Lung Cancer Screening Practices of Primary Care Physicians: Results From a National Survey

INTRODUCTION

Lung cancer is the leading cause of cancer death in the United States,^1,2 with a 5-year survival rate (16%) that is substantially lower than that for other common cancers.³ Despite periodic debate about the utility of screening patients for lung cancer,^4–7 no randomized controlled trial has demonstrated that screening with chest radiograph or sputum cytology reduces mortality from lung cancer.^8–11 In the United States, a large, randomized controlled trial comparing chest radiograph vs usual care has completed accrual; results assessing impact on lung cancer mortality are expected in late 2015.¹⁰ A second large, randomized controlled trial comparing low-dose spiral computed tomography (LDCT) vs chest radiograph in current and former heavy cigarette smokers, the National Lung Screening Trial (NLST), was halted in October 2010 after observation of a 20% reduction in lung cancer mortality in the LDCT group.^11,12 To date, because of a lack of evidence from rigorous studies, major expert groups have not recommended screening asymptomatic individuals, even those with heavy or long-term smoking histories, for lung cancer.^13–15 Whether expert groups will modify their lung cancer screening guidelines based on NLST results is at present unknown.

Past practice patterns raise concerns that NLST results may prompt overuse of lung cancer screening among average-risk patients. Despite the lack of strong evidence and guideline recommendations, prior studies have shown that many primary care physicians order chest radiographs to screen patients for lung cancer.^16–19 In the US health care system (which is decentralized), LDCT and other radiologic screening tests are widely available and increasingly marketed, and patients can obtain them through self-referral.^20,21 Because primary care physicians are many patients’ initial point of contact with the health care system and coordinators of preventive care, they may have a role in ordering lung cancer screening tests or interpreting their results; however, national data on these physicians’ knowledge, beliefs, and practices regarding lung cancer screening are sparse and out of date. Four national surveys conducted during the 1980s asked primary care physicians about their use of or recommendations for lung cancer screening with chest radiograph; results showed that up to 45% ordered or recommended it.^16,17,22,23 No recent national survey of health care professionals has focused on lung cancer screening, nor has any assessed the use of other tests that have been considered for lung cancer early detection, such as sputum cytology and LDCT.

The National Cancer Institute (NCI) addressed this gap by sponsoring the National Survey of Primary Care Physicians’ Recommendations & Practices for Breast, Cervical, Colorectal, & Lung Cancer Screening.²⁴ We previously used data from this survey to assess primary care physicians’ knowledge and beliefs related to screening asymptomatic patients for lung cancer.²⁵ In the present study, we use the data to examine US primary care physicians’ self-reported lung cancer screening practices, including characteristics of those physicians who order lung cancer screening tests.

METHODS

RESULTS

Lung Cancer Screening Beliefs and Practices, and Patient Inquiries

One-quarter of primary care physicians believed that 1 or more national expert groups recommend screening asymptomatic patients for lung cancer (Table 2). Physicians’ beliefs about the effectiveness of screening tests in reducing lung cancer mortality in asymptomatic patients varied by test modality and patients’ smoking status. For all 3 patient smoker statuses (never, former, current smokers), lower percentages of physicians expressed the belief that sputum cytology, compared with chest radiograph or LDCT, is effective in reducing lung cancer mortality. Higher percentages perceived LDCT as effective compared with chest radiograph. Physicians tended to rate the 3 tests (chest radiograph, sputum cytology, LDCT) as more effective in current than in never smokers.

View this table:

• View inline
• View popup

Table 2

Primary Care Physicians’ Lung Cancer Screening Beliefs, Self-Reported Practices, and Patient Influence (N=962)

More than one-half of physicians reported that, in the past 12 months, they had ordered chest radiograph to screen an asymptomatic patient for lung cancer, and nearly one-quarter had ordered LDCT (Table 2). In contrast, less than 5% reported ordering sputum cytology. Overall, 38% had ordered no tests, while 57% had ordered at least 1 of them. More than two-thirds of physicians reported that at least 1 patient in the past 12 months had asked them whether they could or should be screened for lung cancer.

In general, the percentages of physicians who had ordered lung cancer screening tests increased with their propensity to recommend screening asymptomatic patients for lung cancer (Figure 1). The proportions who had ordered such tests (chest radiograph and any test) were lowest among those who did not recommend screening and highest among those recommending screening for all patients, even patients without substantial smoking exposure. In contrast, a similar dose-response relationship was not observed for LDCT: although the percentages of physicians who had ordered LDCT were higher among those recommending screening for asymptomatic patients compared with physicians who did not recommend screening, they did not vary by patients’ smoking exposure.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Percentage of primary care physicians who ordered lung cancer screening tests by types of patients for whom they would recommend lung cancer screening.

LDCT = low-dose spiral computed tomography.

Note: Brackets indicate 95% CIs.

^a Chest radiograph, sputum cytology, or LDCT.

DISCUSSION

To our knowledge, this is the first national study of primary care physicians’ lung cancer screening practices to consider sputum cytology and LDCT—the new modality assessed in the recent NLST—in addition to chest radiograph. Four prior national surveys have evaluated primary care physicians’ use of or recommendations for chest radiograph ^15,16,21,22; however, those surveys took place more than 2 decades ago and did not include all 3 primary modalities that have been considered for lung cancer early detection.

Our findings that a majority of primary care physicians reported ordering lung cancer screening tests for asymptomatic patients and that patients have recently asked them about lung cancer screening suggest that ordering of these tests is common among these physicians, as are patient-physician discussions about lung cancer screening, even though at the time of the survey, lung cancer screening was not recommended by major expert groups in the United States. Although direct comparison of our estimates with those of earlier studies^16,17,22,23 is not possible because of differences in study designs, it appears that US primary care physicians have not decreased their practice of ordering chest radiographs to screen asymptomatic patients for lung cancer.

Primary care physicians’ ordering of unproven lung cancer screening technologies has several implications. There is potential for psychological harm from false-positive or false-negative test results, and physical harm from invasive procedures performed to follow up false-positive screening tests. Results from the NLST’s feasibility phase, the Lung Screening Study, indicated that the issue of false-positive results is nontrivial: the percentage of screened participants with a false-positive examination was higher for LDCT than for chest radiograph, and LDCT showed a false-positive rate on an initial screen of 21% and on a second screen of 33%.³⁴ Imaging tests can also contribute to overdiagnosis of lung cancer and other conditions,³⁵ and to radiation-induced cancers, especially with repeated LDCT examinations and the diagnostic imaging that often follows positive examinations.^36,37 Finally, use of unproven lung cancer screening technologies ultimately drives up health care costs.³⁷

To better understand reasons for primary care physicians’ apparent overuse of lung cancer screening, we examined the potential influence of physician and practice setting characteristics, physician beliefs and practice style, and patient demand on physicians’ lung cancer screening test ordering. We found in multivariate modeling that physicians who believe expert groups recommend lung cancer screening or that screening tests are effective in reducing lung cancer mortality, said they would recommend lung cancer screening for asymptomatic patients, or reported that their patients had asked them about being screened for lung cancer were significantly more likely to have ordered chest radiographs or LDCT as lung cancer screening tests in the past year. Physicians who graduated from medical school 20 years ago or more were significantly more likely to have ordered LDCT. Our findings that physician beliefs about screening and older physician age are associated with ordering lung cancer screening tests are consistent with previous studies of chest radiograph ordering.^18,22,23 The finding that patient demand is associated with physicians’ reports of lung cancer screening test ordering parallels earlier work showing that patient requests influence physicians’ ordering of genetic tests for cancer susceptibility.³² Our assessment of physician characteristics associated with the ordering of lung cancer screening tests may aid efforts to target educational interventions to physicians most in need of updated information about lung cancer screening’s evidence base and guidelines.

Strengths of this study include its large, nationally representative sample of practicing primary care physicians and high survey response rate. Our study also has some limitations. Physicians’ self-reports of their lung cancer screening recommendations and practices were not validated with other data sources such as medical records or claims. To minimize respondent burden, the survey module on lung cancer screening was relatively brief, and we were not able to ask more detailed questions about specific characteristics of patients for whom physicians ordered lung cancer screening tests, such as type and extent of smoking exposure. We also did not ask whether the physician’s practice had chest radiography available on site, which may influence chest radiograph ordering. Lastly, we did not ask to what extent physicians may be initiating discussions about lung cancer screening with their patients. All of these are key areas for future research on physicians’ lung cancer screening recommendations and practices.

The disconnect between lung cancer screening evidence and practice that our study documents provides important context for considering the potential consequences of the recent, highly publicized NLST findings.³⁸ Those findings are specific to individuals aged 50 to 75 years who were current or former smokers with at least a 30–pack-year history of smoking. Any change in screening recommendations that might occur as a result of the NLST would apply only to this select, high-risk population and 1 specific technology (ie, LDCT). Our results showing gaps in primary care physicians’ knowledge of lung cancer screening and use of unproven screening modalities suggest that in the United States—where most cancer screening occurs opportunistically rather than through organized programs—a substantial proportion of the adult population could be inappropriately screened unless there are concerted efforts to inform primary care physicians of appropriate interpretation of NLST findings and how best to apply them in practice.

Finally, 3 additional factors heighten the need for educating primary care physicians and patients about lung cancer screening’s evidence base, guidelines, potential harms, and costs. First, anecdotal evidence suggests that announcement of NLST results may have prompted some medical professionals to more widely promote lung cancer screening and their LDCT facilities³⁷; second, the US general public has an overly positive view of cancer screening and limited understanding of its potential harms³⁹; and third, use of computed tomography scans in general is rising rapidly in the United States.⁴⁰ These factors along with our study’s results and the evolving evidence base underscore the importance of continued monitoring of primary care physicians’ knowledge, beliefs, and practices related to lung cancer screening.