In spite of numerous prior studies demonstrating disparities in lung cancer outcomes, several key issues remain unclear. To what extent do survival differences reflect individual or institutional or system factors? To what degree are they related to variance in treatment? Does stage distribution among populations underlie these disparities? We evaluated these and other factors through a cohort study of consecutive patients with stage I-III NSCLC treated at a large, urban North American academic medical center with a diverse patient population. At UT Southwestern, a single medical faculty treats patients in a large safety net hospital system and in a tertiary care NCI-designated cancer center. Health care providers apply a single set of institutional clinical practice guidelines to all patients and present cases at a single multidisciplinary tumor board. Despite the homogeneity of medical care, the present study demonstrates that socioeconomic status was a principal determinant of whether or not a patient received stage-specific “standard” treatment. Specifically, indigent patients were less likely than those with private insurance to undergo surgery for stage I-II NSCLC, and less likely to undergo surgery- or radiation-based therapy for stage III NSCLC. Even when controlling for treatment selection, socioeconomic status remained associated with overall survival in stage I-II disease.
Socioeconomic status was significantly associated with a number of baseline patient and disease characteristics that may impact treatment selection and clinical outcomes. Compared to patients with private insurance, indigent patients were less likely to be female or non-Hispanic white. They presented with more advanced stage disease and were less likely to have adenocarcinoma histology. These findings are consistent with a national study, in which patients without insurance were twice as likely to present with advanced stage disease when compared to individuals with private insurance.(4
) Furthermore, inferior outcomes in male patients and non-adenocarcinoma histology are well established.(23
) The lower rate of adenocarcinoma histology in our indigent population may indicate that this group had a more extensive smoking history, which in turn could potentially influence candidacy for and tolerance of surgery for stage I-II disease. Greater tobacco use could also independently affect overall survival through co-morbidities such as chronic obstructive pulmonary disease and cardiovascular disease, as well as through the development of other tobacco-related malignancies. It has been suggested that, compared to white patients, black patients are less likely to quit smoking after a diagnosis of lung cancer.(27
) Despite these potential explanations, when we controlled for all other variables in multivariate models, socioeconomic status remained independently associated with both treatment selection and overall survival in stage I-II disease. Specifically, indigent patients were less likely to undergo surgery than were individuals with private insurance (OR for undergoing surgery 0.13; 95% CI, 0.04 to 0.43), and their risk of death was twice that of individuals with private insurance (HR 1.98; 95% CI, 1.16 to 3.37).
These findings echo those of large, population-based studies for a variety of malignancies. In the 1970s, a retrospective chart review of over 1,400 patients with NSCLC Stage I-III in the New England area found that patients with private insurance were more likely to undergo surgery for lung cancer than others even when controlling for disease stage.(28
) Coburn et al
found that 96% of patients with private insurance received surgical treatment for breast cancer, compared with only 85% of women without insurance.(29
) Because these studies reported data from state registries or from multiple medical centers, it is difficult to discern whether these findings result from system or from individual behaviors. Our current study suggests that individual patient factors contribute to both treatment selection and clinical outcomes, findings that contrast those of prior single-institution analyses.(15
) However, these results are consistent with studies of first- and second-line chemotherapy administration for advanced NSCLC performed at our institution.(30
Historically, there have been fewer investigations into treatment and outcome disparities for locally advanced NSCLC. Hardy and colleagues found that black patients were significantly less likely than white patients to receive surgery in stage III NSCLC, but found no differences in the receipt of radiation.(32
) In multivariate analysis, they also found that patients in the lowest income quartile were less likely to receive radiation in stage III-IV disease. In our study, older patients and indigent patients were less likely to receive “standard” (surgery- and/or radiation-based) treatment. While non-white patients were also less likely to receive “standard” therapy, this trend did not reach statistical significance (P
=0.23). In contrast to stage I-II NSCLC, only treatment selection remained associated with overall survival for stage III NSCLC in multivariate analysis. It is possible that underlying patient and disease characteristics may have less influence on clinical outcomes in this setting due to a clinical course that is generally shorter and less variable.
This study suggests that treatment and outcome disparities persist in an environment designed to provide uniform care across populations. However, the underlying reasons for these findings remain unclear. It is possible that under-represented populations elect to receive less aggressive care. For instance, some studies have shown that black patients are less likely to proceed with surgery for lung cancer, even when it is recommended.(6
) However, other studies have demonstrated that minorities seek more aggressive care, such as in the acute care setting.(34
) Physician bias could be another explanation for these findings. One previous study reported that black patients were less likely to be recommended surgery for NSCLC than white patients.(36
) However, when limited to a single center—as is our current study—these findings were not reproducible.(37
) Potentially limiting individual physician biases, early stage and locally advanced NSCLC cases undergo multidisciplinary tumor board review at our institution, thereby incorporating consensus treatment recommendations. Perhaps, following definitive therapy, differences in clinical and radiographic monitoring schedules and adherence among populations could contribute to survival disparities. Alternatively, greater rates of surgery in the private insurance group might result in upstaging of a greater proportion of these patients and, consequently, relative improvement in their stage-specific survival compared to clinically staged patients. Another reasonable explanation is that socioeconomically disadvantaged patients may be less fit for specific treatments. As mentioned previously, the greater proportion of squamous tumors in our indigent population may indicate greater prevalence and intensity of tobacco use, which in turn could predispose to other medical conditions and limit lung cancer therapeutic options. Unfortunately, a detailed analysis of patient fitness lies beyond the scope of this report. Smoking history, performance status, and medical comorbidities are not routinely captured in tumor registries. Nor are they consistently documented in the primary medical record. We are currently planning an evaluation of these variables. Identification of the causes of treatment and survival disparities—and measures to address them—will be essential to improving the care and outcome of socioeconomically disadvantaged patients with lung cancer.
Principal limitations of this study include its retrospective nature, relatively small size, and single-institution setting. However, it was precisely the single-center setting that permitted a focus on individual, rather than system, disparities. While our center provides access to a diverse patient cohort, certain characteristics may not be fully representative of the larger lung cancer population. Notably, the median age at diagnosis in this study was 64 years, approximately 6 years younger than the national average. We have reported similar age distributions in studies of advanced NSCLC at our institution.(30
) This trend may be driven by patients with an indigent health plan, who had a median age of 58 years. Whether this reflects the greater tobacco use in medically underserved populations(38
) and hence earlier tumor initiation and development, is unclear. In the current study, the younger age distribution clearly impacts insurance designation (with Medicare generally available to individuals age 65 years and older), our surrogate marker of socioeconomic status. Additionally, our treatment categorization did not fully capture the nature of a patient’s therapy, such as whether adjuvant chemotherapy was administered for early stage disease (relatively unlikely given the study time period), or whether definitive thoracic radiation was administered alone or with chemotherapy for locally advanced disease. Compared to our primary treatment categorizations (any surgery, any radiation, or neither), though, one would expect these further distinctions to have less impact on outcomes. Key strengths of this study include diversity of the patient population, a uniform approach to NSCLC treatment in the study setting, completeness of data and follow-up, and inclusion of several potentially confounding factors in the multivariable models.