 |
|
Flexible Sigmoidoscopy Training and Its Impact on Colorectal Cancer Screening by Primary Care Physicians
James D. Lewis, MD, MSCE;
Gregory G. Ginsberg, MD;
Timothy C. Hoops, MD;
Michael L. Kochman, MD;
Warren B. Bilker, PhD;
Brian L. Strom, MD, MPH
Arch Fam Med. 2000;9:420-425.
ABSTRACT
| |
Background Colorectal cancer is an ideal disease for prevention with screening programs. Efforts to increase compliance with screening recommendations have included training primary care physicians to perform flexible sigmoidoscopy.
Objective To assess the impact of flexible sigmoidoscopy training on compliance with current screening recommendations.
Methods We performed a cross-sectional study of 232 patients cared for by physicians in a primary care network.
Main Outcome Measures Rates of screening for colorectal cancer and rates of undergoing flexible sigmoidoscopy were compared across patient groups according to the physician's training and whether the physician performs flexible sigmoidoscopy in his or her practice.
Results Among 217 patients included in the analysis, 122 (56%) were cared for by physicians who were trained in flexible sigmoidoscopy, of whom 79 (36%) were cared for by physicians who perform flexible sigmoidoscopy in their practice. Patients cared for by physicians trained in flexible sigmoidoscopy were not significantly more likely to receive any colorectal cancer screening than were patients cared for by physicians not trained in flexible sigmoidoscopy (odds ratio, 1.16; 95% confidence interval, 0.67-2.01). However, patients cared for by physicians who perform flexible sigmoidoscopy in their practice were more likely to have undergone any colorectal cancer screening (odds ratio, 1.73; 95% confidence interval, 1.02-2.95) and flexible sigmoidoscopy (odds ratio, 2.69; 95% confidence interval, 1.14-6.36).
Conclusion Performance of flexible sigmoidoscopy by primary care physicians has the potential to increase the rate of colorectal cancer screening with flexible sigmoidoscopy.
INTRODUCTION
COLORECTAL cancer is the second most common cause of cancer-related deaths in the United States.1 The disease has a high prevalence, passes through a long asymptomatic yet detectable phase, has higher cure rates when detected at an early stage, and has a high mortality rate when diagnosed at an advanced stage.2 As such, colorectal cancer is well suited for prevention with screening programs. Screening with flexible sigmoidoscopy and fecal occult blood testing have been demonstrated to reduce colorectal cancer–related mortality.3-8 Currently, most professional organizations recommend screening for colorectal cancer with flexible sigmoidoscopy every 5 years and/or annual fecal occult blood testing beginning at age 50 years.2, 9-10 Recently published guidelines2 also recognize a role for barium enema and colonoscopy as screening modalities.
Despite these strong recommendations, few patients actually undergo these examinations. Only 9.4% to 30.4% of Americans 50 years and older have had a sigmoidoscopic examination in the preceding 3 to 5 years.2-11 Similarly, in the 1997 Behavioral Risk Factor Surveillance System data,11 only 20.9% of women and 18.3% of men 50 years or older reported undergoing fecal occult blood testing in the preceding year.
Much research12-21 has focused on identifying barriers to completion of colorectal cancer screening and methods to increase compliance. In most cases, primary care physicians are responsible for ensuring that patients receive the recommended colorectal cancer screening. For multiple reasons, primary care physicians have been trained to perform flexible sigmoidoscopy, and results of recent studies22-24 suggest that 23% to 67% of primary care physicians routinely perform flexible sigmoidoscopy as part of their practice. Although it is logical that training primary care physicians to perform flexible sigmoidoscopy should increase the rate of colorectal cancer screening, there are few data to support this hypothesis.15, 21 In addition, previous studies generally do not provide information on each of the currently available screening modalities but focus specifically on completion of flexible sigmoidoscopy. The present study was designed to test the hypothesis that patients cared for by primary care physicians trained in flexible sigmoidoscopy are more likely to receive colorectal cancer screening than are patients whose physician has not received training.
PARTICIPANTS AND METHODS
PHYSICIAN DATA
Clinical Care Associates is a group of primary care physicians and specialists who are employed by the University of Pennsylvania Health System to maintain community-based medical practices. Physicians participating in this program were contacted via a mailed survey to determine whether they had received training in flexible sigmoidoscopy and whether they were currently performing the procedure as part of their practice. Physicians also rated on a 5-point Likert scale the strength of their belief that screening with flexible sigmoidoscopy every 5 years and with annual fecal occult blood testing reduces colorectal cancer mortality. These results are reported elsewhere.25
PATIENT SELECTION
We performed a cross-sectional study of patients cared for by physicians who returned the mailed questionnaire. Study patients were identified through a 3-stage sampling process using a University of Pennsylvania Health System database that includes patient demographic information, patients' care providers, and International Classification of Diseases, Ninth Revision–coded26 diagnostic information from billing records. Stage 1 of the sampling procedure identified all patients aged 50 to 75 years as of January 1, 1997, identified by the database as receiving care from a physician who had completed the initial survey.
Because some clinicians might choose to discontinue colorectal cancer screening for patients with limited life expectancy, we attempted to design a cohort with few comorbid conditions.27 In addition, we attempted to minimize the number of patients in the cohort who had been evaluated for gastrointestinal tract symptoms, thus increasing the likelihood of completing colorectal cancer screening procedures as part of an evaluation of symptoms rather than for screening. To accomplish this, in stage 2 of the sampling procedure we used the diagnostic coding information from the database to exclude patients with the following diagnoses: congestive heart failure, end-stage renal disease, chronic obstructive pulmonary disease, cancer other than nonmelanoma skin cancer, stroke, human immunodeficiency virus infection, Alzheimer disease, Parkinson disease, iron deficiency anemia, Crohn disease, ulcerative colitis, rectal bleeding, lower abdominal pain, and irritable bowel syndrome.
In stage 3 of the sampling procedure we selected a stratified random sample of 7 patients per physician. We then collected data from outpatient charts of the first 5 of the 7 patients for whom the outpatient chart was available. Often the computerized database did not identify the correct primary care physician. As such, some clinicians cared for more or less than 5 study patients (median, 3 patients per physician; interquartile range, 2-6 patients).
Patients were excluded from the analysis if they had received care from their primary care physician for less than 1 year (n=48), did not have an office visit after December 31, 1996 (n=5), or had a diagnosis of inflammatory bowel disease identified during chart abstraction (n=5). Patients with other comorbid conditions identified during chart abstraction were not excluded (Table 1).
|
|
|
|
Table 1. Patient Characteristics According to Primary Care Physician Category*
|
|
|
DEFINITION OF OUTCOMES
Patients were considered to have undergone colorectal cancer screening if they had completed any of the following tests: home fecal occult blood testing between January 1, 1995, and December 31, 1997; flexible sigmoidoscopy between January 1, 1992, and December 31, 1997; a barium enema between January 1, 1987, and December 31, 1997; or colonoscopy between January 1, 1987, and December 31, 1997. Patients were considered to have been offered these screening tests if the physician documented in the medical chart scheduling the test, completion of the test, or that the patient refused the test during the corresponding period. We also recorded whether patients were offered and underwent a rectal examination, serum cholesterol measurement, and a mammogram in women during calendar year 1997 and prostate-specific antigen testing in men between January 1, 1995, and December 31, 1997.
STATISTICAL ANALYSES
Comparisons of continuous and dichotomous variables between multiple groups were performed using the Kruskal-Wallis test and the Pearson  2 test, respectively.28 To test the association of physicians' training and practice characteristics with rate of colorectal cancer screening we compared the proportion of patients receiving screening according to whether their physician was trained to perform flexible sigmoidoscopy and whether the physician performed the procedure in his or her practice. These comparisons were performed using logistic regression analysis with a robust estimate of the variance to adjust for clustering on the primary care physician and to adjust for other potential confounding variables.28 All analyses used 2-sided tests of statistical significance, with significance at P=.05. Statistical analyses were performed using computer software (STATA 5.0; Stata Corp, College Station, Tex).
RESULTS
Of 290 patients whose medical records were abstracted, 232 met the inclusion criteria. These patients were cared for by 68 physicians, of whom 38 (56%) were trained to perform flexible sigmoidoscopy and 23 (34%) were currently performing flexible sigmoidoscopy in their practice. Of 30 physicians (44%) who had not received training in flexible sigmoidoscopy, 12 (18%) reported interest in obtaining training. Table 1 details demographic characteristics of the patients stratified by the characteristics of the physician caring for the patient. There were no significant differences in age, sex, or duration of care among the 4 patients groups.
Table 2 contains data on the proportion of patients in each group completing screening tests. In unadjusted analyses, the overall rate of colorectal cancer screening did not differ significantly between patients cared for by physicians trained (48%) vs not trained (46%) in flexible sigmoidoscopy (odds ratio[OR], 1.09; 95% confidence interval [CI], 0.63-1.90). In bivariate analyses, there was evidence that the association between physician training and completion of colorectal cancer screening may differ according to whether the physician believes in the effectiveness of flexible sigmoidoscopic examination and fecal occult blood testing to reduce cancer-related mortality (ie, possible effect modification). Physicians who did not believe in the ability of either of these modalities to reduce colorectal cancer mortality cared for only 15 patients, of whom 4 were cared for by physicians who had received training in flexible sigmoidoscopy. None of these 4 patients had completed any form of colorectal cancer screening. As such, all further analyses were limited to the 217 patients whose physicians believe that screening for colorectal cancer reduces mortality.
|
|
|
|
Table 2. Patients Receiving Screening According to Primary Care Physician Catagory*
|
|
|
In analyses limited to the remaining 217 patients, the overall rate of colorectal cancer screening did not differ significantly between patients cared for by physicians trained in vs not trained in flexible sigmoidoscopy (OR, 1.16; 95% CI, 0.67-2.01). A logistic regression model adjusting for patient age, patient sex, presence of comorbid illness, and duration of care and for physician age and sex demonstrated no evidence of further confounding (fully adjusted OR, 1.03; 95% CI, 0.57-1.87). However, in the fully adjusted model, there was a strong association between patient sex and receiving colorectal cancer screening (OR for men, 2.54; 95% CI, 1.31-4.95).
There was a higher rate of screening by any method among patients cared for by physicians who perform flexible sigmoidoscopy in their practice (57%) than among patients whose physician does not perform flexible sigmoidoscopy (44%) (OR, 1.73; 95% CI, 1.02-2.95). Adjusting individually for patient age, patient sex, presence of comorbid illness, and duration of care and for physician age and sex did not significantly affect this estimate.
There was a higher rate of screening with flexible sigmoidoscopy among patients whose physician was trained (17%) vs not trained (8%) to perform flexible sigmoidoscopy, although this difference was not statistically significant (OR, 2.26; 95% CI, 0.78-6.57). Similarly, patients whose physicians perform flexible sigmoidoscopy were significantly more likely to receive a flexible sigmoidoscopic examination (21%) than were patients whose physician does not perform flexible sigmoidoscopy (9%) (OR, 2.69; 95% CI, 1.14-6.36). Adjusting individually for patient age, patient sex, presence of comorbid illness, and duration of care and for physician age and sex did not significantly affect either estimate.
There were no significant differences in the proportion of patients receiving fecal occult blood testing, cholesterol screening, prostate cancer screening with serum prostate-specific antigen testing, or breast cancer screening with mammogram when stratified according to (1) the physician's training in flexible sigmoidoscopy or (2) whether the physician performs flexible sigmoidoscopy in his or her practice (Table 3).
|
|
|
|
Table 3. Comparison of Compliance With Screening Recommendations According to Primary Care Physician Training and Practice Characteristics*
|
|
|
COMMENT
Our data demonstrate that the overall rate of colorectal cancer screening does not differ between patients cared for by physicians who are or are not trained to perform flexible sigmoidoscopy. However, there was a trend toward increased screening with flexible sigmoidoscopy among patients cared for by physicians trained to perform this procedure. Patients cared for by physicians choosing to perform flexible sigmoidoscopy as part of their practice were more likely to receive any colorectal cancer screening and specifically to undergo flexible sigmoidoscopy. These data suggest that training primary care physicians to perform flexible sigmoidoscopy may not result in an overall increase in the proportion of the population receiving colorectal cancer screening. However, such a program is likely to increase the proportion of patients receiving screening with flexible sigmoidoscopy. Furthermore, creating an environment in which physicians trained to perform flexible sigmoidoscopy choose to perform this procedure in their practice might increase the overall colorectal cancer screening rate.
There are several possible explanations for the finding that the overall screening rate did not differ according to physicians' training. Our definition of colorectal cancer screening included several screening tests that do not require special training. In addition, physicians not trained to perform flexible sigmoidoscopy may view screening with fecal occult blood testing to be equally or more effective in reducing colorectal cancer mortality than screening with flexible sigmoidoscopy. As such, these physicians may be more likely to screen patients with fecal occult blood testing alone.13 Our finding that the rate of screening with fecal occult blood testing did not differ according to whether the patient's physician was trained in flexible sigmoidoscopy is consistent with this hypothesis.
The increased rate of screening with flexible sigmoidoscopy by physicians who perform flexible sigmoidoscopy might also be explained by several factors. Results of previous studies15-16 suggest that patients' willingness to undergo screening flexible sigmoidoscopy is influenced by the emphasis that their primary care physician places on this screening test. Physicians who perform flexible sigmoidoscopy may be better prepared to explain the procedure to their patients. In addition, providing this service in the primary care office may be more convenient for patients. Finally, it is also possible that those physicians currently performing flexible sigmoidoscopy are more conscious of the need for this procedure and thus recommend it more often.21
Several limitations to this study need to be recognized. The patients and physicians included in the study are all employed by a university-based health system in a large urban area. As such, the patient profile and physician characteristics might differ slightly from other populations of physicians and patients. We believe that this is unlikely to represent a substantial bias because the proportion of patients in our cohort who had completed fecal occult blood testing and flexible sigmoidoscopy was comparable to that of previous studies.2, 11 In addition, the clinicians in our study all participated in community-based practices.
This study is potentially subject to selection bias for additional reasons. Physicians who choose to be trained in flexible sigmoidoscopy and who choose to perform this procedure in their practice might differ from other physicians. Physicians who believe strongly in the role of preventive medicine could be more likely to seek training in flexible sigmoidoscopy. This group would then be expected to emphasize more strongly the need for colorectal cancer screening and particularly the need for flexible sigmoidoscopy. However, the absence of a difference between the groups in the utilization of cholesterol screening, mammography, and prostate-specific antigen testing suggests that these physicians do not perform other screening tests more frequently. Furthermore, we excluded from the analyses all patients cared for by physicians reporting that they do not believe in the effectiveness of colorectal cancer screening. As such, this form of selection bias is unlikely to significantly affect the results of this study.
It is possible that physicians who perform flexible sigmoidoscopy differ in other ways. Lewis et al25 previously demonstrated that physicians choosing not to perform flexible sigmoidoscopy in their practice emphasize that the time required to perform the procedure and the lack of availability of adequately trained support staff are barriers to performing flexible sigmoidoscopy. These physicians also emphasize the availability of other clinicians to perform flexible sigmoidoscopy.25 Lewis and Asch29 also previously demonstrated that the current reimbursement rates for flexible sigmoidoscopy may be inadequate to cover the primary care physician's costs. Furthermore, some managed care plans do not provide any additional reimbursement to primary care physicians for performing flexible sigmoidoscopy. As such, it is possible that unmeasured practice characteristics, including the insurance status of the patient population, the proportion of patients participating in managed care plans, and the size of the physician's patient panel, may have affected the results. However, these factors would only further emphasize the need to design health care systems that encourage primary care physicians to perform flexible sigmoidoscopy.
We do not have data on patients whose physicians did not complete the questionnaire. Nonresponders may differ from responders, particularly regarding interest in colorectal cancer screening and belief in the effectiveness of colorectal cancer screening. We limited our analyses to physicians who reported that they believed in the effectiveness of this screening. Given the growing evidence of the effectiveness of colorectal cancer screening, the proportion of physicians not believing in the effectiveness of this intervention is likely to be small, and that is indeed what we observed.
Another potential limitation to our study is that we were only able to analyze data recorded in the primary care record. We assumed that tests not documented in the primary care record were not completed. As such, we may have underestimated the performance rate of flexible sigmoidoscopy, particularly for patients having the examination performed by clinicians other than their primary care physician. This might partially explain the observed difference in the flexible sigmoidoscopy rates between the groups and would further emphasize that programs designed merely at training physicians to perform flexible sigmoidoscopy will not be adequate to increase screening rates.
Finally, because this was a cross-sectional study, we do not know how long the physicians have been performing flexible sigmoidoscopy in their practice. Similarly, we do not know whether any physicians who reported training in flexible sigmoidoscopy but currently choosing not to perform the procedure previously performed this procedure as part of their practice. If some physicians recently began or discontinued performing flexible sigmoidoscopy in their practice, our study may underestimate the true impact of such training on compliance with screening recommendations.
Despite these limitations, the results of this study demonstrate that training primary care physicians to perform flexible sigmoidoscopy has the potential to increase the rate of colorectal cancer screening with sigmoidoscopy. Creating an environment whereby physicians trained in flexible sigmoidoscopy will choose to perform the procedure in their practice would be expected to further increase the rate of screening with flexible sigmoidoscopy, and potentially the overall colorectal cancer screening rate. Further research should examine the impact of training primary care physicians to perform flexible sigmoidoscopy in other geographic locations and health care settings. In addition, it will be important to establish factors that affect whether a physician chooses to perform flexible sigmoidoscopy after receiving training.
AUTHOR INFORMATION
Accepted for publication December 15, 1999.
This study was funded by a grant from the Office of the Associate Dean for Health Services Research of the University of Pennsylvania, Philadelphia; in part by training grant 1-T32-DK07740-0 from the National Institutes of Health, Bethesda, Md (Dr Lewis); and by an Outcomes Research Training award from the American Digestive Health Foundation.
Corresponding author: James D. Lewis, MD, MSCE, Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, 423 Guardian Dr, Eighth Floor, Blockley Hall, Philadelphia, PA 19104 (e-mail: lewis{at}cceb.med.upenn.edu).
From the Division of Gastroenterology (Drs Lewis, Ginsberg, Hoops, and Kochman), the Center for Clinical Epidemiology and Biostatistics (Drs Lewis, Bilker, and Strom), the Division of General Internal Medicine (Dr Strom), and the Department of Biostatistics and Epidemiology (Drs Bilker and Strom), University of Pennsylvania, Philadelphia.
REFERENCES
| |
1. Ries L, ed, Kosary C, ed, Hankey B, ed, Miller B, ed, Edwards B, ed. SEER Cancer Statistics Review 1973-1995. Bethesda, Md: National Cancer Institute; 1998.
2. Winawer ST, Fletcher RH, Miller L, et al. Colorectal cancer screening: clinical guidelines and rationale. Gastroenterology. 1997;112:594-642.
FULL TEXT
|
ISI
| PUBMED
3. Newcomb P, Norfleet R, Sorer B, Surawiscz T, Marcus P. Screening sigmoidoscopy and colorectal cancer mortality. J Natl Cancer Inst. 1992;84:1572-1575.
FREE FULL TEXT
4. Mueller A, Sonnenberg A. Prevention of colorectal cancer by flexible sigmoidoscopy and polypectomy: a case-control study of 32702 veterans. Ann Intern Med. 1995;123:904-910.
FREE FULL TEXT
5. Selby JV, Friedman GD, Quesenberry CP Jr, Weiss NS. A case-control study of screening sigmoidoscopy and mortality from colorectal cancer. N Engl J Med. 1992;326:653-657.
ABSTRACT
6. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood: Minnesota Colon Cancer Control Study. N Engl J Med. 1993;328:1365-1371.
FREE FULL TEXT
7. Kronborg O, Fenger C, Olsen J, Jorgensen OD, Sondergaard O. Randomised study of screening for colorectal cancer with faecal-occult-blood test. Lancet. 1996;348:1467-1471.
FULL TEXT
|
ISI
| PUBMED
8. Hardcastle JD, Chamberlain JO, Robinson MH, et al. Randomised controlled trial of faecal-occult-blood screening for colorectal cancer. Lancet. 1996;348:1472-1477.
FULL TEXT
|
ISI
| PUBMED
9. Byers T, Levin B, Rothenberger D, Dodd GD, Smith RA. American Cancer Society guidelines for screening and surveillance for early detection of colorectal polyps and cancer: update 1997: American Cancer Society Detection and Treatment Advisory Group on Colorectal Cancer. CA Cancer J Clin. 1997;47:154-160.
ABSTRACT
10. US Preventative Services Task Force. Guide to Clinical Preventive Services. 2nd ed. Baltimore, Md: Williams & Wilkins; 1996.
11. Screening for colorectal cancer—United States 1997 MMWR Morb Mortal Wkly Rep. 1999;48:116-121.
PUBMED
12. Cockburn J, Thomas RJ, McLaughlin SJ, Reading D. Acceptance of screening for colorectal cancer by flexible sigmoidoscopy. J Med Screen. 1995;2:79-83.
PUBMED
13. Cooper G, Fortinsky R, Hapke R, Landefeld S. Primary care physicians' recommendations for colorectal cancer screening: patient and practitioner factors. Arch Intern Med. 1997;157:1946-1950.
FREE FULL TEXT
14. Faivre J, Arveux P, Milan C, Durand G, Lamour J, Bedenne L. Participation in mass screening for colorectal cancer: results of screening and rescreening from the Burgundy Study. Eur J Cancer Prev. 1991;1:49-55.
15. Holt WS Jr. Factors affecting compliance with screening sigmoidoscopy. J Fam Pract. 1991;32:585-589.
ISI
| PUBMED
16. Lewis S, Jensen N. Screening sigmoidoscopy: factors associated with utilization. J Gen Intern Med. 1996;11:542-544.
ISI
| PUBMED
17. Lindholm E, Berglund B, Haglind E, Kewenter J. Factors associated with participation in screening for colorectal cancer with faecal occult blood testing. Scand J Gastroenterol. 1995;30:171-176.
ISI
| PUBMED
18. McCarthy B, Moskowitz M. Screening flexible sigmoidoscopy: patient attitudes and compliance. J Gen Intern Med. 1993;8:120-125.
ISI
| PUBMED
19. Petravage J, Swedberg J. Patient response to sigmoidoscopy recommendations via mailed reminders. J Fam Pract. 1988;27:387-389.
ISI
| PUBMED
20. Pye G, Christie M, Chamberlain JO, Moss SM, Hardcastle JD. A comparison of methods for increasing compliance within a general practitioner–based screening project for colorectal cancer and the effect on practitioner workload. J Epidemiol Community Health. 1988;42:66-71.
FREE FULL TEXT
21. Schoen RE, Weissfeld JL, Kuller LH. Sigmoidoscopy use among primary care physicians. Prev Med. 1995;24:249-254.
FULL TEXT
|
ISI
| PUBMED
22. Saad JA, Pirie P, Sprafka JM. Relationship between flexible sigmoidoscopy training during residency and subsequent sigmoidoscopy performance in practice. Fam Med. 1994;26:250-253.
PUBMED
23. Buckley RL, Smith MU, Katner HP. Use of rigid and flexible sigmoidoscopy by family physicians in the United States. J Fam Pract. 1988;27:197-200.
ISI
| PUBMED
24. American Cancer Society. 1989 Survey of physicians' attitudes and practices in early cancer detection. CA Cancer J Clin. 1990;40:77-101.
ISI
| PUBMED
25. Lewis J, Asch D, Ginsberg G, et al. Primary care physicians' decisions to perform flexible sigmoidoscopy. J Gen Intern Med. 1999;14:297-302.
FULL TEXT
|
ISI
| PUBMED
26. World Health Organization. International Classification of Diseases, Ninth Revision (ICD-9). Geneva, Switzerland: World Health Organization; 1977.
27. Kiefe C, Funkhouser E, Fouad M, May D. Chronic disease as a barrier to breast and cervical cancer screening. J Gen Intern Med. 1998;13:357-365.
FULL TEXT
|
ISI
| PUBMED
28. Armitage P, Berry G. Statistical Methods in Medical Research. 2nd ed. Oxford, England: Blackwell Scientific Publications; 1987.
29. Lewis JD, Asch DA. Barriers to office-based screening sigmoidoscopy: does reimbursement cover costs? Ann Intern Med. 1999;130:525-530.
FREE FULL TEXT
RELATED ARTICLE
The Archives of Family Medicine Continuing Medical Education Program
Arch Fam Med. 2000;9(5):463-465.
FULL TEXT
THIS ARTICLE HAS BEEN CITED BY OTHER ARTICLES
Detection of Proximal Adenomatous Polyps With Screening Sigmoidoscopy: A Systematic Review and Meta-analysis of Screening Colonoscopy
Lewis et al.
Arch Intern Med 2003;163:413-420.
ABSTRACT
| FULL TEXT
|
