Mind the clinical-analytic gap: Electronic health records and COVID-19 pandemic response.

Data quality is essential to the success of the most simple and the most complex analysis. In the context of the COVID-19 pandemic, large-scale data sharing across the US and around the world has played an important role in public health responses to the pandemic and has been crucial to understanding and predicting its likely course. In California, hospitals have been required to report a large volume of daily data related to COVID-19. In order to meet this need, electronic health records (EHRs) have played an important role, but the challenges of reporting high-quality data in real-time from EHR data sources have not been explored. We describe some of the challenges of utilizing EHR data for this purpose from the perspective of a large, integrated, mixed-payer health system in northern California, US. We emphasize some of the inadequacies inherent to EHR data using several specific examples, and explore the clinical-analytic gap that forms the basis for some of these inadequacies. We highlight the need for data and analytics to be incorporated into the early stages of clinical crisis planning in order to utilize EHR data to full advantage. We further propose that lessons learned from the COVID-19 pandemic can result in the formation of collaborative teams joining clinical operations, informatics, data analytics, and research, ultimately resulting in improved data quality to support effective crisis response.

Trends in Detection of Adenoma and Sessile Serrated Lesions Over a Decade in a Community-Based Healthcare System.

INTRODUCTION: Adenoma detection rate (ADR) is an accepted benchmark for screening colonoscopy. Factors driving ADR and its relationship with sessile serrated lesions detection rate (SSLDR) over time remain unclear. We aim to explore patient, physician, and procedural influences on ADR and SSLDR trends. METHODS: Using a large healthcare system in northern California from January 2010 to December 2020, a total of 146,818 screening colonoscopies performed by 33 endoscopists were included. ADR and SSLDR were calculated over time using natural language processing. Logistic regression was used to calculate the odd ratios of patient demographics, physician attributes, and procedural details over time. RESULTS: Between 2010 and 2020, ADR rose from 19.4% to 44.4%, whereas SSLDR increased from 1.6% to 11.6%. ADR increased by 2.7% per year (95% confidence interval 1.9%-3.4%), and SSLDR increased by 1.0% per year (95% confidence interval 0.8%-1.2%). Higher ADR was associated with older age, male sex, higher body mass index, current smoker, higher comorbidities, and high-risk colonoscopy. By contrast, SSLDR was associated with younger age, female sex, white race, and fewer comorbidities. Patient and procedure characteristics did not significantly change over time ( P -interaction >0.05). Longer years in practice and male physician were associated with lower ADR and SSLDR in 2010, but significantly attenuated over time ( P -interaction <0.05). DISCUSSION: Both ADR and SSLDR have increased over time. Patient and procedure factors did not significantly change over time. Male endoscopist and longer years in practice had lower initial ADR and SSLDR, but significantly lessened over time.

Drivers of Variation in Opioid Prescribing after Common Surgical Procedures in a Large Multihospital Healthcare System.

BACKGROUND: Misuse of prescription opioids is a well-established contributor to the United States opioid epidemic. The primary objective of this study was to identify which level of care delivery (i.e. patient, prescriber, or hospital) produced the most unwarranted variation in opioid prescribing after common surgical procedures. STUDY DESIGN: Electronic health record (EHR) data from a large multihospital healthcare system was used in conjunction with random-effect models to examine variation in opioid prescribing practices following similar inpatient and outpatient surgical procedures between October 2019 and September 2021. Unwarranted variation was conceptualized as variation resulting from prescriber behavior unsupported by evidence. Covariates identified as drivers of warranted variation included characteristics known to influence pain levels or patient safety. All other model variables, including prescriber specialty and patient race, ethnicity, and insurance status were characterized as potential drivers of unwarranted variation. RESULTS: Among 25,188 procedures with an opioid prescription at hospital discharge, 53.5% exceeded guideline recommendations, corresponding to 13,228 patients receiving the equivalent of >140,000 excess 5mg oxycodone tablets following surgical procedures. Prescribing variation was primarily driven by prescriber-level factors, with approximately half of the total variation in morphine milligram equivalents (MMEs) prescribed observed at the prescriber level and not explained by any measured variables. Unwarranted covariates associated with higher prescribed opioid quantity included non-Hispanic black race, Medicare insurance, smoking history, later hospital discharge times, and prescription by a surgeon rather than a hospitalist or primary care provider. CONCLUSION: Given the large proportion of unexplained variation observed at the provider level, targeting prescribers through education and training may be an effective strategy for reducing postoperative opioid prescribing.

Homelessness Among Acute Care Patients Within a Large Health Care System in Northern California.

The impacts of homelessness on health and health care access are detrimental. Intervention and efforts to improve outcomes and increase availability of affordable housing have mainly originated from the public health sector and government. The role that large community-based health systems may play has yet to be established. This study characterizes patients self-identified as homeless in acute care facilities in a large integrated health care system in Northern California to inform the development of collaborative interventions addressing unmet needs of this vulnerable population. The authors compared sociodemographic characteristics, clinical conditions, and health care utilization of individuals who did and did not self-identify as homeless and characterized their geographical distribution in relation to Sutter hospitals and homeless resources. Between July 1, 2019 and June 30, 2020, 5% (N = 20,259) of the acute care settings patients had evidence of homelessness, among which 51.1% age <45 years, 66.4% males, and 24% non-Hispanic Black. Patients experiencing homelessness had higher emergency department utilization and lower utilization of outpatient and urgent care services. Mental health conditions were more common among patients experiencing homelessness. More than half of the hospitals had >5% of patients who identified as homeless. Some hospitals with higher proportions of patients experiencing homelessness are not located near many shelter resources. By understanding patients who self-identify as homeless, it is possible to assess the role of the health system in addressing their unmet needs. Accurate identification is the first step for the health systems to develop and deliver better solutions through collaborations with nonprofit organizations, community partners, and government agencies.

Risk of Gastric Adenocarcinoma in a Multiethnic Population Undergoing Routine Care: An Electronic Health Records Cohort Study.

BACKGROUND: Gastric adenocarcinoma (GAC) is often diagnosed at advanced stages and portends a poor prognosis. We hypothesized that electronic health records (EHR) could be leveraged to identify individuals at highest risk for GAC from the population seeking routine care. METHODS: This was a retrospective cohort study, with endpoint of GAC incidence as ascertained through linkage to an institutional tumor registry. We utilized 2010 to 2020 data from the Palo Alto Medical Foundation, a large multispecialty practice serving Northern California. The analytic cohort comprised individuals ages 40-75 receiving regular ambulatory care. Variables collected included demographic, medical, pharmaceutical, social, and familial data. Electronic phenotyping was based on rule-based methods. RESULTS: The cohort comprised 316,044 individuals and approximately 2 million person-years (p-y) of observation. 157 incident GACs occurred (incidence 7.9 per 100,000 p-y), of which 102 were non-cardia GACs (incidence 5.1 per 100,000 p-y). In multivariable analysis, male sex [HR: 2.2, 95% confidence interval (CI): 1.6-3.1], older age, Asian race (HR: 2.5, 95% CI: 1.7-3.7), Hispanic ethnicity (HR: 1.9, 95% CI: 1.1-3.3), atrophic gastritis (HR: 4.6, 95% CI: 2.2-9.3), and anemia (HR: 1.9, 95% CI: 1.3-2.6) were associated with GAC risk; use of NSAID was inversely associated (HR: 0.3, 95% CI: 0.2-0.5). Older age, Asian race, Hispanic ethnicity, atrophic gastritis, and anemia were associated with non-cardia GAC. CONCLUSIONS: Routine EHR data can stratify the general population for GAC risk. IMPACT: Such methods may help triage populations for targeted screening efforts, such as upper endoscopy.

Peer Comparison or Guideline-Based Feedback and Postsurgery Opioid Prescriptions: A Randomized Clinical Trial.

Importance: Excess opioid prescribing after surgery can result in prolonged use and diversion. Email feedback based on social norms may reduce the number of pills prescribed. Objective: To assess the effectiveness of 2 social norm-based interventions on reducing guideline-discordant opioid prescribing after surgery. Design, Setting, and Participants: This cluster randomized clinical trial conducted at a large health care delivery system in northern California between October 2021 and October 2022 included general, obstetric/gynecologic, and orthopedic surgeons with patients aged 18 years or older discharged to home with an oral opioid prescription. Interventions: In 19 hospitals, 3 surgical specialties (general, orthopedic, and obstetric/gynecologic) were randomly assigned to a control group or 1 of 2 interventions. The guidelines intervention provided email feedback to surgeons on opioid prescribing relative to institutionally endorsed guidelines; the peer comparison intervention provided email feedback on opioid prescribing relative to that of peer surgeons. Emails were sent to surgeons with at least 2 guideline-discordant prescriptions in the previous month. The control group had no intervention. Main Outcome and Measures: The probability that a discharged patient was prescribed a quantity of opioids above the guideline for the respective procedure during the 12 intervention months. Results: There were 38 235 patients discharged from 640 surgeons during the 12-month intervention period. Control-group surgeons prescribed above guidelines 36.8% of the time during the intervention period compared with 27.5% and 25.4% among surgeons in the peer comparison and guidelines arms, respectively. In adjusted models, the peer comparison intervention reduced guideline-discordant prescribing by 5.8 percentage points (95% CI, -10.5 to -1.1; P = .03) and the guidelines intervention reduced it by 4.7 percentage points (95% CI, -9.4 to -0.1; P = .05). Effects were driven by surgeons who performed more surgeries and had more guideline-discordant prescribing at baseline. There was no significant difference between interventions. Conclusions and Relevance: In this cluster randomized clinical trial, email feedback based on either guidelines or peer comparison reduced opioid prescribing after surgery. Guideline-based feedback was as effective as peer comparison-based feedback. These interventions are simple, low-cost, and scalable, and may reduce downstream opioid misuse. Trial Registration: ClinicalTrials.gov NCT05070338.

Disparities in postoperative opioid prescribing by race and ethnicity: an electronic health records-based observational study from Northern California, 2015-2020.

OBJECTIVES: To examine racial and ethnic disparities in postoperative opioid prescribing. DATA SOURCES: Electronic health records (EHR) data across 24 hospitals from a healthcare delivery system in Northern California from January 1, 2015 to February 2, 2020 (study period). STUDY DESIGN: Cross-sectional, secondary data analyses were conducted to examine differences by race and ethnicity in opioid prescribing, measured as morphine milligram equivalents (MME), among patients who underwent select, but commonly performed, surgical procedures. Linear regression models included adjustment for factors that would likely influence prescribing decisions and race and ethnicity-specific propensity weights. Opioid prescribing, overall and by race and ethnicity, was also compared to postoperative opioid guidelines. DATA EXTRACTION: Data were extracted from the EHR on adult patients undergoing a procedure during the study period, discharged to home with an opioid prescription. PRINCIPAL FINDINGS: Among 61,564 patients, on adjusted regression analysis, non-Hispanic Black (NHB) patients received prescriptions with higher mean MME than non-Hispanic white (NHW) patients (+ 6.4% [95% confidence interval: 4.4%, 8.3%]), whereas Hispanic and non-Hispanic Asian patients received lower mean MME (-4.2% [-5.1%, -3.2%] and - 3.6% [-4.8%, -2.3%], respectively). Nevertheless, 72.8% of all patients received prescriptions above guidelines, ranging from 71.0 to 80.3% by race and ethnicity. Disparities in prescribing were eliminated among Hispanic and NHB patients versus NHW patients when prescriptions were written within guideline recommendations. CONCLUSIONS: Racial and ethnic disparities in opioid prescribing exist in the postoperative setting, yet all groups received prescriptions above guideline recommendations. Policies encouraging guideline-based prescribing may reduce disparities and overall excess prescribing.

Medication non-adherence and therapeutic inertia independently contribute to poor disease control for cardiometabolic diseases.

Poorly controlled cardiometabolic biometric health gap measures [e.g.,uncontrolled blood pressure (BP), HbA1c, and low-density lipoprotein cholesterol (LDL-C)] are mediated by medication adherence and clinician-level therapeutic inertia (TI). The study of comparing relative contribution of these two factors to disease control is lacking. We conducted a retrospective cohort study using 7 years of longitudinal electronic health records (EHR) from primary care cardiometabolic patients who were 35 years or older. Cox-regression modeling was applied to estimate how baseline proportion of days covered (PDC) and TI were associated with cardiometabolic related health gap closure. 92,766 patients were included in the analysis, among which 89.9%, 85.8%, and 73.3% closed a BP, HbA1c, or LDL-C gap, respectively, with median days to gap closure ranging from 223 to 408 days. Patients who did not retrieve a medication were the least likely to achieve biometric control, particularly for LDL-C (HR = 0.58, 95% CI: 0.55-0.60). TI or uncertainty of TI was associated with a high risk of health gap persistence, particularly for LDL-C (HR ranges 0.46-0.48). Both poor medication adherence and TI are independently associated with persistent health gaps, and TI has a much higher impact on disease control compared to medication adherence, implying disease management strategies should prioritize reducing TI.

Do Patients Continue to Use Video Visits? Factors Related to Continued Video Visit Use.

Many studies have assessed the factors associated with overall video visit use during the COVID-19 pandemic, but little is known about who is most likely to continue to use video visits and why. The authors combined a survey with electronic health record data to identify factors affecting the continued use of video visit. In August 2020, a stratified random sample of 20,000 active patients from a large health care system were invited to complete an email survey on health care seeking preferences during the COVID. Weighted logistic regression models were applied, adjusting for sampling frame and response bias, to identify factors associated with video visit experience, and separately for preference of continued use of video visits. Actual video visit utilization was also estimated within 12 months after the survey. Three thousand three hundred fifty-one (17.2%) patients completed the survey. Of these, 1208 (36%) reported having at least 1 video visit in the past, lowest for African American (33%) and highest for Hispanic (41%). Of these, 38% would prefer a video visit in the future. The strongest predictors of future video visit use were comfort using video interactions (odds ratio [OR] = 5.30, 95% confidence interval [95% CI]: 3.57-7.85) and satisfaction with the overall quality (OR = 3.94, 95% CI: 2.66-5.86). Interestingly, despite a significantly higher satisfaction for Hispanic (40%-55%) and African American (40%-50%) compared with Asian (29%-39%), Hispanic (OR = 0.46, 95% CI: 0.12-0.88) and African American (OR = 0.54, 95% CI: 0.16-0.90) were less likely to prefer a future video visit. Disparity exists in the use of video visit. The association between patient satisfaction and continued video visit varies by race/ethnicity, which may change the future long-term video visit use among race/ethnicity groups.

Evaluating the Impact of a Point-of-Care Cardiometabolic Clinical Decision Support Tool on Clinical Efficiency Using Electronic Health Record Audit Log Data: Algorithm Development and Validation.

BACKGROUND: Electronic health record (EHR) systems are becoming increasingly complicated, leading to concerns about rising physician burnout, particularly for primary care physicians (PCPs). Managing the most common cardiometabolic chronic conditions by PCPs during a limited clinical time with a patient is challenging. OBJECTIVE: This study aimed to evaluate a Cardiometabolic Sutter Health Advanced Reengineered Encounter (CM-SHARE), a web-based application to visualize key EHR data, on the EHR use efficiency. METHODS: We developed algorithms to identify key clinic workflow measures (eg, total encounter time, total physician time in the examination room, and physician EHR time in the examination room) using audit data, and we validated and calibrated the measures with time-motion data. We used a pre-post parallel design to identify propensity score-matched CM-SHARE users (cases), nonusers (controls), and nested-matched patients. Cardiometabolic encounters from matched case and control patients were used for the workflow evaluation. Outcome measures were compared between the cases and controls. We applied this approach separately to both the CM-SHARE pilot and spread phases. RESULTS: Time-motion observation was conducted on 101 primary care encounters for 9 PCPs in 3 clinics. There was little difference (<0.8 minutes) between the audit data-derived workflow measures and the time-motion observation. Two key unobservable times from audit data, physician entry into and exiting the examination room, were imputed based on time-motion studies. CM-SHARE was launched with 6 pilot PCPs in April 2016. During the prestudy period (April 1, 2015, to April 1, 2016), 870 control patients with 2845 encounters were matched with 870 case patients and encounters, and 727 case patients with 852 encounters were matched with 727 control patients and 3754 encounters in the poststudy period (June 1, 2016, to June 30, 2017). Total encounter time was slightly shorter (mean -2.7, SD 1.4 minutes, 95% CI -4.7 to -0.9; mean -1.6, SD 1.1 minutes, 95% CI -3.2 to -0.1) for cases than controls for both periods. CM-SHARE saves physicians approximately 2 minutes EHR time in the examination room (mean -2.0, SD 1.3, 95% CI -3.4 to -0.9) compared with prestudy period and poststudy period controls (mean -1.9, SD 0.9, 95% CI -3.8 to -0.5). In the spread phase, 48 CM-SHARE spread PCPs were matched with 84 control PCPs and 1272 cases with 3412 control patients, having 1119 and 4240 encounters, respectively. A significant reduction in total encounter time for the CM-SHARE group was observed for short appointments (≤20 minutes; 5.3-minute reduction on average) only. Total physician EHR time was significantly reduced for both longer and shorter appointments (17%-33% reductions). CONCLUSIONS: Combining EHR audit log files and clinical information, our approach offers an innovative and scalable method and new measures that can be used to evaluate clinical EHR efficiency of digital tools used in clinical settings.

Assessment of Postoperative Opioid Prescriptions Before and After Implementation of a Mandatory Prescription Drug Monitoring Program.

Importance: Legislation mandating consultation with a prescription drug monitoring program (PDMP) was implemented in California on October 2, 2018. This mandate requires PDMP consultation before prescribing a controlled substance and integrates electronic health record (EHR)-based alerts; prescribers are exempt from the mandate if they prescribe no more than a 5-day postoperative opioid supply. Although previous studies have examined the consequences of mandated PDMP consultation, few have specifically analyzed changes in postoperative opioid prescribing after mandate implementation. Objective: To examine whether the implementation of mandatory PDMP consultation with concurrent EHR-based alerts was associated with changes in postoperative opioid quantities prescribed at discharge. Design Setting and Participants: This cross-sectional study performed an interrupted time series analysis of opioid prescribing patterns within a large health care system (Sutter Health) in northern California between January 1, 2015, and February 1, 2020. A total of 93 760 adult patients who received an opioid prescription at discharge after undergoing general, obstetric and gynecologic (obstetric/gynecologic), or orthopedic surgery were included. Exposures: Mandatory PDMP consultation before opioid prescribing, with concurrent integration of an EHR alert. Prescribers are exempt from this mandate if prescribing no more than a 5-day opioid supply postoperatively. Main Outcomes and Measures: The primary outcome was the total quantity of opioid medications (morphine milligram equivalents [MMEs] and number of opioid tablets) prescribed at discharge before and after implementation of the PDMP mandate, with separate analyses by surgical specialty (general, obstetric/gynecologic, and orthopedic) and most common surgical procedure within each specialty (laparoscopic cholecystectomy, cesarean delivery, and knee arthroscopy). The secondary outcome was the proportion of prescriptions with a duration of longer than 5 days. Results: Of 93 760 patients (mean [SD] age, 46.7 [17.6] years; 67.9% female) who received an opioid prescription at discharge, 65 911 received prescriptions before PDMP mandate implementation, and 27 849 received prescriptions after implementation. Most patients received general or obstetric/gynecologic surgery (48.6% and 30.1%, respectively), did not have diabetes (90.3%), and had never smoked (66.0%). Before the PDMP mandate was implemented, a decreasing pattern in opioid prescribing quantities was already occurring. During the quarter of implementation, total MMEs prescribed at discharge further decreased for all 3 surgical specialties (eg, medians for general surgery: ß = -10.00 [95% CI, -19.52 to -0.48]; obstetric/gynecologic surgery: ß = -18.65 [95% CI, -22.00 to -15.30]; and orthopedic surgery: ß = -30.59 [95% CI, -40.19 to -21.00]) after adjusting for the preimplementation prescribing pattern. The total number of tablets prescribed also decreased across specialties (eg, medians for general surgery: ß = -3.02 [95% CI, -3.47 to -2.57]; obstetric/gynecologic surgery: ß = -4.86 [95% CI, -5.38 to -4.34]; and orthopedic surgery: ß = -4.06 [95% CI, -5.07 to -3.04]) compared with the quarters before implementation. These reductions were not consistent across the most common surgical procedures. For cesarean delivery, the median number of tablets prescribed decreased during the quarter of implementation (ß = -10.00; 95% CI, -10.10 to -9.90), but median MMEs did not (ß = 0; 95% CI, -9.97 to 9.97), whereas decreases were observed in both median MMEs and number of tablets prescribed (MMEs: ß = -33.33 [95% CI, -38.48 to -28.19]; tablets: ß = -10.00 [95% CI, -11.17 to -8.82]) for laparoscopic cholecystectomy. For knee arthroscopy, no decreases were found in either median MMEs or number of tablets prescribed (MMEs: ß = 10.00 [95% CI, -22.33 to 42.33; tablets: ß = 0.83; 95% CI, -3.39 to 5.05). The proportion of prescriptions written for longer than 5 days also decreased significantly during the quarter of implementation across all 3 surgical specialties. Conclusions and Relevance: In this cross-sectional study, the implementation of mandatory PDMP consultation with a concurrent EHR-based alert was associated with an immediate decrease in opioid prescribing across the 3 surgical specialties. These findings might be explained by prescribers' attempts to meet the mandate exemption and bypass PDMP consultation rather than the PDMP consultation itself. Although policies coupled with EHR alerts may be associated with changes in postoperative opioid prescribing behavior, they need to be well designed to optimize evidence-based opioid prescribing.

Impact of implementing electronic prior authorization on medication filling in an electronic health record system in a large healthcare system.

OBJECTIVE: Medications frequently require prior authorization from payers before filling is authorized. Obtaining prior authorization can create delays in filling prescriptions and ultimately reduce patient adherence to medication. Electronic prior authorization (ePA), embedded in the electronic health record (EHR), could remove some barriers but has not been rigorously evaluated. We sought to evaluate the impact of implementing an ePA system on prescription filling. MATERIALS AND METHODS: ePA was implemented in 2 phases in September and November 2018 in a large US healthcare system. This staggered implementation enabled the later-implementing sites to be controls. Using EHR data from all prescriptions written and linked information on whether prescriptions were filled at pharmacies, we 1:1 matched ePA prescriptions with non-ePA prescriptions for the same insurance plan, medication, and site, before and after ePA implementation, to evaluate primary adherence, or the proportion of prescriptions filled within 30 days, using generalized estimating equations. We also conducted concurrent analyses across sites during the peri-implementation period (Sept-Oct 2018). RESULTS: Of 74 546 eligible ePA prescriptions, 38 851 were matched with preimplementation controls. In total, 24 930 (64.2%) ePA prescriptions were filled compared with 26 731 (68.8%) control prescriptions (Adjusted Relative Risk [aRR]: 0.92, 95%CI: 0.91-0.93). Concurrent analyses revealed similar findings (64.7% for ePA vs 62.3% control prescriptions, aRR: 1.03, 95%CI: 0.98-1.09). DISCUSSION: Challenges with implementation, such as misfiring and insurance fragmentation, could have undermined its effectiveness, providing implications for other health informatics interventions deployed in outpatient care. CONCLUSION: Despite increasing interest in implementing ePA to improve prescription filling, adoption did not change medication adherence.

Imputation of missing values for electronic health record laboratory data.

Laboratory data from Electronic Health Records (EHR) are often used in prediction models where estimation bias and model performance from missingness can be mitigated using imputation methods. We demonstrate the utility of imputation in two real-world EHR-derived cohorts of ischemic stroke from Geisinger and of heart failure from Sutter Health to: (1) characterize the patterns of missingness in laboratory variables; (2) simulate two missing mechanisms, arbitrary and monotone; (3) compare cross-sectional and multi-level multivariate missing imputation algorithms applied to laboratory data; (4) assess whether incorporation of latent information, derived from comorbidity data, can improve the performance of the algorithms. The latter was based on a case study of hemoglobin A1c under a univariate missing imputation framework. Overall, the pattern of missingness in EHR laboratory variables was not at random and was highly associated with patients' comorbidity data; and the multi-level imputation algorithm showed smaller imputation error than the cross-sectional method.

Persistent Cardiometabolic Health Gaps: Can Therapeutic Care Gaps Be Precisely Identified from Electronic Health Records.

The objective of this study was to determine the strengths and limitations of using structured electronic health records (EHR) to identify and manage cardiometabolic (CM) health gaps. We used medication adherence measures derived from dispense data to attribute related therapeutic care gaps (i.e., no action to close health gaps) to patient- (i.e., failure to retrieve medication or low adherence) or clinician-related (i.e., failure to initiate/titrate medication) behavior. We illustrated how such data can be used to manage health and care gaps for blood pressure (BP), low-density lipoprotein cholesterol (LDL-C), and HbA1c for 240,582 Sutter Health primary care patients. Prevalence of health gaps was 44% for patients with hypertension, 33% with hyperlipidemia, and 57% with diabetes. Failure to retrieve medication was common; this patient-related care gap was highly associated with health gaps (odds ratios (OR): 1.23-1.76). Clinician-related therapeutic care gaps were common (16% for hypertension, and 40% and 27% for hyperlipidemia and diabetes, respectively), and strongly related to health gaps for hyperlipidemia (OR = 5.8; 95% CI: 5.6-6.0) and diabetes (OR = 5.7; 95% CI: 5.4-6.0). Additionally, a substantial minority of care gaps (9% to 21%) were uncertain, meaning we lacked evidence to attribute the gap to either patients or clinicians, hindering efforts to close the gaps.