Background Information

FreeKnee offers information for those wishing to learn about the possible condition(s) causing their knee problem.

The factors necessary for making an accurate diagnosis of a patient's knee problem requires the judgment of a physician. However the medical literature documents that even the most expert surgeons are not perfect in their clinical diagnoses.^1-6 It is also reported that use of various single physical exam tests will not produce 100% clinical accuracy.^7,8 Even extensive physical examination findings for a single diagnosis were not very accurate.⁹ It may be surprising to learn that even the very sensitive MRI may not be better than the surgeon's clinical impression.^10,11

The diagnostic challenge is complicated by the fact that it is rare that there is only one problem in the knee. For instance, the patient may have an injury that resulted not only in a torn cartilage, but a torn ligament. If there were previous injuries, then arthritis of variable magnitude would complicate the issue.¹² Therefore, taking multiple factors into consideration is more accurate in determining a diagnosis.¹³

Traditionally, the diagnostic process is initiated with a physician taking a medical history by questioning and examining the patient. The physician forms what is called a clinical impression following the initial patient encounter. The clinical impression is speculative. The physician then selects one or more additional tests to better define the diagnosis. He/she must choose the tests that have the greatest predictive value for any given patient's condition. This again requires judgment since predictive values of various tests are either not known or provide little direction.⁹

Laboratory blood tests and or various imaging studies may be recommended. These may be plain film x-rays, arthrogram (dye injection into the joint), CAT scan, bone scan, and more recently MRI. This information must be considered circumstantial evidence. A definitive diagnosis is hopefully achieved by the various tests and combined with the surgeon's judgment results in the appropriate treatment recommendation. Direct evidence for the diagnostic verdict would include surgery with visual inspection. Presently this is accomplished by arthroscopy.

However, all of these diagnostic methods are subject to human error, equipment failure and/or misinterpretation. Inaccuracy may accompany every method. For instance, the patient may not be an accurate historian. They may forget or even withhold information. The physician may not ask a uniform set of questions, let alone knowing whether the questions have predictive value. The physician often relies only on antidotal evidence based upon experience or lack thereof.

The physician may not request the necessary or definitive laboratory tests or imaging evaluations. The laboratory tests are subject to interpretation. The imaging documents are subject to human error in creation or by interpretation. Even arthroscopy providing for direct inspection of the joint may be limited by the physician's technical ability or instrumentation. Visual arthroscopic documentation may not be permanently stored for subsequent study. Even the pathologic diagnosis may be subject to error in what tissue was submitted and the experience of the pathologist. There can be many factors that lead to uncertainty in making a diagnosis and the subsequent treatment plan.

It is apparent that new means of determining the diagnosis would be welcome. Hence, the introduction of FreeKnee.

The information age brings such a new and powerful tool to assist in making a knee joint diagnosis. It is based upon the power of data. Utilizing massive databases from clinical practices, it is now possible to select questions that have scientifically established predictive power to make a diagnosis.¹⁴ The differentiation was possible in that a comparison could be made to a cohort of men and women who had always considered their knees normal.¹⁵ A random set of patients in the database was used in the analysis. The remaining patients were used to validate the method and establish the reliability. The method was further validated by applying it to similar databases from other surgeons. This is all accomplished by computer computation.

Unlike most Internet site diagnostic methods that provide general information, FreeKnee provides a personalized report concerning your knee problem. This is accomplished by assessing your responses to a set of uniform questions, each selected for its predictive value in determining a known knee joint diagnosis. For your convenience we offer links to Internet sites with general information on knee problems.

FreeKnee was developed by Information Health Network—a small medical software research and development company founded by Lanny L. Johnson, MD, of East Lansing, Michigan, in 1982. Dr. Johnson, a pioneer in arthroscopic surgery, initially developed a proprietary computerized medical record to better communicate with his patients and to facilitate his clinical research. Andrew W. Pittsley, an electrical engineer by training, has been the Chief Software Architect. Information Health Network's software and various information systems facilitated the development of an extensive clinical database unlikely to exist elsewhere. This database provided the material for many scientific publications, some of which are listed under References.

In a 1996 pilot study based on the database, the authors confirmed their hypothesis that a differential diagnosis could be made from medical history data alone.¹⁴ An analysis was made of the data to select the fewest possible questions with strong predictive value that would produce consistently reliable results for making a differential diagnosis of a knee problem.¹⁴

A method similar to the one used in the 1996 report was used to construct a medical history questionnaire—the central focus of KneeProblem.com. KneeProblem.com was launched in 1999. Using scientific methods and statistical analysis, the authors listed the hundreds of questions in this extensive database in order of their diagnostic predictive value, greatest to least. In general, the greater the number of questions used, the higher the reliability. An unlimited number of questions would not be practical, however, so the number selected was balanced against the degree of overall reliability. In the end, the top 70 predictors were chosen. These were validated and found to yield 80% reliability in predicting the presence or absence of surgical indications. The FreeKnee analysis produces a list of probable, possible, and unlikely diagnoses, including the specific reliability of each diagnosis for each person. This is something no other method, even a physician or MRI is able to do.

Comparison with Reports in the Medical Literature: Reports in the literature show great variation in overall accuracy from one surgeon to another.^1-6 Accuracy also varies for different diagnoses—probably due to the different methods used by different authors in their studies. The main variation has been in the criteria used to determine accuracy. The stricter the criteria, the lower the rate of accuracy.

The literature does not indicate what factors various surgeons considered in making their diagnoses. There is no apparent uniformity in the data collected by any individual physician or among physicians. There may not have been uniformity of data collection from one patient to another. Perhaps more important, the statistical power of the questions asked by various surgeons in making their diagnoses has not traditionally been presented; the method of data collection for clinical assessment may have been different for each patient. This method is known as the "intuitive practice of medicine"—with the surgeon's diagnosis based upon experience and opinion.

This intuitive practice of medicine is now being replaced by "evidence-based medicine"—made possible by modern computer technology. FreeKnee is an example of evidence-based medicine. Real data, science, and statistics were used in developing the FreeKnee analysis. The report that is generated from responses to the questionnaire is more reliable than most published studies.^1-6

In fact, a FreeKnee report is more likely to be accurate than a clinical impression by the surgeon whose data was analyzed.¹ Yes, although one might think the surgeon would be more accurate—having seen, interviewed, and examined the patient and also reviewed x-rays including MRI, the facts prove otherwise.

For what possible reason might an analysis based on data out-perform an actual surgeon? How could an analysis based on data be more reliable than that of experts reporting their clinical accuracy in the literature? It is the difference between opinion and fact, between intuition and evidence.

The report of diagnosis from FreeKnee is 80% accurate overall. This compares favorably with a report in literature that compared the physician's clinical impression with MRI results in 238 patients. Both were 79% accurate when compared to what was seen at arthroscopy. There was a 77% agreement with the physician's clinical impression and the MRI. They also reported that MRI did not reduce the negative arthroscopy any more than the clinical impression would have determined.⁶

While reports in the literature lack vital information, the algorithms and software used to construct FreeKnee are based upon evidence.**

Efficiency: FreeKnee is time efficient. Completing the questionnaire requires about ten minutes—less time than it takes to travel to and from the doctor's office in most cities; less time than is required to make an appointment or fill out forms in the surgeon's office, let alone read a magazine and wait undressed in an exam room; less time than is spent getting the obligatory x-ray or elective MRI; less time than one spends listening to the surgeon explain what may be wrong with one's knee.

Cost Effective: FreeKnee meets the strictest definition of cost effective, since there is no charge to the patient. By comparison, an initial orthopedic office visit for a private patient in Michigan typically costs between $100 and $200. Elsewhere it may be more. In Workers' Compensation cases, an independent medical exam may cost $350 to $500. Add to this the usual plain film x-ray at $100 and the possible MRI at $750. Also, with the traditional evaluation method the patient may have to pay for parking, or take time off from work to see the doctor. FreeKnee reduces these costs to the patient and/or employer. Likewise, it is cost effective for patients who must justify going out of their health plan for a second opinion or treatment. The service is provided by the sponsoring clinic or physician. Of course if the patient decides to see the sponsoring physician or any other for further opinion, there would be a charge for those services.

Surgical Indications: Undergoing surgery is not to be taken lightly. Patients rightly desire and deserve to know that surgical intervention is really necessary and will benefit them.

Yet many accepted surgical indications are based on physician opinions—not on evidence. When medical information is statistically analyzed to determine its predictive value, the critical predictors frequently turn out to be different from the ones surgeons use.¹⁴ This may be hard to believe, but it is none the less true in the Information Age.

Also, it has been recognized in our studies that certain traditional surgical indications exist in people who have never had knee problems.¹⁵ This may be hard to believe, but is none the less true. Therefore, the issue of surgical indications must be reexamined in light of this evidence. An alternative means of determining the presence or absence of surgical indications would be welcome and is present in FreeKnee.

The FreeKnee software offers one such alternative—enabling patients with knee problems to solve their problem more quickly, more accurately, and without expense.

**As with any other single diagnostic tool, FreeKnee cannot do everything. It should be noted that the FreeKnee analysis is unable to predict the presence of a torn lateral meniscus based upon lack of questions with high predictive value. In the case of rheumatoid synovitis or tumor, they were not present in sufficient numbers in the database to permit statistical validation. The analysis does not cover patients who previously had a total knee replacement. The judgment of the physician is the critical factor.

References

The following scientific publications serve as background for comparing this new diagnostic instrument, FreeKnee, with reports in the medical literature on the accuracy of the orthopedic surgeons' opinions as well as reports on MRI evaluations for determining the cause of knee problems.