As science invents new cures and tools to fight disease, an intractable problem continues to plague American health care: medical errors by doctors and hospitals.
The issue of medical error is hardly a new one: We have just passed the 20th anniversary of the release of “To Err is Human,” a landmark government report that exposed the shocking toll of medical error in the United States. It sparked widespread media attention, public hearings and action by the Clinton administration.
The report put the annual number of deaths resulting from medical error at 98,000. More recent studies have indicated an amount more than twice as high, making it the third leading cause of death in the United States. About one in 20 patients are exposed to preventable harm, according to a recent study, with 12 percent of those mistakes resulting in permanent disability or death.
Yet there are no marches, fundraisers, or lobbyists in Washington waging a fight against medical errors. Funding for research is negligible compared to that spent on diseases like cancer. The issue does not receive the attention it so desperately deserves.
The crisis in medical errors persists and even continues to grow. Although some progress has been made in a few areas of medicine, deadly medical errors continue to put millions of patients at risk every year.
One particularly troublesome area is diagnostics. Most of us will be given the wrong diagnosis at least once during our lifetime. Three conditions – cancer, vascular events, and infection – accounted for three-quarters of the cases, causing immeasurable human harm and suffering and an estimated $1.8 billion in malpractice suits.
But haven’t the advances in technology, such as the introduction of electronic medical records and 3D imaging, reduced the number of medical errors in diagnosing disease? Sadly, the advent of such technologies can create more problems than they solve. The tidal wave of new data that these technologies create and propagate is both a blessing and the bane of health-care providers and hospitals.
For example, electronic medical records have led physicians to often spend twice as much time with their screens than with their patients, contributing to the physician burnout that affects more than half of doctors, according to one recent study.
The result: more than half of the doctors in one recent study showed symptoms of burnout. And they were also more than twice as likely to have reported a diagnostic or other medical error in the previous three months.
Or take radiology. The growing demand for increasingly complex imaging has meant more data and images for practitioners. Radiologists at one institution saw a 10-fold increase in the number of images from 1999 to 2010, which resulted in them getting just four seconds on average to review each image.
And the pace of expansion in imaging has only increased over the past 10 years. About 4 percent of interpretations by radiologists end up having errors, according to some estimates, which translates to roughly 40 million errors a year globally.
But science is just now beginning to provide promising solutions. In radiology and other fields, artificial intelligence uses algorithms to optimize the analysis of images and continuously improve the detection of abnormalities. Machine learning and artificial intelligence can detect errors that people – particularly overworked people – miss.
A comprehensive review of published studies recently found that deep learning algorithms had performed roughly on par with health-care professionals in making diagnoses based on imaging. In contrast, other studies have shown even more promise for the technology. For example, one study of CT lung cancer screenings found that combining a radiologist with a computer detection system resulted in better performance than having a second radiologist do an analysis.
To be sure, these new technologies have risks, such as how they affect patient privacy or contain hidden biases against certain demographic groups, and we must be cautious as we bring them online. But these technologies can and should be used to reduce medical errors.
There is plenty that government, industry, hospitals, and health plans can do to support the fight against medical errors. The Food and Drug Administration rightly is cautious in approving new technologies that diagnose illnesses.
But it could streamline pathways for the approval of new non-diagnostic technologies and tools that counter medical errors. The government and industry could invest more in research and development of those technologies. And hospitals and health plans could give medical errors a higher priority.
Although medical errors have received far too little attention in recent years, we are entering an era where it may be possible to make real progress. The continued march forward of the digital revolution – and with its artificial intelligence and machine learning – should be able to turn the tide on this intractable epidemic.
Pelu Tran is the chief executive officer of Ferrum Health, a health technology company based in San Francisco.