Adam Sturts, MSIV

Artificial intelligence (AI) is transforming health care. The global impact of AI, for better or for worse, will be profound because it is widely applicable across all fields of medicine. Reform of modern medical education curricula is essential for the safe and effective integration of AI into our future health care system. 

Undoubtably, AI has the potential to enhance many aspects of health care in the United States. Hailed by the National Academy of Medicine, “the emergence of AI as a tool for better health care offers unprecedented opportunities to improve patient and clinical team outcomes, reduce costs and impact population health” in the 2019 

Artificial Intelligence in Health Care: The Hope, the Hype, the Promise, the Peril

Today’s clinicians have access to an immense volume of medical data and information. In fact, the wealth of medical information is so great that it far exceeds human interpretation capacity.

The tremendous value of AI lies in its ability enhance clinician decision-making by interpreting this uninterpretable information.

Despite recent promising examples of AI application in health care, the National Academy of Medicine cautions “we believe it is imperative to proceed with caution, else we may end up with user disillusionment and another AI winter, and/or further exacerbate existing health and technology driven disparities.”

Many concerning AI development and implementation issues have been highlighted in the literature, including biased data sets, poorly generalizable training and validation data sets, unintended consequences, risk/benefit considerations, and inadequate cost of care analysis. Such developmental miscalculations have dangerous health and health care economic implications. Unfortunately, today’s clinician has not been adequately trained to develop and interpret AI, limiting widespread safety assessment and utilization.

The American Medical Association published an article in 2019, 

Reimagining Medical Education in the Age of AI

. The authors recommended an overhaul of medical school curricula with a focus on knowledge management and communication rather than information acquisition.

Such a transformation would resolve several significant issues pertaining to modern medical education and AI. Firstly, it would alleviate some of the psychological burden that medical students currently face as a result of “information overload and concerns about never knowing enough,” factors that have been linked to mental health deterioration among medical students and trainees.

Rather than focusing solely on information acquisition, learners should be taught to collaborate with the AI systems that interpret such information through stochastic and probability-based techniques compatible with AI system development and interpretation.

Across the country, we see the great benefits that AI can provide to our health care; however, medical curriculum reform has not yet caught up to this evolution in medical care.

Despite the continued integration and consideration of AI use in our health care system, one thing that must never leave sight of our care is humanism in medicine. No machine can disentangle the biopsychosocial complexities of our society.

Medicine will remain humanistic at its core. By learning to collaborate with AI, future clinicians will have more opportunity to focus on the non-analytical, humanistic, individualized patient care where currently, technology takes the forefront.

Adam Sturts, MSIV is a fourth-year medical student at Rowan University School of Osteopathic Medicine. This piece reflects the author’s views and not necessarily those of the publication.

Articles

Artificial intelligence has the potential to enhance many aspects of healthcare in the US, including patient and clinical outcomes, while reducing costs.

Telemedicine Tools Emerging in Medical School

A myriad of scientific reports, news articles, and editorials have evoked both discussion and confusion for clinicians trying to make sense of 

The Centers for Disease Control and Prevention (CDC) recommends against the use of chest CT for the screening and diagnosis of COVID-19. Notwithstanding, the role of CT has been a subject of debate during the ongoing outbreak. The following perspectives and recommendations provide direction for frontline clinicians.

Chest CT and Coronavirus Disease (COVID-19): A Critical Review of the Literature to Date

, sought to determine the role of CT in screening and/or diagnosis of COVID-19 based on review of the most frequently cited studies reporting CT sensitivity and specificity. A group of 10 authors from 6 US institutions contributed to the “Clinical Perspective.”

In the study, which was led by Constantine Raptis, MD, of Washington University School of Medicine, authors highlighted the wide range of CT sensitivity values that have been reported. Some of these measurements may be over-estimated in the context of selection bias. For example, the authors noted in their review that it was often unclear why patients were selected for CT testing—this is crucial information for clinicians as clinical context affects sensitivity measurement.

“If a study cohort contains patients who are more likely to have a true-positive finding and less likely to have a false-negative finding, sensitivity will be overestimated,” authors noted.

Also contributing to sensitivity over-estimation was the use of CT as a binary test with an undefined and low threshold for positivity. As described in a previous article, the data was collected from sites of outbreak. Therefore, radiologists may have been more likely to make a diagnosis of COVID-19 resulting in high sensitivity measurement.

None of the studies reviewed by the authors reliably reported high specificity of CT in differentiating COVID-19 pneumonia from other diseases with similar CT findings; thus limiting the use of CT as a confirmatory diagnostic test.

Raptis suggested the aforementioned studies still hold value and reporting the CT features of COVID-19 remains “an important first step” in helping radiologists identify infected patients, but noted the danger of over-interpretation of certain results.

“Test performance and management issues arise when inappropriate and potentially overreaching conclusions regarding the diagnostic performance of CT for COVID-19 pneumonia are based on low-quality studies with biased cohorts, confounding variables, and faulty design characteristics.”

Based on their review, the authors concluded that the radiology literature on COVID-19 does not substantiate the use of CT as a diagnostic test for COVID-19.

This critical review supports the notion that low quality observational and hospital based series should not be used for population inference. These studies include only a small percentage of the globally aggregated cases and, because of that, the findings are not generalizable to larger populations living in different environments.

To provide guidance in the absence of quality evidence, the CDC, 

 have released recommendation statements to guide CT use.

Most recently, the Fleischner Society shared “

The Role of Chest Imaging in Patient Management during the COVID-19 Pandemic: A Multinational Consensus Statement from the Fleischner Society

, provides context for the use of CXR and CT imaging in the management of COVID-19 based on three clinical scenarios.

Representing opinions and perspectives of radiology, pulmonology, intensive care, emergency medicine, laboratory medicine and infection control experts practicing in 10 countries, the guidance that it provides is broadly applicable and perhaps the most clinically useful thoracic imaging recommendations thus far. 

The panel answered 14 key questions corresponding to eleven decision points within 3 clinically driven scenarios based on the anticipated value of the information that thoracic imaging would be expected to provide. The 3 clinically driven scenarios were defined by the severity of the patients’ presenting features, pre-test probability, and the presence of medical resource constraints. The scenarios are presented as easily interpretable decision pathways.

Results were aggregated, resulting in 5 main and 3 additional recommendations to guide the use of CXR and CT in the management of COVID-19.

Essential recommendations are listed below:

Imaging is not indicated in patients with suspected COVID-19 and mild clinical features unless they are at risk for disease progression.

Imaging is indicated in a patient with COVID-19 and worsening respiratory status.

In a resource-constrained environment, imaging is indicated for medical triage of patients with suspected COVID-19 who present with moderate-severe clinical features and high pre-test probability of disease.

It is important to note that the diagnostic algorithm always begins with either RT-PCR or the point of care rapid COVID-19 test, which are both considered first-line tests for active infection.

Clinicians faced with the circumstances described in essential recommendation 3 should use imaging for triage—however, suspicion for other conditions must remain high.

The recommendations did not distinguish between CXR and CT. Clinical context and resource availability should guide modality selection. CXR appears to be poorly sensitive early in the disease course. Comparatively, CT has a better sensitivity at this stage. However, in a resource-constrained environment where CT access is limited, CXR may be preferred unless features of respiratory worsening warrant the use of CT.

The authors acknowledged that the statement represents opinion at a moment in time within an environment where the status of regional epidemics and the availability of critical resources can vary daily.

As an aside, I would like to thank the expert clinicians and researchers from around the world who are working overtime to synthesize vital guidance from the limited data that is currently available. I would also like to thank all of the healthcare workers risking their lives to preserve the health of others.

Adam Sturts, MSIV is a fourth-year medical student at Rowan University School of Osteopathic Medicine. This piece reflects the author’s views and 

not necessarily those of the publication.

Notes from Rounds is a contributor platform from HCPLive® where aspiring medical students, fellows, and early career providers provide written insight and perspective for the medical community. Those interested in contributing to COVID-19 or our coverage of other topics can

A fourth-year medical student writes about the use of CT for diagnosing COVID-19. 

Despite Early Data, CT Alone Not Appropriate for COVID-19 Diagnosis

Results of a new study are offering clinicians guidance on optimal timing of menopausal hormone replacement therapy to attenuate the risk of cardiac events.

The study, which was presented as part of the American College of Cardiology’s Annual Scientific Session Together with World Congress of Cardiology 

, supports the “Timing Hypothesis” of hormone replacement therapy, which suggests benefits and harms of hormone replacement therapy correlate to how soon after the start of menopause the therapy is initiated.

“This is observational research so the conclusions need to be interpreted cautiously,” said lead investigator Matthew Nudy, MD, of Penn State Hershey Medical Center, in an interview. “But this study can provide reassurance to younger menopausal women who are taking hormone replacement therapy that they are not at an increased risk of having a cardiac event.”

To further evaluate the impact of timing on outcomes in hormone replacement therapy in postmenopausal women, Nudy and colleagues conducted a retrospective analysis using the Truven MarketScan Database comparing the risk of cardiac and thromboembolic events between women initiated on hormone replacement therapy who continued therapy after 6 months to those who did not—a design intended to minimized selection bias.

“Six months was used because we wanted to capture women that were selected for hormone replacement therapy, and that continued it for some time,” Nudy commented.

The analysis included women aged 50 or 60 years of age who were given a new prescription for hormone replacement therapy in 2009. Outcomes of interest in the current analysis included 5-year cardiac events, which was defined as a composite of revascularization, nonfatal myocardial infarction, and hospitalization for unstable angina, and thromboembolic events, which was a composite of stroke, deep venous thrombosis, pulmonary embolism, and transient ischemic attacks. 

A total of 6903 individuals were included in the 50-year-old group. Of these, 2529 discontinued hormone replacement therapy and 4374 continued. After 5 years, investigators pointed out there was no statistically significant difference observed in regard to cardiac (1.6% vs 1.8%; OR, 1.2; 95% CI, 0.81-1.76) or thromboembolic events (4.11% vs 4.3%; OR, 1.11; 95% CI, 0.86-1.41).

A total of 3628 individuals were included in the 60-year-old group. Of these, 1208 discontinued hormone replacement therapy and 2420 continued. After 5 years, investigators observed significantly increased risk of cardiac events among those who continued therapy as compared to those who discontinued (2% vs 2.7%; OR, 1.67; 95% CI, 1.01-2.76). There was no apparent difference in thromboembolic events (5.5% vs 4.3%; OR, 0.94; 95% CI, 0.68-1.31). Nudy suggested these findings support the “Timing Hypothesis” for cardiac but not thromboembolic events.

Nudy also pointed out that decision-support tools, such as the North American Menopause Society (NAMS) MenoPro mobile app, are available for free to help clinicians determine which patients are candidates for pharmacologic treatment of menopausal symptoms.

Hormone Replacement Therapy and Cardiac Events in Postmenopausal Women aged 50 and 60 years

Adam Sturts, MSIV, is a fourth-year medical student at Rowan University School of Osteopathic Medicine. His contributions to our coverage are made possible through our Notes from Rounds program. Anyone interested in contributing to our coverage can 

The latest piece coming to us through Notes from Rounds examines how the timing of hormone replacement therapy can impact potential for adverse outcomes. 

Timing of Hormone Replacement Therapy Impacts Outcomes in Older Women

 was debatably the most impactful research of 2019 because it revealed a novel approach to the treatment of heart failure (HF). In this study, dapagliflozin significantly reduced cardiovascular death and worsening HF (hospitalization or urgent outpatient visits requiring IV therapy) for symptomatic HFrEF patients with and without diabetes. Evidence supporting SGLT2 inhibitor use for the treatment of HF continued to accumulate at the American College of Cardiology’s Annual Scientific Session Together with World Congress of Cardiology 

A study presented by Piotr Ponikowski, MD, PhD, of the Center for Heart Disease at University Hospital, Medical University in Wroclaw, Poland on behalf of the Executive Committee of the DAPA-HF Study titled “

Benefit of Dapagliflozin on First and Repeat Events in Patients with HFrEF in DAPA-HF

” was a highlight of the Featured Clinical Research trials on March 30, 2020. The pre-specified secondary analysis of DAPA-HF was designed to capture all HF hospitalizations —including both first and repeat events&mdash;through time-to-first event and recurrent events analysis. The study adds another patient-centric endpoint to DAPA-HF.

“Conventional time-to-first event analysis very often ignores the issue of heart failure burden, which is mainly [centered] around repeated hospital admissions,” Piotr explained during his presentation.

During the follow-up period, 18.2 months, 12% of patients had at least one hospitalization and 4% had at least 2 hospital admissions. 32% of HF hospitalizations were a second or subsequent event, which would have been missed in conventional time-to-first event analysis.

Dapagliflozin significantly reduced the risk of both first and recurrent HF hospitalizations. The time-to-first event analysis demonstrated a 25% risk reduction in the composite CV death or HF hospitalization (HR, 0.75; 95% CI, 0.65-0.85;  

<.0001). Lin-Wei-Yang-Ying recurrent events analysis revealed a 25% risk reduction in the composite CV death or HF hospitalization (HR, 0.75 [0.65, 0.88], 

=.0002). Additionally, joint frailty recurrent events analysis revealed a 29% risk reduction in HF hospitalization (HR, 0.71 [0.61, 0.82], 

<.0001) and a 19% risk reduction in CV death (HR, 0.81; 95% CI, 0.67-0.98; 

Compared with time-to-first event analysis, recurrent events analysis demonstrated a greater absolute risk reduction for the composite CV death or HF hospitalization (3.9 vs 5.3 per 100 patient-years) and HF hospitalization (2.9 vs 3.8 per 100 patient-years), suggesting the therapeutic effects of dapagliflozin are sustained and HF burden reduction achieved. Benefits were consistent across baseline medication use, diabetic/non-diabetic, and age subgroups.

Despite reductions in mortality, hospitalization, and heart failure symptoms, as well as a number of positive cardiovascular and renal outcomes highlighted in recent trials, SGLT2 inhibitors are under-prescribed to diabetic patients in the US. Nicole Chahine, MD, a resident at University Hospitals Cleveland Medical Center co-authored a study presented at ACC.20/WCC titled “

Under-Prescription of SGLT2 Inhibitors in Patients with Diabetes and Cardiovascular Disease in the United States

,” which sought to identify SGLT2 inhibitor prescribing patterns.

The retrospective study included patient data from 2016-2019. 383,750 patients with diabetes on metformin who also had cardiovascular disease were included. Patients with stage 4-5 chronic kidney disease and end-stage renal disease were excluded. Of note, 9.3% (35,590/383,750) of the patients were prescribed SGLT2 inhibitors.

Multivariable logistic regression modeling identified factors associated with prescribing. Factors included being younger than 65 years of age, private insurance, Caucasian, male, and an endocrinology or cardiology visit. The authors pointed out the statistically significant race and sex prescribing disparities and noted that there may be utility in specialty referral to cardiology or endocrinology.

Sometimes the clear benefits of these drugs are overlooked as a result of confusion between prescribing physicians. In addition to specialty referral, robust cross-specialty conversation might also serve as a broader solution to the SGLT2 inhibitor prescribing dilemma.

The patient population selected in the aforementioned study includes some of the highest risk patients. Established SGLT2 inhibitor trials have shown greatest risk reduction within the diabetic population affected by cardiovascular disease. Therefore, the low prescribing rate reported within this population is alarming.

Diabetes affects 30 million in the US and its prevalence is expected to increase by 55% over the next two decades. The 

2020 ADA Standards of Medical Care in Diabetes

 include SGLT2 inhibitor prescribing recommendations for diabetic patients. Under these new recommendations, it is suggested that SGLT2 inhibitors and GLP1 agonists should be considered for patients when atherosclerotic cardiovascular disease, heart failure or chronic kidney disease predominates independent of a1c.

Nevertheless, the present and the future of cardiometabolic health has never been brighter—thanks to the tireless work of clinicians and researchers for their contributions towards advances and major organizations like the ACC and WCC for providing platforms to disseminate the new information.

Adam Sturts, MSIV, is a fourth-year medical student at Rowan University School of Osteopathic Medicine and set to become an internal medicine resident at Penn State Hershey Medical Center.

Notes from Rounds is a contributor platform from HCPLive® where aspiring medical students, fellows, and early career clinicians and researchers to provide written insight and perspective for the medical community. Those interested in contributing to our coverage in the future can 

*The piece reflects the views of the author and not necessarily those of the publication.

A look at the state of cardiometabolic care based on research at ACC.20/WCC from a fourth-year medical student.