Convergence of Medical Data and IT

The Hospital Information Management Systems Society (HIMSS) invited me to make a presentation at their midyear educational conference in June.

The audience was primarily health care chief information officers and health care information technology vendors. My presentation was in the “Emerging Technologies” track and was titled “Convergence of Medical Device Technologies With IT.” The premier society of hospital information systems realizes that, although it appears that medical technology is now computer driven, there is still a critical gap between clinical data and the information systems to which these data are being provided.

In 1983, the three great lies in health care were: Radiology will be filmless in 3 years, health care will be paperless in 5 years, and the US Food and Drug Administration will approve our product in 90 days.

In 1993, the three great lies in health care were: PACS will make us filmless, the electronic medical record will make us paperless, and FDA approval will make it interface with other medical equipment.

In 2003, the three great lies in health care are: This equipment is fully compliant with DICOM standards, this system is fully HL7-compliant, and FDA approval assures compliance.

In 2013, the three great lies in health care will be: You won’t need a laser camera for this imaging equipment, you won’t need a laser printer on this computer network, and FDA approval assures compliance with the Health Insurance Portability and Accountability Act of 1996.

So what is so difficult about making clinical computer systems provide information systems with data?

The difficulty seems to be in expressing data from the clinical domain in an information domain. Patient monitoring in the clinical domain is composed of electrocardiogram waveforms and blood pressures traces. When we try to transpose this continuous visual information into the information domain, we end up with only fixed data points like heart rate and systolic, diastolic, and mean blood pressure in a database file.

In medical imaging, computed tomography (CT), magnetic resonance imaging, nuclear medicine, positron emission tomography, and ultrasound images are all clinical domain products. The transposition to the information domain results in calculations like percentage of vessel occlusion, flow rates, and tumor volume. If a picture is worth a thousand words, it should take an entire book—not an 80-word interpretation—to describe a 500-image study from a new 16-slice CT scanner.

In our biomedical service and support area we have spent years developing asset management systems that work in a clinical domain. We identify equipment by application, age, and technical performance. When asset management is transposed into the information domain, all we get is ownership demographics and an automated reminder of the date on which we need to locate it again for testing or service. We lose data on clinical applications, amount of utilization, and appropriate training for users.

If clinical data are ever going to converge with information data and become a transparent transition, we are going to have to use the information technology to improve, not restrict, the clinical data. For 30 years, telemetry central stations have required a trained scope watcher to interpret the ECG waveforms and separate noise artifacts from life-threatening arrhythmias. The growing utilization of combined ECG and SaO2 telemetry can allow computer interpretation systems to track both oxygen saturation and arrhythmia detection for an automated and transparent analysis in clinical decisions. If the ECG appears to be changing dramatically, but the oxygen saturation is not changing, an information alarm can be initiated to check the ECG electrodes and patient wires. If the automated monitoring program detects a steady reduction in oxygen saturation, it can increase the sensitivity in the ECG analysis. The goal of an automated telemetry monitor should be to identify every cardiac arrest 2 minutes before it happens rather than 2 seconds after it happens. The next step will be to determine the patient’s location and notify members of the arrest team closest to the patient without monitoring staff interaction.

In medical imaging, the new automated mammography image-review programs are changing the tedious job of reviewing 3 years of repetitive images looking for subtle changes that are consistent with tumors. These automated image comparison systems are demonstrating significant improvements in early diagnosis. But will “spell-check” for mammography or chest films improve patient diagnosis at the expense of radiologist proficiency the way spell check for word processing has improved printed text at the expense of spelling skills for this writer?

C. Wayne Hibbs is a principal of Hibbs & Associates, a medical technology and equipment planning consulting group with offices in Dallas and Indianapolis. cwhibbs@cwhibbs.com.