Carl van Walraven, MD, FRCPC, MSc

carlv@ohri.ca

When physicians don't communicate properly with each other, what is the global effect on patients? If medical data is neither gathered systematically nor stored methodically, how can resources be allocated appropriately? How can the potential research areas, which lead directly to patient benefits and health-care system improvements, be accurately defined?

These are the questions studied by Dr. Carl van Walraven, whose particular interests within the field of health services research are inter-physician communication and the systematic collection of health-related data. In projects exploring how health care is delivered and where potential inefficiencies lie, Dr. van Walraven hopes to bring his two interests together.

Dr. van Walraven's first goal is to create a communications infrastructure for physicians. This would be designed not only to improve communication between physicians, but also to simultaneously collect health-related information in a standardized fashion. As a first step of this project, Dr. van Walraven is collecting specific information on every patient who goes through the Ottawa Hospital. The resulting data base will then allow him to study how health-care is delivered.

But the data base will also make it easier for physicians to create a report--known as a "discharge summary"--that is generated for each patient for whom they have cared in hospital. Using codes identified in the data base, the physicians will use simple forms to record specific information. This will relieve them of having to dictate the full discharge summaries that the current system requires--an onerous task that is often carried out imperfectly.

Dr. van Walraven's Discharge Database will address another problem with the status quo. When a patient leaves the hospital, the discharge summary created by the physician is essentially a narrative letter containing particular information. This letter tends to linger in the system for a day or two, after which it gets typed up and distributed to various physicians, such as the patient's family doctor. But at the same time, the patient's chart is on its way to Health Records. Here an analyst assigns specific codes for various diagnoses and procedures. These codes are used for various purposes, such as financial remuneration, research, and health planning. The problem is that the two streams of information--the discharge summary from the physician and the coding summary from the chart--move through the system completely independent from one another. When the chart arrives at Health Records, the discharge summary has invariably not yet arrived. Worse yet, the charts are notoriously difficult to read. Thus, the data in the coding summaries can be unreliable. Furthermore, because dictating discharge summaries is viewed as a chore by physicians, their quality is suspect.

Dr. van Walraven's Discharge Database, by contrast, could make life easier for both the physicians and the health records analysts, may save the hospitals money, and should result in better and more efficient coding and discharge summaries. Simultaneously, it will gather high-quality clinical data. Dr. van Walraven's ten-year goal for this project is to see hospitals across whole regions--or better yet, across the whole province--convert to this system, or one that is similar. He points out that administrative databases--of OHIP claims, drugs, hospitalizations, etc.--already exist in Toronto that cover the whole province and everyone in it. However, the quality of this data, which was not collected for research, is questionable. The usefulness of a high-quality clinical database in studying how health-care is delivered cannot be overemphasized.

That such a database would be an invaluable tool is illustrated to a large extent by Dr. van Walraven's second major project, as he will show how comprehensive, high-quality clinical data might be used to reduce inappropriate lab use. It may seem obvious that lab tests, for example, should not be repeated unecessarily. But unnecessary repetition is surprisingly difficult to identify. No one has systematically determined how much time must pass between repeat tests before significantly different results will differ. This is complicated by the fact that while some lab tests can change in value very quickly, others take a long time to change. Blood cholesterol levels are a good example of something that changes slowly. If a patient's cholesterol is measured today, the result should not change significantly over the next two weeks independent of what happens clinically. Even if the same patient has a heart transplant, his or her blood cholesterol level should not change significantly. Or, says Dr. van Walraven, that's the hypothesis.

Dr. van Walraven is currently devising ways to identify inappropriate test repetition, while simultaneously establishing time frames within which results won't change for nine specific lab tests. As the first step, he and his team are collecting data--patient id, physician id, date of test, and test result--for every one of these specific lab tests done in every commercial and hospital lab in Eastern Ontario. While it wasn't easy to convince so many disparate bodies to participate, it was critically important. Methods differ between the commercial and hospital sector, and if either declined to be involved, it would be impossible to obtain an accurate picture. Now that all have agreed to participate, the database will allow the team to discover any duplication that occurs--something that has never been done before. But what excites Dr. van Walraven even more is that he will also be able to analyze the data at a deeper level, for example by determining such things as where hot spots of repetition are occuring geographically. Areas where little repetition occurs will also be revealed, and the reasons investigated. This should be the case in Perth and Smith's Falls, for example, where family doctors have direct electronic connections to a data base.

Dr. van Walraven and his team will also be able to determine how long it takes before test results change significantly. Results for five of the lab tests, including cholesterol, change very slowly, while four of the tests have extremely short windows of stability. As a final step, the team will figure out why duplication is occurring. Clearly, physicians realize that if a particular test (like cholesterol) was done yesterday, it's unnecessary to do it again today. Thus, inappropriate test repetition can be used as what Dr. van Walraven terms a health-system "duplicometer" to identify areas of poor integration of health service delivery. His hope is that simple, routinely collected data will thus allow us to survey the health-care system to discover inefficiencies. By linking his database with others, he'll be able to follow patients through the system over time, and determine whether and why tests are being repeated unnecessarily.

That's good news for everyone concerned about the efficient use of health-care resources.