For more than a decade, Joe Bridges, DHI, witnessed his sister Jan struggle to figure out what was causing bouts of severe swelling; and then he decided to do something about it.
Beginning at age 77, Bridges dedicated three years to researching and implementing a solution that can help physicians consider a wider array of possible diagnoses to get it right the first time to help prevent patients, like his sister, from suffering unnecessarily.
“Some might think that it’s a little odd that a retired mechanical engineer would be interested in health informatics, and thinking about diagnoses, but my journey has been a series of events and opportunities leading quite logically to this point,” he said. “I have seen diagnostic inaccuracy firsthand.”
For Jan, the onset of severe swelling would happen at random and last for days. The root cause eluded doctors for 12 years, but after consulting with four physicians and multiple cross-country specialist excursions, Jan finally got a correct diagnosis — acquired angioedema. Today, she is symptom-free with proper treatment.
The DHI program is designed for working professionals who already have a master’s degree and want to stay in industry, rather than go into research. Instead of a dissertation, candidates in the program must complete a large-scale translational project in a health care organization.
“Having known Joe Bridges for nearly a decade, I am aware that being the first Doctorate of Health Informatics in the nation is a reflection of just one of the many achievements in his remarkable life and career,” said Jiajie Zhang, PhD, School of Biomedical Informatics dean. “Dr. Joe Bridges, as I will soon refer to him, is a credit to our school and the field of informatics, and he is both a friend and source of inspiration to me.”
Bridges’ project centered on diagnostic inaccuracy — instances when a physician makes an incorrect diagnosis during the first encounter with the patient, when all defining symptoms of their illness are present. He dove into how often it occurs, what the root causes are, and how to prevent it.
Diagnosing is one of the most difficult tasks physicians do. There are an estimated 200 symptoms that can present in 10,000 different conditions.
Studies suggest that in primary care about 5% of people are incorrectly diagnosed, equating to roughly 12 million people annually in the United States. Half of those will likely suffer a negative health outcome. Studies show that in three-fourths of cases of inaccurate diagnoses, all elements needed to make a correct diagnosis were present at the initial physician visit.
“Six million is a number so big that it loses its significance,” Bridges said. “A 5% error means 1 in 20 patients are misdiagnosed. If a physician sees four patients an hour over a 10-hour day — that’s 40 patients. Statistically, two of them are misdiagnosed, and one will suffer harm.”
Bridges spent two years evaluating possible solutions to help reduce diagnostic inaccuracy as part of his translational project, deciding that the best solution available with current technology is a computerized diagnostic decision support system. Bridges evaluated products on the market.
“The idea of introducing a system like this is to get ahead of a possible mistake, to present alternatives for consideration beyond an initial impression,” Bridges said. “Researchers call it ‘premature closure,’ when a conclusion is reached too soon. After that, you can see confirmation bias, a tendency to emphasize information that supports the premature diagnosis and dismiss information that may conflict with it.”
Bridges ranked currently available diagnostic decision assistance tools, in terms of diagnostic retrieval accuracy, speed, and usability. Physicians input patient information and symptoms, and the chosen program searches a database of scientific reference literature to create a ranked differential diagnosis list.
Bridges conducted two studies of the program to determine if the diagnostic decision-making tool would be useful to medical providers.
He first did a case study with 120 McGovern Medical School at UTHealth Houston residents. They were divided into groups and asked to evaluate 24 cases, with and without the use of the software.
His results found an absolute 9.1% average improvement in diagnostic accuracy using the software, with 75% reporting the tool would be useful in routine clinical practice.
The next step was to test the program in a live clinical setting with input from UT Physicians, the academic medical practice of McGovern Medical School.
Bridges collaborated with the UTHealth Houston IT department to purchase and install the software, and received approval from the UT Physicians Outpatient Quality Council to install the software into the existing electronic health record system.
Over 86 days following implementation, Bridges collected usage data and then recorded anonymous survey results from physicians.
In total, there were 37 individual faculty users and 37 resident users, with 57% of them using the system a single time. The bulk of uses were for geriatric patients, who have many preexisting conditions.
Physicians reported they were most likely to use the diagnostic decision support system for a challenging diagnosis, often an unusual combination of symptoms, or if a previous treatment was unsuccessful for a returning patient.
Within the professional group, 77.8% of survey respondents said they thought the system would be helpful in routine clinical practice, and 36.8% said the system prompted a change to their original diagnosis.
Bridges’ study results show improved diagnostic accuracy using computerized decision assistance tools, but there are more steps for these types of systems to gain widespread use.
The decision to contract for these kinds of tools is at the individual institution level, so there must be institutional encouragement and a cultural embrace of their use, he said.
Journey to UTHealth Houston
Bridges’ first encounter with UTHealth Houston was through its development board, a group of community leaders that advances the mission and vision of the institution by increasing public awareness and increasing philanthropic support.
He joined the board in 2011 on the recommendation of a friend.
At the time, his sister Jan was struggling with her inaccurate diagnosis (that would not be solved until 2016), and he had recently lost his mother to a hospital-acquired infection.
“I felt helpless. I’m not interested in being misdiagnosed, and I’m not interested in my family being misdiagnosed,” he said. “I thought, ‘UTHealth Houston is a place that can address some of the issues I can’t address myself.’”
Since 2012, Bridges has served on the advisory councils for Cizik School of Nursing at UTHealth Houston, as well as the School of Biomedical Informatics, where he served as chair of the council from 2018 to 2020.
Bridges’ daughter Beth Cozby also serves on the development board’s Executive Committee. She and other family and friends have a special section at the Friday, May 13, graduation ceremony.
“We are all in awe of his dedication and diligence, but not surprised that he is the first graduate of a cutting-edge field of medicine,” Cozby said. “It makes perfect sense to those who know him. To say he is a lifelong learner is an understatement. He loves to read and do research on a wide variety of topics. He has always enjoyed and been interested in technology. So, factoring in his sister’s health care experience, and the data analytics aspect of the field, it just makes sense that his pursuit would be in bioinformatics.”
As council chair, Bridges went to dinner with Dean Zhang, Michael Blackburn, PhD, and UTHealth Houston President Giuseppe Colasurdo, MD, to discuss ideas for programs at the School of Biomedical Informatics — one of them being the idea of addressing diagnostic inaccuracy. Colasurdo suggested Bridges enroll in the DHI program, and he did with vigor, completing the program a year ahead of his cohorts.
An engineer by trade, Bridges spent his career working in the oil and gas industry in Texas and Louisiana, as well as in banking and ranching.
Bridges holds a Bachelor of Science and Master of Science in mechanical engineering from The University of Texas at Austin, and is a graduate of the School of Banking of the South at Louisiana State University in Baton Rouge, Louisiana. His critical thinking skills honed over decades mapping and tapping into oil and gas reservoirs have been helpful in his doctoral journey.
“My journey has brought me from helplessness, to hopefulness, to action — with the opportunity to move from a philanthropic support role to being an agent of change in improving diagnostic accuracy in routine clinical practice. It has been quite a journey, and I hope this will not end with the graduation ceremony as this project is, in my opinion, imminently scalable beyond UTHealth Houston, perhaps to The University of Texas System, and maybe even further.”