Methods

Top Methods Results Discussion Author & Article Info References

All primary care physicians (general internists, family practitioners, and gynecologists) ordering dual x-ray absorptiometry (Hologic 1000W, Waltham, Massachusetts) for bone mineral density measurements between 2 October 1995 and 23 April 1996 at the osteoporosis center of a community teaching hospital were enrolled. When their patients underwent their first procedure, physicians were randomly assigned by a computer-generated allocation sequence to receive either short technical reports or long narrative clinical reports (Appendix) and a printout generated by the bone densitometer software that contained morphometry, bone density data, and statistical comparisons with control groups in tabular format. Specialists with expertise in bone metabolism received the short reports and were not included in the study. Reports were written by two endocrinologists trained in bone metabolism. Clinical information was obtained from a database compiled by the bone density technologist during the procedure. The study was approved by the Human Studies Review Committee of UMass Memorial Health Care (Worcester, Massachusetts).

One of the authors interviewed the physicians by telephone by using an intake form after the physicians had received at least two reports in a given format. Physicians were asked whether they understood various concepts of bone densitometry; if they answered yes, they were asked to explain the concept. Specific questions about individual patients (Appendix) were then posed to each physician with the database available to the investigator and the office record available to the primary care physician. A minimum of two and a maximum of six reports were included in the analysis for each physician. After the interview, physicians who had received the short technical reports began to receive the long clinical reports and were reinterviewed after having received at least two long reports.

Numerical data were analyzed by using the t-test, the Mann-Whitney rank-sum test, one-way analysis of variance, or Kruskal-Wallis analysis of variance on ranks; Bonferroni correction was used where appropriate. Nominal data were analyzed with the Fisher exact test or the z-test, the McNemar test, or the chi-square test; Yates correction for continuity was used. Individual patient data were clustered by physician and averaged before analysis. Data were analyzed by using Sigma Stat (Jandel Scientific, San Rafael, California), and a P value less than 0.05 was considered significant in all analyses.

The funding source for this study had no role in gathering, analyzing, or interpreting the data or in the decision to submit the paper for publication.

Results

Top Methods Results Discussion Author & Article Info References

Sixty-eight physicians were enrolled over 7 months and were followed for an additional 10 months. Eleven physicians were omitted from the analysis because they did not remain in practice in the area or did not meet protocol requirements. Thirty-five physicians were randomly assigned to receive the short reports; they ordered 0.72 ± 0.71 tests per month before the first interview. Twenty-two physicians were randomly assigned to receive the long reports; they ordered 1.30 ± 1.21 tests per month before the first interview (P = 0.002 compared with physicians who received the short reports [95% CI for difference of means, 0.36 to 0.79]).

Thirty-three internists ordered 640 tests, 13 family practitioners ordered 136 tests, and 11 gynecologists ordered 118 tests. Patients referred by internists were older (63.0 ± 8.4 years compared with 56.7 ± 4.7 years; P = 0.044) and more likely to have had fragility fractures (19% compared with 1%; P = 0.044) than patients referred by gynecologists. The indications for testing among all physicians included risk factors (25%), patient concerns (23%), menopause (15%), history of fracture (12%), and steroid treatment (8%). The only difference seen by physician specialty was that more internists (17%) than gynecologists (1%) ordered tests because of history of fracture (P = 0.048).

At the first interview, 32% of physicians who received the long reports and 3% of those who received the short reports understood the concepts of T (SD from young adult reference mean) and Z (SD from age-matched controls) scores (Table 1) (P = 0.005). Eighty-six percent of physicians who received the long reports and 30% of those who received the short reports understood the role of bone mineral density in the definition of osteoporosis (P < 0.001). Forty-five percent of physicians who received the long reports and 12% of those who received the short reports understood how fracture risk related to bone mineral density (P = 0.014). No differences by physician specialty were seen.

We saw trends toward fewer formal referrals to specialists, more testing to rule out secondary causes of osteoporosis, and more changes in the pharmacologic treatment of osteoporosis in patients whose reports were in the long format compared with those whose reports were in the short format, but these differences were not statistically significant (Table 2). When specialty was analyzed, 61% of patients of gynecologists who received long reports and 19% of patients whose reports were in the short format received changes in the pharmacologic treatment of osteoporosis (P = 0.021).

Physicians found that 36% of the 112 short reports and 1% of the 176 long reports were confusing (P = 0.003). Of the 20 physicians who were interviewed after receiving the short reports and again after receiving the long reports, 100% requested that use of the long format be continued. Of the 42 physicians interviewed after receiving the long reports, 88% requested continued use of this format.

Discussion

Top Methods Results Discussion Author & Article Info References

Clinical reporting of bone mineral density test results to primary care physicians compared with brief technical reporting leads to increased use and understanding of bone densitometry and changes in the management of osteoporosis. The Consensus Statement of the International Panel on the Clinical Utility of Bone Mass Measurements includes the suggestion that bone mass data be accompanied by a clinical interpretation, including a narrative report, to allow better decision making by the primary care physician [6]. A survey of 181 primary care physicians in Tucson, Arizona, showed that bone mineral density, relative fracture risk, and the World Health Organization definition of osteoporosis were of primary value, whereas clinical recommendations were less important [10]. In contrast, our physicians preferred to receive reports with treatment and follow-up recommendations, perhaps because they actually received the reports and were able to appreciate their usefulness. The practice patterns of the physicians in Tucson and the relatively small group of physicians referring to our osteoporosis center may differ. Other limitations of our study include the potential bias of the interviewer, who was aware of report style assignment during the interview, and an imbalance between the numbers of physicians initially assigned to each group, which might have been avoided by monitoring and adjusting the randomization process.

Our results do not suggest that bone density reports should be standardized; rather, primary care physicians require a customized report depending on their knowledge base and their needs. For example, although we sent different reports to primary care physicians and specialists, three primary care physicians requested the short report instead of the long report and one specialist requested that clinical recommendations be attached to the short report. Reports generated at different institutions may vary according to local background and individual needs. The clinical format may also be useful for transmitting new guidelines or diagnostic and treatment strategies. For example, during the study, new data became available on the clinical utility of bone markers [11], and suggestions for the use of these markers were included in a few selected reports. Although one may surmise that physicians who order many bone density tests would eventually ask to receive the short report rather than the long report, this did not occur in our study, at least in part because some of the physicians found the long report to be a valuable addition to the medical record or a useful document to send to patients for their review. Most of the tests ordered were initial base-line studies. In the future, the long reports would also be helpful in discussing the significance of changes in bone mineral density.

Our results support the Consensus Statement of the International Panel on the Clinical Utility of Bone Mass Measurements with respect to primary care physicians: Bone mass measurement should be accompanied by a clinical interpretation. To be useful, clinical interpretation must rely on an accurate database of clinical information, and the interpretation should be written by a physician with expertise in bone metabolism. This approach may be useful in achieving appropriate use of bone densitometry and may provide a convenient tool for dissemination of new information and guidelines about osteoporosis testing and treatment. A major concern of primary care physicians in this era of rapidly evolving technology is their ability to interpret results of complex new diagnostic tests and translate these into busy clinical practice. Our results also suggest the potential value of development of clinical, user-friendly reporting of technical diagnostic results in other medical specialties.

Appendix

Technical Reports

The short technical reports were in a memorandum format and included the following information.

1. Density of lumbar spine and hip (given as g/cm²).

2. Percentage of and SD from young adult reference mean (T score).

3. Percentage of age-matched controls.

4. Comments about artifacts.

The long clinical reports were in a letter format and included the following information.

1. Data in the short technical report.

2. Diagnosis by World Health Organization criteria.

3. Clinical risk factors for fracture, including advanced age, history of previous fractures, family history of osteoporosis, and history of frequent falls.

4. Specific suggestions for nonpharmacologic and pharmacologic treatment and follow-up bone densitometry.

Questionnaire

The following questionnaire was administered to primary care physicians about patients referred for bone mineral density testing.

1. Why did you order a bone mineral density test for this patient?

2. Did you review the bone densitometer-generated computer printout attached to the report?

3. As a result of this report, have you referred the patient for further evaluation with a specialist physician?

4. As a result of this report, have you ordered any further tests for the evaluation of osteoporosis?

5. As a result of this report, have you suggested any changes in diet or calcium supplements? If so, what were they?

6. As a result of this report, have you suggested any change in exercise regimen or lifestyle changes? If so, what were they?

7. As a result of this report, have you prescribed any medication or made changes in medication? If so, what were they?

8. Did you find this report confusing? If so, why?

9. Did you agree with the history in this report obtained by our technologist? If not, why? (This question was asked of physicians who received the long clinical reports only.)

Presented in part at the Nineteenth Annual Meeting of the American Society for Bone and Mineral Research, Cincinnati, Ohio, September, 1997.