Hazel Tang A science writer with data background and an interest in the current affair, culture, and arts; a no-med from an (almost) all-med family. Follow on Twitter.
Photo Credit: Anne-Lise Heinrichs
In this digital age, the widespread use of electronic medical records (EMRs) has been transformative to the practice of medicine, and the development and use of clinical decision support tools (CDS) are growing in popularity with great potential to improve both quality and cost of healthcare delivery.
While CDS is not the solution to structural problems in healthcare or a replacement for patient shared decision making, it has the potential to address biases in care, provide less experienced physicians with additional guidance, and provide personalized patient care.
There are both clinical and non-clinical causes for variation in clinical care, and understanding these variations will better help us understand what areas CDS can address.
As clinicians become more experienced, their clinical acumen improves, and diagnosis and treatment become more efficient and accurate. When physicians first finish training and begin practicing independently, it is not uncommon for them to seek out the advice of mentors.
A recent study analyzed the utilization of CT imaging to detect pulmonary embolism and found that greater experience and training was associated with less utilization and greater diagnostic yield.1 Experienced physicians were able to identify high risk patients and order imaging accordingly, saving both healthcare dollars as well as decreasing patient radiation exposure. CDS would be a useful resource for less experienced clinicians who may benefit from additional guidance.
Providing CDS tools for physicians, such as the sleep deprived internist at 3:00 AM, may be appreciated support.
A phenomenon deemed “clinical inertia”, or the resistance of a clinician to intensify treatment for medical conditions that are not well controlled, has been found to affect the treatment of multiple chronic diseases.
The cause of inertia may be influenced by the complexity of patient care and hesitation to act because of an unclear ideal course of action. While determining the first and second antihypertensive medication is typically straightforward, treating complicated patients with multiple comorbidities with polypharmacy is more complex.
Clinical inertia increases the time it takes for patients’ conditions to become controlled, sometimes taking over a year. However, the use of alerting systems to counteract clinical inertia in treatment of chronic diseases, like hypertension and diabetes, have been shown to improve the rate of control. 2,3,4
Thus, CDS helps to identify patients at risk and in the future may even be used to assist creating treatment plans for complicated patients and facilitating timely follow ups.
Physical and emotional exhaustion, sleep deprivation, and illness can all influence decision- making capabilities. Physicians may be particularly susceptible as their jobs are often highly stressful, emotionally taxing, and cause disturbance in circadian rhythm (sleep/wake cycle) due to long shifts and overnight call schedules.
There have been increasing measures to limit work hours and provide additional support to trainees, but once done with training, many physicians are put in positions where they are required to work long hours. Providing CDS tools for physicians, such as the sleep deprived internist at 3:00 AM, may be appreciated support.
As physicians are human, it is not surprising that bias in healthcare treatment has been demonstrated in literature from as early as the 1900s.5 Studies have shown a disparity in treatments in patients of lower socioeconomic groups, racial minorities, and gender.6,7 This likely extends to other patient characteristics as well, such as obesity, non-compliance, and ill manners.
When the only evidence available are case studies or small samples, there is understandably variation in treatment.
Studies have shown that blacks with coronary artery disease were less likely than whites to be referred for coronary artery revascularization.8 They are also more likely than their peers to receive an amputation, versus a salvage procedure, when presenting with significant peripheral vascular disease.5 Women who are found to have abnormal cardiac stress tests are less likely to undergo any subsequent diagnostic testing, including secondary stress testing or angiography.9
Incorporating CDS into routine care provides an opportunity to clearly present the standard of care and help decrease variation due to physician bias.
With rare diseases, there is often a scarcity of high quality data on which to base clinical decisions, and development of a care plan can be challenging. When the only evidence available are case studies or small samples, there is understandably variation in treatment.
With the increasing availability of big data, there is potential to collect enough information to identify superior treatment courses and employ CDS to help make management decisions.
Along the same lines, perhaps the most exciting development is the coming of personalized precision medicine. With the availability of big data, it will be possible to identify cohorts of patients who are similar, which will provide more information to support treatment decisions.
Guidelines are constantly changing and there is sometimes even conflicting information from different studies and different suggestions from various organizations; the ground truth is not clear.
Instead of providing generalized statements about progression, we will be able to provide specific advice and identify the most impactful modifiable factors using details such as: The patient is 57 years old, has a history of diabetes mellitus type 2, stage IV chronic kidney disease (CKD), and hypertension, and his blood pressure has been well controlled for the past 10 years.
While there is amazing potential for CDS to impact the practice of medicine, the route to implementing these tools is not straightforward. Guidelines are constantly changing and there is sometimes even conflicting information from different studies and different suggestions from various organizations; the ground truth is not clear.
Perhaps a way to address this would be to present institutional guidelines on current data but provide the source of the studies so that physicians can review the evidence if they so desire.
Lack of insurance is clearly a barrier to healthcare services, but even the type of insurance coverage dictates which tests, procedures, appointments, and medications are covered. Co-pays and other costs may be prohibitive to implementation care plans; the provider selects the best choices within the options presented. If an insurance does not cover a particular medication, CDS will not be able to solve this, but can help determine the best treatment plan given the options available.
In using large data and precision cohorts as models, we will have vast improvements in ability to provide personal patient advice. However, it is important to remember that the medical record does not necessarily encompass the entire clinical situation, as not all information is documented in the clinical chart. At this point in time, CDS should be considered a tool to be used as a guide as opposed to the ground truth.
The desire to work closely with patients is why many physicians choose their career and encouraging patient centered practice should be guarded in the implementation of CDS.
While grounded in evidence based practice, caring for patients is often termed the “art of medicine” for a reason. Physicians do not only spend years studying the sciences and in specialty clinical training, but they are also required to learn how to expertly communicate with patients.
They learn to understand patient wishes and combine clinical and non-clinical factors so they can work with the patient to develop personalized treatment plans. For example, the patient with brittle type 1 diabetes mellitus with frequent episodes of hypoglycemia may not tolerate a low A1c. Likewise, an 85-year-old man with dementia may not need 25 medications.
The desire to work closely with patients is why many physicians choose their career and encouraging patient centered practice should be guarded in the implementation of CDS.
In summary, CDS has the potential to improve the quality of care although appropriate implementation of CDS is critical to its success and will not be a simple task.
The role of CDS is not to take the place of a physician but be a tool to help physicians make more educated decisions while allowing for freedom of clinical judgement when appropriate.
To be successful, it should not be overly onerous to use, as physicians are already short on time. Ideally it will decrease the workload for physicians and allow for more physician-patient time.
To accomplish this, development should be done with the input of practicing physicians. It will be exciting to see how CDS transforms the practice of medicine for both physicians and patients.
Footnotes
- Venkatesh, A. K. et al. Trends and Variation in the Utilization and Diagnostic Yield of Chest Imaging for Medicare Patients With Suspected Pulmonary Embolism in the Emergency Department. Am. J. Roentgenol. 1–6 (2018). doi:10.2214/AJR.17.18586
- Josiah Willock, R. et al. Therapeutic Inertia and Treatment Intensification. Curr. Hypertens. Rep. 20, 4 (2018).
- Josiah Willock, R. et al. Therapeutic Inertia and Treatment Intensification. Curr. Hypertens. Rep. 20, 4 (2018).
- Khunti, K., Wolden, M. L., Thorsted, B. L., Andersen, M. & Davies, M. J. Clinical inertia in people with type 2 diabetes: a retrospective cohort study of more than 80,000 people. Diabetes Care 36, 3411–7 (2013).
- Stapleton, S. M. et al. Variation in Amputation Risk for Black Patients: Uncovering Potential Sources of Bias and Opportunities for Intervention. J. Am. Coll. Surg. (2018). doi:10.1016/j.jamcollsurg.2017.12.038
- Hajjaj, F. M., Salek, M. S., Basra, M. K. A. & Finlay, A. Y. Non-clinical influences on clinical decision-making: a major challenge to evidence-based practice. J. R. Soc. Med. 103, 178–87 (2010).
- Fearnhead, N. S. et al. Variation in practice of pouch surgery in England – using SWORD data to cut to the chase and justify centralisation. Color. Dis. (2018). doi:10.1111/codi.14036
- Kressin, N. R. & Petersen, L. A. Racial differences in the use of invasive cardiovascular procedures: review of the literature and prescription for future research. Ann. Intern. Med. 135, 352–66 (2001).
- Daugherty, S. L. et al. Implicit Gender Bias and the Use of Cardiovascular Tests Among Cardiologists. J. Am. Heart Assoc. 6, (2017).
BIO
Christina Chen, MD is a physician scientist who is an Instructor at Harvard Medical School, Nephrologist at Beth Israel Deaconess Medical Center (BIDMC), and a research scientist at MIT. She has a background in biomedical engineering and computer science. She attends on the nephrology consult service at BIDMC and has an outpatient renal clinic. Her team’s “sentiment analysis” project won the Big Data competition at the London Datathon in 2016 and it was presented at the Intensive Care Society State of the Art Meeting. With her background in engineering and medicine, she hopes to bridge the gap between data scientists and clinicians to answer innovative questions. Her current research interests include studying acute kidney injury and using echocardiography to determine effects of cardiac dysfunction on outcomes.
