Objective: To assess whether the use of chest radiography for screening changes over time.
Design: Systematic review.
Data Sources: MEDLINE, ISI, Cochrane Central Register of Controlled Trials, and handsearching of selected journals.
Review Methods: We evaluated whether the proportion of primary care physicians using chest radiography to screen for (1) malignancy in the general asymptomatic population, (2) malignancy in a high-risk subgroup, (3) any disease in the general population, and (4) any disease in a high-risk subgroup changed over time, using random-effects meta-regression analysis. Adjustments for the availability of national guidelines were also performed.
Results: Overall, 10% to 90% of primary care physicians reported using chest x-ray for screening. In unadjusted analyses, the proportion of physicians using chest radiography for cancer screening in the general population tended to increase by 0.9% per year (8 studies, n = 4313). The corresponding annual changes were -2.9% for cancer screening in high-risk subgroups (8 studies, n = 2784) and -0.4% regarding screening for any disease in the population (7 studies, n = 2627). No meta-regressions were run for outcome (4) (only 1 study). In the adjusted analyses, there was a decreasing nonsignificant trend for all outcomes.
Conclusions: Despite formal recommendations, many physicians still use chest x-ray for screening, with their number decreasing slowly over time. This practice may be harmful because the positive predictive value of chest radiography is low, and further evaluation of false-positive findings might be associated with increased cost and risk from additional diagnostic or therapeutic interventions.
(Am J Manag Care. 2006;12:650-656)
Public health authorities worldwide discourage the use of chest radiography as a screening modality in primary care.1-5 The diagnostic performance of chest radiography does not justify its application for screening in the general population or among high-risk groups such as smokers or persons with a family history of lung cancer.6,7 Prescription of chest radiography as a screening tool may be harmful7 because patients with false-positive results may experience anxiety and concern.6 Moreover, if they proceed to further evaluation, they are subjected to associated increased cost and risk.6
Despite the accumulated evidence and clear guidelines, findings from various reports8-15 suggest that primary care physicians throughout the world still prescribe chest radiography for screening in the general population and among selected high-risk subgroups. Therefore, we set out to assess in a systematic review of the relevant literature whether this habit of old has been diminishing over time and at what pace.
Identification of Eligible Studies
We searched MEDLINE, ISI, and the Cochrane Central Register of Controlled Trials (last search, January 2005) using combinations of terms such as screening, chest radiography, cancer, baseline test, reference test, primary care, and prescription. We set no language or geographical restrictions. We also handsearched16 the last 4 years of 5 primary care and public health journals (European Journal of General Practice, Journal of Family Practice, American Family Physician, Annals of Family Medicine, and Family Practice) and perused the references of the retrieved relevant articles to identify reports that may have been missed by the electronic searches.
We considered as eligible all cross-sectional surveys or controlled trials providing information on the proportion of primary care physicians who reported using or who actually prescribed chest radiography as a screening tool in asymptomatic individuals (from the general population or among high-risk population subgroups). We evaluated all relevant studies, regardless of whether the corresponding proportion was a primary outcome or not. We excluded all qualitative research reports because their sampling methods and stopping rules do not ensure a representative sample and because the thematic coding of the main findings is formulated post hoc by the researchers. Physicians with specialties that are usually encountered in nonprimary care settings were excluded from the calculations unless it was clearly stated that they were indeed primary care oriented. We did not consider information pertaining to beliefs or personal views of physicians regarding the role of chest radiography, as these may be different from actual practice. Finally, we excluded all studies in which chest radiography was used in the diagnostic workup of persons with signs and symptoms of disease or in the follow-up of patients with pulmonary or systematic diseases. Reports from countries in which prescription of screening chest radiography was a part of any sort of national screening policy or trial were excluded from the analysis.
Definitions and Outcomes
We specified that smokers comprise a high-risk subgroup for lung cancer or nonmalignant pulmonary diseases (eg, chronic obstructive pulmonary disease). We also considered persons at high risk who were characterized as such in the primary reports (because of family history, high-risk profession, or any other reason). Screening for any disease pertained to the use of chest radiography in the diagnosis of any clinical entity; screening for malignancy pertained mostly to lung cancer screening. We evaluated the proportion of primary care physicians who reported using or who actually used chest radiography to screen for (1) malignancy in the general asymptomatic population, (2) malignancy in a high-risk subgroup, (3) any disease in the general asymptomatic population, and (4) any disease in a high-risk subpopulation changed over time, using random-effects meta-regression analysis.
We extracted information from each eligible study. The data recorded included the first author's name, journal and year of publication, place and country of origin, study design, physician inclusion and exclusion criteria, number of enrolled and analyzed physicians, number of physicians who reported using or actually used chest radiography for screening in each setting, method used to measure study outcomes (standard or telephone interview, questionnaire, patient medical record review, or the use of actors paying unannounced visits to the physician), population that would be screened (general population or high-risk subgroup) and for what purpose (screening for any disease or screening for cancer), and definition of high-risk population (if applicable).
In controlled studies that compared the prescription rates between a group of physicians who received educational interventions and a control group not exposed to the educational program, only physicians in the control group were considered eligible. Similarly, in interventional studies in which screening attitudes were evaluated before and after an educational intervention, we considered only baseline data (before the educational intervention). Finally, whenever prescription practice was evaluated by means of different methods in the same study, we recorded the most objective one. For example, we preferred information inferred from patients' medical record surveys over information based on patients' self-reports, and the latter was preferred over physicians' self-reported practices.
The study was the unit of the analyses. For each of the aforementioned outcomes, we calculated whether the proportion of physicians who use chest radiography as a screening tool changed over time, using random-effects meta-regression analyses.17 Meta-regressions are variance-weighted least-square regressions, in which the within-study and between-study variability of the pertinent proportions are accounted for. We used the year of study publication to approximate when each study was conducted.
Because guideline availability and knowledge of recommendations have been enhanced by the introduction of Internet and Web sites, we split the data for a subanalysis before and after 1990. We assessed whether there were any changes over time by comparing with a standardized z score18 the summary proportions of physicians prescribing chest radiographs for screening in studies published before and after 1990. Summary proportions were estimated using general inverse variance random-effects models,19 which allow for betweenstudy heterogeneity (dissimilarity) and incorporate it into the calculations.20 Heterogeneity in each subgroup was assessed using Fisher exact test and was quantified using I2, a metric that expresses the percentage of between-study variability that is attributed to heterogeneity rather than to chance. I2 ranges between 0% and 100%, and values above 75% typically represent high heterogeneity. Because we were interested in the temporal trends of the percentage of physicians who prescribe chest x-rays for screening, we did not consider further subgroup analyses. Instead, we performed additional meta-regressions that included as a covariate whether national guidelines existed at the time the study was completed or published (yes or no).
Analyses were performed in Intercooled STATA 8.2 (StataCorp LP, College Station, Tex) using the metareg module and in Meta-Analyst (Joseph Lau, Boston, Mass). Unless otherwise specified, all P values are 2-tailed, and P < .05 indicates formal statistical significance.
We identified 28 articles8-15,21-40 published between 1970 and 2005 describing 23 potentially eligible studies (5 publications25,29,30,38,39 were excluded because they were covered by 3 other reports of the same studies). Overall, the proportions of physicians who prescribe chest radiography for screening were described in 17 cross-sectional surveys and in 1 randomized trial11 (Table 1). Sample sizes in the studies using surveys ranged from 120 to 1349 analyzed primary care physicians. In all studies, the pertinent information was self-reported, except for the randomized trial, in which a more objective method was used (unannounced visits from standard patients). Three studies were conducted before the announcement of corresponding national guidelines (an American study23 published in 1970) or in countries in which relevant national guidelines do not exist (studies from Greece33 and France40). The percentage of physicians who reported using chest radiography for screening was more than 75% in all these cases. Most studies did not clearly state the exact phrasing of the questions used in the various studies to assess how commonly physicians prescribe chest x-ray for screening and in which setting. Similarly, most studies did not evaluate physician beliefs; therefore, discrepancy in physician beliefs and attitudes could not be analyzed.
Overall, 10% to 90% of primary care physicians actually used or reported that they used chest radiography for screening purposes in the various studies. When all studies were considered, the percentage of physicians who use chest x-ray to screen for malignancy in the general population8,9,14,15,22,27,28,40 seemed to increase by 0.9% per year, although this was not formally statistically significant (P = .55; 8 studies with 4313 analyzed participants) (Table 2). The increasing trend is due to leverage from the French study,40 in which 85% of 480 physicians reported that they use biannual chest x-ray to screen for cancer among patients older than 50 years. This trend was reversed when the availability of national guidelines was taken into account in the calculations (annual decrease, -1.4%; P = .14) (Figure 1A). When screening for cancer in high-risk population subgroups,8-10,14,24,26,28,40 we estimated a -2.9% annual decrease in the corresponding percentage (P = .10; 8 studies with 2784 analyzed participants). The trend became stronger, albeit only marginally and nonsignificantly, when the availability of national guidelines was taken into account (P = .05) (Figure 1B). Similarly, Figure 1C shows a small decreasing tendency of -0.8% per year in the proportion of primary care physicians who use chest radiography to screen for any disease in the general population (P = .41; 7 studies with 2627 analyzed physicians).12,23,31,33-35 In the latter case, the trend disappeared (annual decrease, -0.4%; P = .40) when the presence of national guidelines was taken into account (Table 2). Because only 1 study11 provided data on the use of chest radiography to screen for any clinical entity in a high-risk population, no meta-regressions were run for this outcome. In that study, 15 (27.3%) of 55 primary care physicians reported using chest x-ray for this purpose.
There was substantial between-study heterogeneity in the proportion of physicians prescribing chest radiography for screening (P < .05 for the subgroups of studies published before or after 1990 in all 3 outcomes in which the analysis was meaningful; I2 was universally above 90% in all these analyses). When we compared the corresponding summary proportions before and after 1990 (Figure 2), a diminishing trend was again evident. The decrease was significant only for chest x-ray prescription for cancer screening among high-risk population subgroups (P = .007).
Chest radiography prescription rates were lower in Israel, Australia, and North American countries (United States and Canada). They were higher in Europe (France and Greece).
Despite the available evidence and recommendations, physicians commonly prescribed chest radiography for screening purposes in the general asymptomatic population and among high-risk population subgroups. There are indications that this habit of old diminishes at a slow pace over time.
Screening tests are generally harmful, and only in selected cases does their application outweigh potential harms.41,42 In 2 systematic reviews6,7 of older randomized trials, there was no evidence supporting the use of chest radiography for lung cancer screening. If anything, screening with chest radiography was associated with increased lung cancer mortality,7 although this finding is consistent with overdiagnosis bias, given that overall mortality was not affected. Health hazards are unrelated to radiation exposure, because the delivered dose is low.43 They rather stem from the additional diagnostic or therapeutic interventions during the further evaluation of false-positive findings.6 The proportion of false-positive screening chest radiograms suspicious for lung cancer may be high (range, 40%-60%).7 Therefore, the implementation of even an inexpensive imaging study on a massive scale would greatly burden health economics and the workload of radiology departments. Most patients in the United States and Canada are not screened for lung cancer using chest xrays. However, because recommendations are based on older randomized trials (and the comparisons were against screening less often rather than no screening), the question is revisited. The ongoing Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial44 is investigating approximately 100 000 people aged 55 to 74 years to clarify the value of annual screening with chest x-ray compared with usual care.
There is evidence that chest radiography was prescribed for screening purposes in all identified relevant studies13,21,32,36,37,45,46 (including those that were ineligible for the analyses). Time seemed to change prescription patterns (although statistically nonsignificantly), but not fast enough to obliterate these habits during more than 20 years of follow-up. There may be many reasons for this slow drop. High rates of chest radiography prescription may be explained by economic and professional factors (especially in the context of fee-for-service medical practice)21,45 or by the absence of national guidelines. Judging from the Greek33 and French40 studies, it seems that the recommendations of the European Code Against Cancer1 do not have any effect on prescription practices in the absence of specific national guidelines. The role of the language barrier in such cases is debatable. Ignorance of formal recommendations on the issue might be an explanation, especially in countries without a strong tradition in primary care medicine. Considering the dearth of high-quality studies, strong suspicion arises that the frequency of screening chest radiography in primary care is underestimated, especially in countries without an established tradition in primary care33 and in the absence of national guidelines.23,33,40
There are some study limitations that should be discussed. First, despite the fact that screening chest radiography was being studied from the 1960s onward, only 18 studies were eligible, and all but 4 reports8,12,33,40 were from the United States and Canada. Therefore, these findings may not be globally generalizable. Second, data were derived from cross-sectional studies. This design has a limited internal validity and is sensitive to several biases. Furthermore, it is not uncommon for surveys to have low response rates.47 Overall, 7876 (57.5%) of 13 690 enrolled physicians were analyzed in the eligible studies. This raises further concerns regarding the generalizability of the results, because nonresponders may have systematically different characteristics from those of responders. Nevertheless, cross-sectional surveys are commonly used47 and are considered appropriate and easy to perform. Third, the exact question that was used to assess the outcome of interest was not clearly described in most studies. Unfortunately, the effect of different phrasings on our findings cannot be assessed.
Allowing for the aforementioned caveats, we conclude that chest radiography prescription for screening purposes among asymptomatic subjects is an old habit that dies hard. This ongoing practice may harm the screenees' health and burdens health economics and radiology department activities. Continuing medical education programs and the establishment of relevant national guidelines, where lacking, may reduce the frequency of recommendation of chest radiography for screening purposes.
We acknowledge Calderisi Marina, Rossella Ballarini, and Antonio Florita of the National Cancer Institute of Milan for their invaluable help in the international cooperation of the Panhellenic Association for Continual Medical Research.
From the Section of Oncology and Public Health, Panhellenic Association for Continual Medical Research, Athens (DM, KK, AP, AX, CP, MD, GZ, GK), and Department of Medical Oncology, Ioannina University Hospital, Ioannina (DM, GP, NP), Greece; and Department of Medical Oncology, Multimedica Hospital, Milan, Italy (MD).
Address correspondence to: Davide Mauri, MD, Section of Oncology and Public Health, Panhellenic Association for Continual Medical Research, Kleanthous 47 (TOYMPA), TK 54453, Thess/ki, Greece. E-mail: email@example.com.
1. Boyle P, Autier P, Bartelink H, et al. European Code Against Cancer and scientific justification: third version (2003). Ann Oncol. 2003;14:973-1005.
2. Smith RA, Mettlin CJ, Davis KJ, Eyre H. American Cancer Society guidelines for the early detection of cancer. CA Cancer J Clin. 2000;50:34-49.
3. National Cancer Institute. Screening for lung cancer (PDQ). National Cancer Institute CancerNet Web site. 2001. Available at: http://www.cancer.gov/cancerinfo/pdq/screening/lung/healthprofessional/. Accessed May 14, 2005.
4. Morrison BJ. Interventions other than smoking cessation to prevent lung cancer. Update for the Canadian Task Force on Preventive Health Care. http://www.ctfphc.org/Abstracts_printable/Ch64abs.htm. Accessed May 14, 2005.
5. US Preventive Services Task Force. Lung cancer screening: recommendation statement. Ann Intern Med. 2004;140:738-739.
6. Humphrey LL, Teutsch S, Johnson M; U.S. Preventive Services Task Force. Lung cancer screening with sputum cytologic examination, chest radiography, and computed tomography: an update for the U.S. Preventive Services Task Force. Ann Intern Med. 2004;140:740-753.
7. Manser RL, Irving LB, Stone C, Byrnes G, Abramson M, Campbell D. Screening for lung cancer. Cochrane Database Syst Rev. 2004;(1):CD001991.
8. Sladden MJ, Ward JE. Do Australian family physicians screen smokers for lung cancer? Chest. 1999;115:725-728.
9. Ashford A, Gemson D, Sheinfeld Gorin SN, et al. Cancer screening and prevention practices of inner-city physicians. Am J Prev Med. 2000;19:59-62.
10. Smith HE, Herbert CP. Preventive practice among primary care physicians in British Columbia: relation to recommendations of the Canadian Task Force on the Periodic Health Examination. CMAJ. 1993;149:1795-1800.
11. Beaulieu MD, Rivard M, Hudon E, Beaudoin C, Saucier D, Remondin M. Comparative trial of a short workshop designed to enhance appropriate use of screening tests by family physicians. CMAJ. 2002;167:1241-1246.
12. Nakar S, Vinker S, Neuman S, Kitai E, Yaphe J. Baseline tests or screening: what tests do family physicians order routinely on their healthy patients? J Med Screen. 2002;9:133-134.
13. Lynch GR, Prout MN. Screening for cancer by residents in an internal medicine program. J Med Educ. 1986;61:387-393.
14. Montano DE, Phillips WR. Cancer screening by primary care physicians: a comparison of rates obtained from physician self-report, patient survey, and chart audit. Am J Public Health. 1995;85:795-800.
15. American Cancer Society 1989 Survey of physicians' attitudes and practices in early cancer detection. CA Cancer J Clin. 1990;40:77-101.
16. Hopewell S, Clarke M, Lusher A, Lefebvre C, Westby M. A comparison of handsearching versus MEDLINE searching to identify reports of randomized controlled trials. Stat Med. 2002;21:1625-1634.
17. Higgins JPT, Thompson SG. Controlling the risk of spurious findings from metaregression. Stat Med. 2004;23:1663-1682.
18. Cappelleri JC, Ioannidis JP, Schmid CH, et al. Large trials vs meta-analysis of smaller trials: how do their results compare? JAMA. 1996;276:1332-1338.
19. Cooper H, Hedges LV, eds. The Handbook of Research Synthesis. New York, NY: Russel Sage Foundation; 1994.
20. Lau J, Ioannidis JP, Schmid CH. Quantitative synthesis in systematic reviews. Ann Intern Med. 1997;127:820-826.
21. Abelson J, Lomas J. Do health service organizations and community health centres have higher disease prevention and health promotion levels than fee-for-service practices? CMAJ. 1990;142:575-581.
22. American Cancer Society Survey of physicians' attitudes and practices in early cancer detection. CA Cancer J Clin. 1985;35:197-213.
23. Bates B, Mulinare J. Physicians' use and opinions of screening tests in ambulatory practice. JAMA. 1970;214:2173-2180.
24. Battista RN. Adult cancer prevention in primary care: patterns of practice in Quebec. Am J Public Health. 1983;73:1036-1039.
25. Battista RN, Williams JI, MacFarlane LA. Determinants of primary medical practice in adult cancer prevention. Med Care. 1986;24:216-224.
26. Battista RN, Palmer CS, Marchand BM, Spitzer WO. Patterns of preventive practice in New Brunswick. CMAJ. 1985;132:1013-1015.
27. Bergner M, Allison CJ, Diehr P, Ford LG, Feigl P. Early detection and control of cancer in clinical practice. Arch Intern Med. 1990;150:431-436.
28. Gemson DH, Elinson J. Cancer screening and prevention: knowledge, attitudes, and practices of New York City physicians. N Y State J Med. 1987;87:643-645.
29. Gemson DH, Elinson J. Prevention in primary care: variability in physician practice patterns in New York City. Am J Prev Med. 1986;2:226-234.
30. Gemson DH, Elinson J, Messeri P. Differences in physician prevention practice patterns for white and minority patients. J Community Health. 1988;13:53-64.
31. Hutchison BG, Abelson J, Woodward CA, Norman G. Preventive care and barriers to effective prevention: how do family physicians see it? Can Fam Physician. 1996;42:1693-1700.
32. Mandel IG, Franks P, Dickinson JC. Screening guidelines in a family medicine program: a five-year experience. J Fam Pract. 1982;14:901-907.
33. Proiskos A, Loukidou E, Kamposioras K, et al. Screening chest radiography in primary care: an underestimate belief. Eur J Gen Pract. 2005;11:76-77.
34. Resnicow K, Schorow M, Bloom HG, Massad R, Coll-Barth M. Screening practices of family physicians: a comparison of STFM and AAFP members. Fam Med. 1987;19:341-345.
35. Schwartz JS, Lewis CE, Clancy C, Kinosian MS, Radany MH, Koplan J. Internists' practices in health promotion and disease prevention: a survey. Ann Intern Med. 1991;114:46-53.
36. Simpson FG, Morrison JFJ, Cooke NJ, Pearson SB. General practitioner referrals for static miniature chest radiography: indications and diagnostic yield. Br J Dis Chest. 1988;82:76-78.
37. Woo B, Woo B, Cook EF, Weisberg M, Goldman L. Screening procedures in the asymptomatic adult: comparison of physicians' recommendations, patients' desires, published guidelines, and actual practice. JAMA. 1985;254:1480-1484.
38. Woodward CA, Hutchison BG, Abelson J, Norman G. Do female primary care physicians practise preventive care differently from their male colleagues? Can Fam Physician. 1996;42:2370-2379.
39. Hutchison B, Woodward CA, Norman GR, Abelson J, Brown JA. Provision of preventive care to unannounced standardized patients. CMAJ. 1998;158:185-193.
40. Ganry O, Boche T. Prevention practices and cancer screening among general practitioners in Picardy (France). Bull Cancer. 2004;91:785-791.
41. Eddy D. ACS report on the cancer-related health checkup. CA Cancer J Clin. 1980;30:193-240.
42. Gray JA. New concepts in screening. Br J Gen Pract. 2004;54:292-298.
43. Diederich S, Lenzen H. Radiation exposure associated with imaging of the chest: comparison of different radiographic and computed tomography techniques. Cancer. 2000;89(suppl):2457-2460.
44. Gohagan JK, Prorok PC, Hayes RB, Kramer BS; Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial Project Team. The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial of the National Cancer Institute: history, organization, and status. Control Clin Trials. 2000;21(suppl):251S-272S.
45. Montano DE, Manders DB, Phillips WR. Family physician beliefs about cancer screening: development of a survey instrument. J Fam Pract. 1990;30:313-319.
46. Zitzelsberger L, Grunfeld E, Graham ID. Family physicians' perspectives on practice guidelines related to cancer control. BMC Fam Pract. 2004;5:e25.
47. Stocks N, Braunack-Mayer A, Somerset M, Gunell D. Binners, fillers and filers: a qualitative study of GPs who don't return postal questionnaires. Eur J Gen Pract. 2004;10:146-151.