Enhanced Dynamic Clinical Decision Support System Pragmatic Trial (E-DYNAMIC) (E-DYNAMIC)

E-DYNAMIC - Enhanced Dynamic Clinical Decision Support to Identify Patients With Chronic Kidney Disease and Improve Cardiovascular Health

Chronic kidney disease (CKD) is a highly prevalent, poorly recognized and undertreated and increases risk of atherosclerotic cardiovascular disease (ASCVD) and mortality. ASCVD risk interventions such as statin medications are not effective if initiated when kidney disease is advanced. Thus, early recognition of CKD is important for effective ASCVD risk management. Patient centered medical homes (PCMH)s (clinics which include nurse educators, dietitians, pharmacists and social workers) were designed to address gaps in care for complex chronic diseases such as CKD by increasing availability of ancillary services for patients. However, PCMH models have not been shown to improve the recognition and treatment of CKD and its associated ASCVD risk. The E DYNAMIC CDS retrieves real-time patient data from the electronic health record (EHR) every 24 hours to help primary care providers (PCP) identify patients with CKD and assess ASCVD risk and provide appropriate treatment. E-DYNAMIC also delegates CKD care with utilization of an opt-out approach for nurse education and dietitian referral. The overall objective of this pragmatic trial is to examine whether the E-DYNAMIC CDS increases PCP recognition of CKD and use of ASCVD risk management interventions when implemented within a PCMH. This pragmatic trial will be conducted within the Hines VA Hospital and community-based outpatient clinics designed as PCMH called teamlets. Teamlets include several PCPs, a nurse educator, a dietitian, a pharmacist, and a social worker. We will randomize 51 teamlets to the E-DYNAMIC CDS or to standard care. This pragmatic trial will address the following aims: 1) Determine the difference in PCP diagnosis of CKD stage 3-5 non-dialysis dependent CKD by allocation to the E-DYNAMIC CDS; 2) Determine the difference in PCPs ASCVD risk management of patients with stage 3-5 non-dialysis dependent CKD by teamlet allocation to the E-DYNAMIC CDS; 3) Determine the difference in patient use of ASCVD risk interventions and patient activation measures by their teamlet allocation to the E-DYNAMIC CDS. The primary outcomes of the pragmatic trial will be ascertained from the EHR. The E-DYNAMIC CDS tool may be transferred into other health systems that utilize an EHR and improve the diagnosis and management of CKD.

The E-DYNAMIC trial a pragmatic randomized two-arm parallel trial that will randomize 51 teamlets at the Hines Veterans Affairs (VA) hospital outpatient and community based outpatient clinics to either the E-DYNAMIC CDS vs. standard care. The E-DYNAMIC CDS will be activated for PCPs who practice in teamlets allocated to the E-DYNAMIC CDS group and this CDS will be kept active for 18 months to maximize the number of patients with potential CKD who complete a clinic visit with teamlets enrolled in the trial.The index date represents the first visit with the PCP after the date of switching on E-DYNAMIC CDS in intervention and standard care groups. E-DYNAMIC will be active for 18 months to maximize the number of index visits of CKD patients with their PCP during the trial; patients with early stage 3 CKD may not visit their PCP annually. Randomization Scheme: The unit of randomization will be at the teamlet level. PCPs within a teamlet provide coverage for each other's patients so randomization of teamlets will help prevent contamination. We will match the teamlets in pairs based on their potential patients volumes, # of PCPs, and location (hospital based clinic vs. community based outpatient clinic). A computer generated randomization scheme will then be used by the biostatistician to randomize the pairs to either intervention or control. We will analyze the data for all three aims at the patient level, clustered by teamlets; therefore, our analyses will account for intra-cluster correlation among patients within the randomized cluster (PCPs practicing in teamlets). After randomization, we will turn on the E-DYNAMIC CDS for the PCPs working within teamlets allocated to the E-DYNAMIC CDS. The E-DYNAMIC CDS will be seen by the PCPs at the point-of-care for their highly likely chronic kidney disease (CKD) patients identified by up-to-date laboratory data. Teamlets assigned to the standard care group will have no change to their clinical practice. The trial is not blinded and PCPs will be aware if they receive the E-DYNAMIC clinical decision support (CDS). Written individual consent will not be obtained from the providers or patients. All PCPs working in eligible teamlets which could be randomized to the E-DYNAMIC CDS will be contacted several months before trial initiation to inform them of the study and provide them with opportunity to opt-out of the study. All providers who do not opt-out will be eligible to be randomized to the E-DYNAMIC CDS or standard care groups. Study cohort: Our PCP cohort includes all PCPs practicing in PACT teamlets. Our analyses will include patients ≥ 50 years old with two eGFR < 60 mL/min/1.73 m2 for ≥ 3 months apart with no intermittent eGFR> 60 mL/min/1.73 m2 in the electronic health record because these patients have confirmed CKD. Patient level data from administrative records will be queried to obtain information on PCP prescribing practices and patient use of medications and services for 12 months before trial initiation and for 12 months after trial initiation. We will ascertain patient demographics; co-morbidities; mention of CKD in the problem list or an international classification of disease (ICD 9/10) code for CKD, prescription and proportion of days covered (PDC) for statin medications and angiotensin converting enzyme inhibitors, angiotensin receptor blockers (ACE/ARB); and prescription and use of nurse education services, medical nutrition therapy (MNT) from a dietitian, PharmD consult for medication management, nurse visit for blood pressure check and smoking cessation by social worker for smokers. Data sources: Patient level data will be obtained from the following administrative national Veterans Health Administration (VHA) datasets including the corporate data warehouse (CDW), Veterans Affairs Vital Status File, and Managerial Cost Accounting National Data Extracts where information on medication prescriptions are stored. Our analyses will exclude patients receiving dialysis or with a kidney transplant. In order to identify patients receiving dialysis or with a previous kidney transplant, we will link data files with the United States Renal Data System (USRDS) database to further identify these patients. Patient demographics, patient co-morbidities, CKD status based on laboratory data and CKD diagnoses (in the problem list or ICD 9/10 code) and statin and anti-hypertensive medication prescriptions will all be obtained from the CDW, VA Vital Status File and Managerial Cost Accounting National Data Extracts. Reason(s) for not prescribing statins will be collected with E-DYNAMIC and will be saved in the CDW as a health factor for future analysis. Because a small percentage of patients receive their medication from non-VA pharmacies, we will also use data from the Center for Medicare & Medicaid Services (CMS) Medicare Part D 'Slim' file obtained from the VA Information Resource Center to capture statins and anti-hypertensive medications including medications dispensed from non- VA pharmacies. The VA Status Files is updated quarterly. Patient surveys: Using data from the CDW, we will identify eligible patients receiving care at the Hines VA hospital and community based outpatient clinics who have upcoming PCP (in the trial teamlets) appointment in the next 0-6 months. We will stratify the patients by teamlet and use random number generator to select equal number of patients from each teamlet for a total number of 750 eligible patients to participate in the survey (pre-test). These same patients will be followed up for 6 months from the Index visit date and contacted again to re-take the survey (post-test). Our sampling strategy will account for patients who will not show up for their appointments (20%), non-response (20%) and loss to follow-up during post-test (30%). We anticipate that 400 Veterans will complete the pre and post surveys. We will mail a short, informational letter explaining the purpose of the survey, their role and the voluntary nature of participation. We will inform the patient that our study coordinator will contact them in the next couple of weeks to conduct the survey. Patients will be provided a phone number to call to opt-out of the survey. We will use a verbal consent script to obtain verbal consent before proceeding with patient surveys. The survey instruments will consist of the standardized and validated Patient Activation Measure (PAM). Survey will also include questions about statin and/or ACE/ARB use and other hypertension use, and why the patient thinks the doctor prescribed them and if they have received CKD education from a nurse or attended a group CKD education class. Provider perception surveys: We will conduct provider (PCP and nurse) perception surveys to evaluate E-DYNAMIC only in the intervention arm after one year of trial enrollment. The survey will be available via-email, on paper in the clinic or by phone whichever is convenient to the provider. The survey instrument will include short battery of statements asking providers to indicate on a 5 point scale how strongly they agree or disagree with statements regarding experiences about the E-DYNAMIC CDS (PCPs) and the trial (PCPs and nurses), followed by limited number of open ended questions asking for their feedback and recommendations for improvement. We will use a written consent script at the beginning of the provider surveys and assume consent if the provider proceeded to take the on-line survey. Statistical analysis and hypothesis testing primary outcomes:The data analysis for the outcomes will be conducted at the patient level clustered by teamlets; therefore, our analyses will account for intra-cluster correlation among patients within the randomized cluster (PCPs practicing in teamlets). All the primary outcomes in this study are binary (1-Yes/ 0-No). We will assess the outcome variables for each patient at two time points: during 12 months prior to the index date and postintervention 12 months after the index date. The preintervention data will be used as a covariate to adjust for differences in baseline values. For all three aims, we propose to use the two-level mixed-effect logistic regression model to analyze the data with patients at level 1 nested within clustered teamlets (PCPs) as level 2. Although some teamlets may have more than 1 PCP, the PCPs will not be treated as a different level since PCPs within the same teamlets share the same patients and the same clinical nurses. In addition, the associations between teamlets within hospitals are assumed to be very minimal and ignorable given that the E-DYNAMIC system is turned on specifically and limited to PCPs in the randomized intervention group. We will formulate the probability of a positive outcome (1- Yes) for a given patient in a given clustered teamlet using the multilevel mixed effect model. The analysis will be adjusted for patient level (e.g. age, gender, race, and baseline CKD stage or presence of diabetes, hypertension, ASCVD) and teamlet level (e.g. proportion of baseline CKD recognition, number of providers in teamlet, average PCP age in teamlet, median years in practice of teamlet PCPs) covariates to eliminate potential impact on outcomes. Patient surveys: In the initial analyses, we will analyze the raw scores (0-100) from the PAM survey and answers to the medication awareness and education participation. We will use the difference in differences approach to compare change in patient activation level from pre to post between intervention and control groups. A mixed-effect regression model will be applied to account for the intra-cluster correlations. Our analyses will control for patient demographic and medical characteristics. Provider perception surveys: We will use descriptive statistics to describe results from the provider surveys separately for nurses and PCPs and report summary descriptive statistics. Established qualitative analytic techniques will be used to examine results from the open-ended questions, albeit staff answers may be brief, it will involve identifying key themes and concepts emergent from the data to generate meaningful categorization. Sample size justification Primary outcomes: An estimated 7,825 patients with CKD in 51 teamlets receive care from the Hines VA based clinics and its associated 6 CBOC clinics. We have conducted power calculations under various scenarios for changing the effect size (proportion difference in the outcome between intervention and standard care groups) and the outcome level in the standard care group. To test the primary outcomes in 3 specific aims, the significance level of alpha was decreased to 0.0167 accounting for the multiple hypothesis testing in each aim. As such, a total sample size of 50 clustered teamlets (25 in each group) and 4000 patients in each clustered teamlet is anticipated to have 82% power to detect at least a 10% difference in the outcome between the two groups, with an intra-cluster correlation of 0.05 and 20% of patients with the outcome in the control group. Our study will also be powered at 80-90% to detect a 20% difference in the outcome between the intervention and control groups with the half available number of clusters (20 clusters). Patient surveys: For the patient surveys, at a significance level of 0.05, and accounting for an intra-cluster (teamlet) correlation coefficient (ICC) of 0.01, a sample size of 400 patients [10 clusters in each group (intervention and control) and 20 patients in each cluster] is estimated to achieve enough power (at least 90%) to detect a difference of 5 points (with SD =10 ) change in the pre- and post-intervention score of PAM-13 survey between the intervention and standard care control groups.

E-DYNAMIC clinical decision support (CDS) retrieves near-real time patient data from the electronic health record to help primary care providers identify patients with stage 3-5 CKD, present ASCVD risk, statin use and clinical recommendations for ASCVD risk reduction at point of care. E-DYNAMIC directs referrals to nurses for CKD education and to dietitians for MNT. Providers are also nudged to prescribe statin medications and address hypertension management.

Proportion of a primary care providers' patients with chronic kidney disease (CKD) with clinically recognized chronic kidney disease (CKD).

We define CKD as two estimated glomerular filtration rate (eGFR) values < 60 spaced ≥90 days apart with no intervening eGFR values ≥ 60 ml/min/1.73 m2 based on outpatient laboratory values within 18 months of the index primary care visit. Using administrative data, we will determine the proportion of a primary care providers' patients with CKD who have clinically recognized CKD defined as presence of a ICD9/10 billing code for CKD in the administrative records or mention of CKD in the problem list. The outcome variable is binary and categorized as recognized vs. not recognized CKD.

Proportion of a primary care providers' patients with CKD prescribed at least two of the atherosclerotic cardiovascular disease- (ASCVD) CKD care metrics.

Using administrative data, we will determine the proportion of a primary care providers' patients with CKD who are prescribed at least two of the ASCVD-CKD care metrics over a 12 month period: 1) Urine albumin level, including up to 1 month before the index primary care visit, 2) Documentation of nurse education. 3) Medical nutrition therapy with a registered dietitian. 4) Statin prescription (new or renewal of existing prescription), 5) Angiotensin Converting Enzyme Inhibitor or Angiotensin Receptor Blocker (ACE/ARB) prescription (new or renewal) for patient with clinic blood pressure ≥ 130/80mm Hg or ICD9/10 code for hypertension in the 6 months before the index primary care visit, 6) Referral to nephrology for CKD stage 4-5 (eGFR < 30 ml/min/1.73 m2). The outcome variable is binary and defined as being prescribed at least two ASCVD-CKD care metrics versus < 2 ASCVD-CKD care metrics.

Using administrative data, we will determine the proportion of patients with CKD who used ASCVD-CKD care interventions. ASCVD-CKD care interventions include: 1) Documentation of nurse education. 2) Documentation of medical nutrition therapy with a registered dietitian. 3) Statin prescription (new or renewal of existing prescription) with proportion of days covered ≥ 80%, 4) Angiotensin Converting Enzyme Inhibitor or Angiotensin Receptor Blocker (ACE/ARB) prescription (new or renewal) with proportion of days covered ≥ 80% (for patient with clinic blood pressure ≥ 130/80mm Hg or ICD9/10 code for hypertension in the 6 months before the index primary care visit), 5) Referral to nephrology for CKD stage 4-5 (eGFR < 30 ml/min/1.73 m2). The outcome variable is binary and defined as being prescribed at least two ASCVD-CKD care metrics versus < 2 ASCVD-CKD care metrics.

Patient activation will be measured by the Patient Activation Measure, a validated survey which consists of 13 statements to which participants indicate their level of agreement on a 4-point Likert scale. Higher scores indicate higher levels of activation to adopt and maintain health behaviors and strategies for self-managing for their illness. The raw score is transformed to a total score ranging from 0 to 100. Data will be collected from telephone surveys of patients.

Percent of patients who used1) PharmD consult for hypertension. 2) Nurse visit for blood pressure check. 3) Smoking cessation by social worker for smokers. These outcomes are binary. Data will be assessed from administrative records.

Inclusion Criteria: All patients with chronic kidney disease (50 years and older, non-dialysis dependent and who had not received kidney transplant) seen by the primary care providers during the study period will be included in this study. Practicing primary provider at any of the Edward Hines Jr. hospital outpatient clinics or community based outpatient clinics Willing to be involved in the E-DYNAMIC clinical trial Exclusion Criteria: - Provider not willing to be involved in the E-DYNAMIC clinical trial

A web-site that contains details about the clinical trial including a short video presentation, power point slides, and contact information will be built. All findings from the trial along with the E-DYNAMIC clinical decision support tool will be provided on a web-site so that the E-DYNAMIC clinical decision support can be shared with the public. The concept of E-DYNAMIC is readily transferable to other health systems that utilize an electronic medical record. Information on individual participant data may be shared with investigators who obtain approval from the Hines VA Institutional Review Board.

The website will be built once funding has been secured for the trial. The study protocol and consent and the E-DYNAMIC clinical decision support tool will be available on the web-site once the web-site is built. Funding is anticipated by October of 2019. Results of the clinical trial will be posted after publication of the main results. This is anticipated in October of 2022.

Saran R, Robinson B, Abbott KC, Agodoa LY, Albertus P, Ayanian J, Balkrishnan R, Bragg-Gresham J, Cao J, Chen JL, Cope E, Dharmarajan S, Dietrich X, Eckard A, Eggers PW, Gaber C, Gillen D, Gipson D, Gu H, Hailpern SM, Hall YN, Han Y, He K, Hebert H, Helmuth M, Herman W, Heung M, Hutton D, Jacobsen SJ, Ji N, Jin Y, Kalantar-Zadeh K, Kapke A, Katz R, Kovesdy CP, Kurtz V, Lavalee D, Li Y, Lu Y, McCullough K, Molnar MZ, Montez-Rath M, Morgenstern H, Mu Q, Mukhopadhyay P, Nallamothu B, Nguyen DV, Norris KC, O'Hare AM, Obi Y, Pearson J, Pisoni R, Plattner B, Port FK, Potukuchi P, Rao P, Ratkowiak K, Ravel V, Ray D, Rhee CM, Schaubel DE, Selewski DT, Shaw S, Shi J, Shieu M, Sim JJ, Song P, Soohoo M, Steffick D, Streja E, Tamura MK, Tentori F, Tilea A, Tong L, Turf M, Wang D, Wang M, Woodside K, Wyncott A, Xin X, Zang W, Zepel L, Zhang S, Zho H, Hirth RA, Shahinian V. US Renal Data System 2016 Annual Data Report: Epidemiology of Kidney Disease in the United States. Am J Kidney Dis. 2017 Mar;69(3 Suppl 1):A7-A8. doi: 10.1053/j.ajkd.2016.12.004. No abstract available. Erratum In: Am J Kidney Dis. 2017 May;69(5):712.

Matsushita K, Coresh J, Sang Y, Chalmers J, Fox C, Guallar E, Jafar T, Jassal SK, Landman GW, Muntner P, Roderick P, Sairenchi T, Schottker B, Shankar A, Shlipak M, Tonelli M, Townend J, van Zuilen A, Yamagishi K, Yamashita K, Gansevoort R, Sarnak M, Warnock DG, Woodward M, Arnlov J; CKD Prognosis Consortium. Estimated glomerular filtration rate and albuminuria for prediction of cardiovascular outcomes: a collaborative meta-analysis of individual participant data. Lancet Diabetes Endocrinol. 2015 Jul;3(7):514-25. doi: 10.1016/S2213-8587(15)00040-6. Epub 2015 May 28.

Keith DS, Nichols GA, Gullion CM, Brown JB, Smith DH. Longitudinal follow-up and outcomes among a population with chronic kidney disease in a large managed care organization. Arch Intern Med. 2004 Mar 22;164(6):659-63. doi: 10.1001/archinte.164.6.659.

Plantinga LC, Boulware LE, Coresh J, Stevens LA, Miller ER 3rd, Saran R, Messer KL, Levey AS, Powe NR. Patient awareness of chronic kidney disease: trends and predictors. Arch Intern Med. 2008 Nov 10;168(20):2268-75. doi: 10.1001/archinte.168.20.2268.

Palmer SC, Navaneethan SD, Craig JC, Johnson DW, Perkovic V, Hegbrant J, Strippoli GF. HMG CoA reductase inhibitors (statins) for people with chronic kidney disease not requiring dialysis. Cochrane Database Syst Rev. 2014 May 31;(5):CD007784. doi: 10.1002/14651858.CD007784.pub2.

Prabhakaran D, Jha D, Prieto-Merino D, Roy A, Singh K, Ajay VS, Jindal D, Gupta P, Kondal D, Goenka S, Jacob P, Singh R, Kumar BGP, Perel P, Tandon N, Patel V; Members of the Research Steering Committee,Investigators,Members of the Data Safety and Monitoring Board. Effectiveness of an mHealth-Based Electronic Decision Support System for Integrated Management of Chronic Conditions in Primary Care: The mWellcare Cluster-Randomized Controlled Trial. Circulation. 2019 Jan 15;139(3):380-391. doi: 10.1161/CIRCULATIONAHA.118.038192. Epub 2018 Nov 10.

Sinaiko AD, Landrum MB, Meyers DJ, Alidina S, Maeng DD, Friedberg MW, Kern LM, Edwards AM, Flieger SP, Houck PR, Peele P, Reid RJ, McGraves-Lloyd K, Finison K, Rosenthal MB. Synthesis Of Research On Patient-Centered Medical Homes Brings Systematic Differences Into Relief. Health Aff (Millwood). 2017 Mar 1;36(3):500-508. doi: 10.1377/hlthaff.2016.1235.

Hibbard JH, Stockard J, Mahoney ER, Tusler M. Development of the Patient Activation Measure (PAM): conceptualizing and measuring activation in patients and consumers. Health Serv Res. 2004 Aug;39(4 Pt 1):1005-26. doi: 10.1111/j.1475-6773.2004.00269.x.