Utility of Electronic Health Records in Randomized Controlled Trials
Beyond interventions, EHRS may be used for recruitment and outcome assessment
Michelle Dotzert is the creative services manager for our partner brand, Lab Manager. She holds a PhD in kinesiology (specializing in exercise biochemistry) from the University of Western Ontario. Her research examined the effects of exercise training on skeletal muscle lipid metabolism and insulin resistance in a rodent model of Type 1 Diabetes. She has experience with a variety of molecular and biochemistry techniques, as well as HPLC-MS.
ViewFull ProfileLearn about ourEditorial Policies.
The randomized controlled trial (RCT) is the standard study design used to evaluate the safety and efficacy of novel treatments and interventions. In addition to demonstrating the potential superiority of a novel treatment over an existing treatment or placebo, RCTs are also associated with a low risk of bias. However, RCTs are timeconsuming and costly to carry out, and have limited generalizability as they are conducted in standardized settings. As such, there is growing interest in the use of routinely collected data, in the form of disease registries, clinical databases, and electronic health records (EHRs), for research purposes.1
EHRs are electronic platforms that contain real-time, patient-centered records including medical histories, diagnoses, medications, allergies, laboratory results, and more that can be created, managed, and shared by multiple health care providers and organizations. EHRs can facilitate pragmatic RCTs designed to evaluate the effectiveness of interventions in real-life settings rather than specialized environments. These pragmatic designs produce more generalizable data that may be implemented in clinical practice.
EHRs are commonly used as part of the intervention, but more recently they have also been successfully leveraged to support recruitment and outcome assessment, and have the potential to support more efficient, cost-effective RCTs.
Intervention delivery
EHRs are often used in RCTs for intervention delivery. System-level interventions such as computerized physician order entry systems (CPOE) and clinical decision support systems (CDSS) have also been incorporated into EHRs to support ordering practices and decision making.
There are numerous trials demonstrating the successful implementation of EHRs as part of the study intervention. One example is the Isotonic Solutions and Major Adverse Renal Events Trial (SMART),2 which coordinated the intervention—balanced crystalloids vs saline in patients admitted to the intensive care unit for a variety of critical illnesses—through EHRs. An electronic advisor within the order-entry system informed providers about the trial and guided them to order the assigned crystalloid when appropriate.
Similarly, a clinical decision support tool was developed to improve medication prescribing for patients with kidney disease.3 This tool used the alert functionality within the EHR, and 20 medications were included with unique alerts based on primary literature review, tertiary references, institutional renal dosage adjustment guidelines, and a third-party EHR drug database.
In addition to guiding prescribing, EHR-based tools have been developed to guide screening. One such tool was developed to help primary care providers identify, evaluate, and treat patients who are overweight or obese.4 The EHR tool included reminders for providers to measure height and weight, alerts asking whether overweight or obesity should be added to a problem list, and reminders with tailored management recommendations based on the patient’s other risk factors.
Facilitating recruitment
Recruitment represents a major hurdle for clinical trials. EHRs can help with recruitment and enhance sample size by enabling researchers to recruit patients during routine clinical visits. This process facilitates consecutive enrollment (a sampling technique by which each subject meeting the inclusion criteria is enrolled until the desired sample size is achieved), and dramatically accelerates the recruitment process, as the EHR may be used to screen and identify patients based on inclusion criteria, send a notification to the clinician, obtain informed consent electronically, and randomize the participant into a study group, all during a single visit.5
Several RCTs have successfully recruited participants with EHR-based methods. For example, a retrospective study was conducted using pharmacy dispensing software to recruit patients with uncontrolled asthma.6 An application was installed at participating pharmacies to query pharmacy EHRs to identify patients who had received at least three short-acting beta-2-agonist canisters over the previous six months, who then received study materials from a pharmacist. Another RCT examined whether an EHR-linked and mailed colorectal cancer screening program improved colorectal cancer screening adherence compared to usual care.7 Compared to the usual care group, the EHR-linked screening program led to twice as many individuals being screened over a two year period.
Overcoming logistical challenges
Implementing outcome assessment via EHR offers significant logistical advantages for RCTs, namely, it eliminates the time and resources required to measure outcomes. In the asthma and colorectal cancer screening studies, EHRs were the sole method used for outcome assessment. In the asthma study, the EHR was used to obtain medication dispensing data for the study period, and the colorectal cancer screening trial based primary outcomes on evidence from the EHR.
In addition to primary outcome assessment, RCTs must also report major clinical events and serious adverse effects. EHRs can be used to simplify this process, as many of these outcomes, including stroke, myocardial infarction, hospital admissions, and other adverse events are documented within the EHR and may be extracted for this purpose.
Improving efficiency and cost savings
Two major challenges associated with traditional RCTs are financial costs and the length of the process. EHRsupported trials have been proposed as a more cost-effective alternative to traditional RCTs, with respect to total and per-patient costs. In a systematic review of costs and resource use of RCTs, the overall cost of an RCT was shown to range between $0.2 million and $611.5 million (USD), with per-patient costs between $41 and $6,990.8 A systematic descriptive study of the use of EHRs in RCTs reported total costs for four trials ranging from $67,750 to approximately $5 million with per-patient costs ranging between $44 and $2,000.9 The authors attributed variations in cost to manual versus automated data extraction from the EHR, with a potential for further per-patient cost reductions with automatic data extraction. Using EHRs for outcome assessment may also enable real-time result collection, which has the potential to accelerate time to first results.6
Some limitations remain
In addition to facilitating intervention delivery, leveraging EHRs for RCTs has the potential to enhance recruitment and outcome assessment, while reducing costs and removing logistical barriers. There are, however, several limitations to their widespread use at this time.
While EHRs have been shown to reduce overall health care costs, including savings on drug expenditures, radiology tests, transcription and dictation costs, among others, the initial costs associated with establishing the necessary infrastructure, as well as maintenance costs, pose barriers to their implementation.10 The initial costs of adoption have been estimated to range between $15,000 and $70,000 per provider.11 Conducting RCTs within EHRs is also limited due to a lack of interoperability. There are a multitude of EHR products in use across North America, and a standard interoperability format has not yet been developed to facilitate the exchange and use of data. Improved operability will be essential to help collaboration across multiple sites.
The nature of the data contained within an EHR is also often unsuitable for research purposes. EHRs contain clinician notes written as free-text, as well as reports and summaries, and this type of data is not easily repurposed.12 While it is possible to extract the necessary data, manual techniques are time-consuming. Artificial intelligence tools, such as natural language processing (NLP), are being implemented to overcome this challenge. NLP is designed to read and contextualize medical terms and expressions throughout EHRs and can compile diverse data ranging from medical history to laboratory results and imaging.13
Finally, there are privacy and ethics concerns surrounding the use of EHRs for research purposes. Data breaches may have significant consequences for those affected, ranging from stigma and discrimination to difficulties obtaining health insurance.
Currently, national policies including the HIPAA Privacy Rule, HITECH Act of 2009, and the Federal Policy for the Protection of Human Subjects are in place to protect health data, and EHR systems are supported by encryption, access limits, and audit logs, among other strategies to ensure privacy.14 Further measures have been recommended, including data validation by researchers and EHR software developers, ethics training for all those involved in data handling, and the development and enforcement of consequences in the event of data breaches.14
Widespread adoption of the EHR by health care providers has created new opportunities for clinical research. Serving as a mobile health record, and containing data from multiple providers, the EHR can facilitate more efficient, affordable RCTs. In addition to their utility in intervention delivery, EHRs are increasingly used for recruitment and outcome assessment. EHRs also support pragmatic trials and may be incorporated into a hybrid design with both traditional and pragmatic elements. While there are many benefits to leveraging this digital platform, ensuring privacy and ethics will be fundamental to protect patients.
References:
1. Hemkens, Lars G, et al. “Routinely collected data and comparative effectiveness evidence: Promises and limitations.” Canadian Medical Association Journal, 188, (2016): E158–64.
2. Semler, Matthew W., et al. “Balanced crystalloids versus saline in critically ill adults.” New England Journal of Medicine, 378 (2018): 829–39.
3. Awdishu, Linda et al. “The impact of real-time alerting on appropriate prescribing in kidney disease: a cluster randomized controlled trial.” Journal of the American Medical Informatics Association 23,3 (2016): 609-16.
4. Baer, Heather J., et al. “Design of a cluster-randomized trial of electronic health record-based tools to address overweight and obesity in primary care.” Clinical Trials, 12 (2015): 374–83.
5. McCord, Kimberly A., and Lars G. Hemkens. “Using electronic health records for clinical trials: Where do we stand and where can we go?” Canadian Medical Association Journal, 191 (2019): E128–33.
6. Bereznicki, Bonnie J., et al. “Pharmacist-initiated general practitioner referral of patients with suboptimal asthma management.” Pharmacy World & Science, 30 (2008) 869–75.
7. Green, Beverly B., et al. “Automated intervention with stepped increases in support to increase uptake of colorectal cancer screening: A randomized trial.” Annals of Internal Medicine, 158 (2013): 301–11.
8. Speich, Benjamin, et al. “Systematic review on costs and resource use of randomized clinical trials shows a lack of transparent and comprehensive data.” Journal of Clinical Epidemiology, 96 (2018): 1–11.
9. McCord, Kimberly A., and Lars G. Hemkens. “Using electronic health records for clinical trials: Where do we stand and where can we go?” Canadian Medical Association Journal, 191 (2019): E128–33.
10. Choi, Jong Soo, et al. “Cost-benefit analysis of electronic medical record system at a tertiary care hospital.” Healthcare Informatics Research, 19, (2013): 205–14.
11. Reisman, Miriam. “EHRs: The challenge of making electronic data usable and interoperable.” Pharmacology and Therapeutics Journal, 42 (2017): 572–75.
12. Capurro, Daniel, et al. “Availability of structured and unstructured clinical data for comparative effectiveness research and quality improvement: A multi-site assessment.” EGEMs (Generating Evidence & Methods to Improve Patient Outcomes), vol. 2, no. 1, Ubiquity Press, Ltd., July 2014, p. 11.
13. Miller, D. Douglas, and Eric W. Brown. “Artificial intelligence in medical practice: The question to the answer?” American Journal of Medicine, 131 (2018): 129–33.
14. Lee, Lisa M. “Ethics and subsequent use of electronic health record data.” Journal of Biomedical Informatics, 71 (2017): 143–46.
