effecta | Clinic random effect, ICC (95% CI)a | |
|---|---|---|---|
| OR (95% CI) | P value | ||
| Completed social risk screening | 7.3 (1.5-36.0) | .01 | 0.50 (0.33-0.99) |
| Social risks documentation | 11.3 (3.1-40.7) | <.001 | 0.39 (0.25-0.72) |
ICC = intraclass correlation coeffient; OR = odds ratio.
Estimates from the covariate-adjusted generalized linear mixed models. Each outcome had a separate model.
DISCUSSION
This study evaluated the impact of SCDS tools on social risk screening, documentation and patient outcomes in community clinics. To our knowledge, this is the first study to integrate structured social risk data into decision support tools for diabetes and hypertension care in these settings.
Tool availability was associated with significantly higher rates of social risk screening and Z code documentation, suggesting that alerts may enhance social risk documentation, a critical step toward improving outcomes. As screening and documentation are increasingly incentivized, such tools may help clinics meet evolving requirements.
Tool availability was also associated with improved BP control. While the observed association is promising, our study was unable to directly assess causality. We could not determine whether the tools prompted care plan adjustments that facilitated adherence for patients and improved their disease management outcomes as plan adjustments were rarely documented. However, qualitative findings indicated that staff valued the increased visibility of social risk data and felt the data supported clinician autonomy, competence, relationship building, and patient-centered care.26 Further research is needed to determine the extent to which SCDS tools directly impact care plan adaptations and underlying mechanisms of impact.
Despite significant changes in BP control, the intervention was not associated with significant changes in mean BP, likely due to how these outcomes are measured and their sensitivity to change. Blood pressure control is a categorical measure, so small improvements among patients near the threshold can meaningfully affect the proportion classified as controlled, while continuous measures reflect changes across the entire population, including patients with severe hypertension.
No significant changes were observed in HbA1c control or mean HbA1c, possibly because achieving glycemic control is more challenging and often requires multi-component interventions.27 Additionally, HbA1c is a biomarker that reflects average blood glucose over the preceding 3 months. As such, changes in clinical management or patient behavior might not have been detectable during follow-up. Longer-term studies may be needed to observe meaningful changes.
The generally low uptake of the tools specifically designed to facilitate documentation of care adjustments aligns with prior research showing limited use of other CDS tool use, even co-designed tools.28 Here, despite co-design efforts, some clinicians perceived the adjustment documentation tools as redundant or intrusive, particularly in clinics where social risk-informed care is already embedded in the practice. Qualitative feedback indicated that many staff viewed suggested adjustments as routine and found prompts unnecessary, with some disliking prompts for care changes.26 These insights point to a misalignment between tool functionality and clinical context. In community clinics, tools that support nonclinician workflows or automate backend processes may be more impactful than those prompting clinician care changes and documentation practices.29
Our findings contribute to understanding the challenges of operationalizing social risk–informed care, particularly in resource-constrained settings serving populations with multiple barriers to care, and underscore the complexity of designing interventions that effectively address patients’ social and medical needs. Future SCDS design should prioritize seamless integration, minimize cognitive load, and consider tailoring interventions to settings where social risk awareness is less routine or less embedded in care delivery.
Limitations
These findings should be interpreted in the context of several limitations. Selection bias may have influenced results, as only clinics that agreed to participate were included in the intervention arm. Although clinics were stratified by prior screening volume, clinic-level randomization did not ensure balance in patient characteristics, and baseline differences in race, ethnicity, and language preference may have affected outcomes. Additionally, unmeasured clinic-level factors, such as staffing, workflows, and engagement, may have influenced implementation and outcomes.
The pragmatic trial design limited standardization across clinics, contributing to variability in tool usage. Patients were only included if they had at least 2 visits during the study period, potentially biasing the sample toward those with ongoing or moderate health needs. Patients with severe illnesses or improved health may have been excluded due to lack of follow-up. Tool activation at the individual level required documented social risks, meaning patients who did not report a risk, even if they had one, were not flagged by the tools. This introduced potential selection bias, as patients who report social risks may differ systematically from those who do not, possibly due to concerns about stigma, privacy, or uncertainty about how the information will be used.26 Moreover, patients who report social risks may be more engaged with care or have stronger relationships with providers.30
Regression to the mean may have contributed to improvements in both intervention and control groups, since patients were selected based on poor baseline indicators. This statistical phenomenon, where extreme values tend to move closer to the average over time, could have attenuated the observed effect of the intervention, particularly in a small sample with a relatively short follow-up period. The small sample size, in this case constrained by the study budget, limits statistical power and precision, particularly for outcomes like HbA1c control, where null findings may reflect insufficient power rather than absence of effect. The small number of intervention clinics also limits generalizability, as clinic-level factors may have disproportionately influenced results. The study period may have been too short to detect meaningful changes in chronic disease outcomes. Lastly, although clinicians may have regularly adjusted care plans based on social risk, in this study we relied on EHR data, which meant we could only assess documented adjustments. This restricted our ability to evaluate the impacts of the tools on clinical decision-making.
CONCLUSION
Social clinical decision support tools were associated with significant increases in social risk screening, social risk documentation, and BP control. However, this trial provides limited insights into how the tools influenced care team decision making. Future research should explore clinician preferences, ways to enhance clinical impact, and mechanisms through which SCDS tools affect health outcomes.
Acknowledgments
The research reported in this work was powered by PCORnet. PCORnet has been developed with funding from the Patient-Centered Outcomes Research Institute (PCORI) and conducted with the Accelerating Data Value Across a National Community Health Center Network (ADVANCE) Clinical Research Network (CRN). ADVANCE is a Clinical Research Network in PCORnet led by OCHIN in partnership with Health Choice Network, Fenway Health, University of Washington, and Oregon Health & Science University. ADVANCE’s participation in PCORnet is funded through the PCORI Award RI-OCHIN-01-MC. The authors also express their appreciation for the OCHIN member community-based health centers and their staff who tested these clinical decision support tools.
Footnotes
Conflicts of interest: authors report none.
Funding support: Research reported was supported by the National Institute on Minority Health and Health Disparities of the National Institutes of Health under Award Number R01MD014886. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Previous presentations: The preliminary results of the Contextualize Care in Community Health Centers’ Electronic Health Records trial were presented at the NAPCRG Annual Meeting; November 23, 2024; Quebec City, Canada; the Conference on the Science of Dissemination and Implementation in Health; December 9, 2024; Arlington, Virginia; and the SIREN 2025 National Research Meeting; February 4, 2025; San Diego, California.
References
- 1.Alderwick H, Gottlieb LM.. Meanings and misunderstandings: a social determinants of health lexicon for health care systems. Milbank Q. 2019; 97(2): 407-419. doi: 10.1111/1468-0009.12390 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Council On Community Pediatrics . Poverty and child health in the United States. Pediatrics. 2016; 137(4): e20160339. doi: 10.1542/peds.2016-0339 [DOI] [PubMed] [Google Scholar]
- 3.Wyatt R, Laderman M, Botwinick L, Mate K, Whittington J.. Achieving Health Equity: A GuidefFor Health Care Organizations. Institute for Healthcare Improvement White Paper. https://www.ihi.org/library/white-papers/achieving-health-equity-guide-health-care-organizations [Google Scholar]
- 4.Andermann A. Screening for social determinants of health in clinical care: moving from the margins to the mainstream. Public Health Rev. 2018; 39(1): 19. doi: 10.1186/s40985-018-0094-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Adler NE, Stewart J.. Health disparities across the lifespan: meaning, methods, and mechanisms. Ann N Y Acad Sci. 2010; 1186: 5-23. doi: 10.1111/j.1749-6632.2009.05337.x [DOI] [PubMed] [Google Scholar]
- 6.Daniel H, Bornstein SS, Kane GC, et al. ; Health and Public Policy Committee of the American College of Physicians . Addressing social determinants to improve patient care and promote health equity: an American College of Physicians position paper. Ann Intern Med. 2018; 168(8): 577-578. doi: 10.7326/M17-2441 [DOI] [PubMed] [Google Scholar]
- 7.Cole MB, Nguyen KH.. Unmet social needs among low-income adults in the United States: Associations with health care access and quality. Health Serv Res. 2020; 55(Suppl 2)(Suppl 2): 873-882. doi: 10.1111/1475-6773.13555 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kushel MB, Gupta R, Gee L, Haas JS.. Housing instability and food insecurity as barriers to health care among low-income Americans. J Gen Intern Med. 2006; 21(1): 71-77. doi: 10.1111/j.1525-1497.2005.00278.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.National Academies of Sciences, Engineering, and Medicine, Health and Medicine Division, Board on Health Care Services, Committee on Integrating Social Needs Care into the Delivery of Health Care to Improve the Nation’s Health . Integrating Social Needs Care into the Delivery of Health Care to Improve the Nations Health. Published 2019. Accessed Jun 2, 2023. https://www.nationalacademies.org/our-work/integrating-social-needs-care-into-the-delivery-of-health-care-to-improve-the-nations-health
- 10.De Vito Dabbs A, Myers BA, McCurry KR, et al. User-centered design and interactive health technologies for patients. Comput Inform Nurs. 2009; 27(3): 175-183. doi: 10.1097/NCN.0b013e31819f7c7c [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Henderson VA, Barr KL, An LC, et al. Community-based participatory research and user-centered design in a diabetes medication information and decision tool. Prog Community Health Partnersh. 2013; 7(2): 171-184. doi: 10.1353/cpr.2013.0024 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Steele Gray C, Khan AI, Kuluski K, et al. Improving patient experience and primary care quality for patients with complex chronic disease using the electronic patient-reported outcomes tool: adopting qualitative methods into a user-centered design approach. JMIR Res Protoc. 2016; 5(1): e28. doi: 10.2196/resprot.5204 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gold R, Sheppler C, Hessler D, et al. Using electronic health record–based clinical decision support to provide social risk–informed care in community health centers: protocol for the design and assessment of a clinical decision support tool. JMIR Res Protoc. 2021; 10(10): e31733. doi: 10.2196/31733 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Gunn R, Pisciotta M, Gold R, et al. Partner-developed electronic health record tools to facilitate social risk-informed care planning. J Am Med Inform Assoc. 2023; 30(5): 869-877. doi: 10.1093/jamia/ocad010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Pisciotta M, Morrissey S, Bunce A, et al. Help us document what we already do: pilot study of clinical decision support tools targeting social risk-informed care. JAMIA Open. 2025; 8(3): ooaf045. doi: 10.1093/jamiaopen/ooaf045 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gold R, Bunce A, Cottrell E, et al. Study protocol: a pragmatic, stepped-wedge trial of tailored support for implementing social determinants of health documentation/action in community health centers, with realist evaluation. Implement Sci. 2019; 14(1): 9. doi: 10.1186/s13012-019-0855-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Gold R, Bunce A, Cowburn S, et al. Adoption of social determinants of health EHR tools by community health centers. Ann Fam Med. 2018; 16(5): 399-407. doi: 10.1370/afm.2275 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Gold R, Cottrell E, Bunce A, et al. Developing electronic health record (EHR) strategies related to health center patients’ social determinants of health. J Am Board Fam Med. 2017; 30(4): 428-447. doi: 10.3122/jabfm.2017.04.170046 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.LaForge K, Gold R, Cottrell E, et al. How 6 organizations developed tools and processes for social determinants of health screening in primary care: an overview. J Ambul Care Manage. 2018; 41(1): 2-14. doi: 10.1097/JAC.0000000000000221 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Cottrell EK, Dambrun K, Cowburn S, et al. Variation in electronic health record documentation of social determinants of health across a national network of community health centers. Am J Prev Med. 2019; 57(6)(Suppl 1): S65-S73. doi: 10.1016/j.amepre.2019.07.014 [DOI] [PubMed] [Google Scholar]
- 21.Cassel CM, Sarndal CE, Wretman JH.. Some results on generalized difference estimation and generalized regression estimation for finite populations. Biometrika. 1976; 63(3): 615-620. doi: 10.2307/2335742 [DOI] [Google Scholar]
- 22.Gao Y, Liu Y, Matsouaka R.. When does adjusting covariate under randomization help? A comparative study on current practices. BMC Med Res Methodol. 2024; 24(1): 250. doi: 10.1186/s12874-024-02375-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Li X, Ding P.. Rerandomization and Regression Adjustment. J R Stat Soc Series B Stat Methodol. 2020; 82(1): 241-268. doi: 10.1111/rssb.12353 [DOI] [Google Scholar]
- 24.Lin W. Agnostic notes on regression adjustments to experimental data: reexamining Freedman’s critique. Ann Appl Stat. 2013; 7(1). doi: 10.1214/12-AOAS583 [DOI] [Google Scholar]
- 25.Ye T, Shao J, Yi Y, Zhao Q.. Toward better practice of covariate adjustment in analyzing randomized clinical trials. J Am Stat Assoc. 2023; 118(544): 2370-2382. doi: 10.1080/01621459.2022.2049278 [DOI] [Google Scholar]
- 26.Bunce A, Morrissey S, Gold R, et al. “The start of something that I hope could be greater”: health information technology tools for social care. SSM Qual Res Health. 2025; 7: 100544. doi: 10.1016/j.ssmqr.2025.100544 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Cotter TG, Dinneen SF, Healy DA, et al. Glycaemic control is harder to achieve than blood pressure or lipid control in Irish adults with type 1 diabetes. Diabetes Res Clin Pract. 2014; 106(3): e56-e59. doi: 10.1016/j.diabres.2014.09.036 [DOI] [PubMed] [Google Scholar]
- 28.Asgari E, Kaur J, Nuredini G, et al. Impact of electronic health record use on cognitive load and burnout among clinicians: narrative review. JMIR Med Inform. 2024; 12(1): e55499. doi: 10.2196/55499 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Doshani M, Weng M, Moore KL, Romero JR, Nelson NP.. Recommendations of the Advisory Committee on Immunization Practices for use of hepatitis A vaccine for persons experiencing homelessness. MMWR Morb Mortal Wkly Rep. 2019; 68(6): 153-156. doi: 10.15585/mmwr.mm6806a6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.De Marchis EH, Alderwick H, Gottlieb LM.. Do patients want help addressing social risks? J Am Board Fam Med. 2020; 33(2): 170-175. doi: 10.3122/jabfm.2020.02.190309 [DOI] [PubMed] [Google Scholar]