Search results
The search identified 17,047 citations. After removal of duplicates, titles and abstracts of 10,722 records were screened for eligibility. After title and abstract screening, 551 records progressed to full-text review. Overall, 62 papers (reporting 58 independent studies) met the inclusion criteria (Fig. 1) and were included in the review. Eight studies reported utilization data for predictive biomarker tests [21,22,23,24,25,26,27,28], thirty-seven studies (41 papers) reported utilization data for biological and precision therapies [29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65], and 3 studies reported both [66,67,68]. Ten papers (Additional file 1: Table S1) had no denominator populations or only reported an average measure of socio-economic status (e.g., mean household income), and were excluded from inclusion in the meta-analysis and are not discussed further [69,70,71,72,73,74,75,76,77,78].
Study characteristics
The 48 included studies covered 7 cancers, 5 predictive biomarker tests, and 11 biological and precision therapy classifications, of which bevacizumab (12 studies) [41,42,43,44,45, 54,55,56, 58, 59, 64, 65] and trastuzumab (11 studies) [29,30,31,32,33,34,35,36,37,38,39] were most common. Most studies were in the USA (n = 42) [21, 22, 25,26,27,28,29,30,31,32,33,34,35, 40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68], and a majority analyzed SEER registry data (n = 27) [21, 22, 25, 29,30,31,32,33, 41, 42, 45, 47, 49, 50, 54,55,56,57,58,59, 61,62,63,64, 66,67,68] (Additional file 1: Fig. S1). Of the SEER data studies, 19 [29,30,31,32, 41, 42, 47, 49, 54,55,56,57,58,59, 61,62,63,64, 68] were SEER Medicare (i.e., included patients ≥ 65). The remaining studies were from Canada (4 studies) [23, 36,37,38], China (1 study) [39], and Ireland (1 study) [24]. Forty-six studies reported one or more area-based socio-economic status measure, and only two utilized individual-based measures [34, 68]. Six SES measures (poverty, income, education, employment, deprivation, and socio-economic status aggregate score) were reported. For nine studies, utilization was only available as percentages [24, 29, 32, 52, 54, 56, 61, 66, 67]. Study characteristics are summarized in Additional file 1: Table S2.
Seven papers, pertaining to four studies, reported the same data from the same registry [38, 43, 45, 79,80,81,82]. Sixteen papers (covering 8 studies) overlapped in their study populations (cancer site, stage, years of diagnosis time frames, patients’ age) [29,30,31,32, 36, 37, 41,42,43,44, 49, 50, 54, 55, 67, 68]. Two studies did not report unadjusted drug and/or test utilization data [40, 58]. This left 38 studies (including 1,036,125 patients) which were included in the meta-analysis [21,22,23,24,25,26,27,28,29, 31, 33,34,35, 37,38,39, 42, 44,45,46,47,48,49, 51,52,53,54, 56, 57, 59,60,61,62,63,64,65,66, 68].
Quality appraisal
The 48 studies scored in the range 4–10, out of a possible 10 (mean = 6.9, median = 6.5) (Additional file 1: Table S3). Papers scored well regarding data source(s), study populations, and reporting socio-economic definition(s). Discussion of results with reference to the role of socio-economic status, statistical analysis with summary measures like OR, and explanations for confounder selection were often reported poorly.
Predictive biomarker testing
Eleven studies reported data of interest for five predictive biomarker tests [21,22,23,24,25,26,27,28, 66,67,68]. Ten studies were included in the meta-analysis [21,22,23,24,25,26,27,28, 66, 68]. These covered the following cancers: breast (4 studies) [21,22,23,24], colorectal (3 studies) [25,26,27], melanoma (1 study) [66], and non-small cell lung (2 studies) [28, 68]. The pooled OR for predictive biomarker test receipt for low socio-economic status was 0.86 (95% CI 0.71–1.05; I2 = 86%; 10 studies) (Fig. 2). This pattern was consistent across cancer sub-groups (4 breast cancer studies, 2 lung cancer studies, and 1 melanoma study) but was only significant in colorectal cancer (0.76, 95% CI 0.65–0.88; 3 studies).
Biological and precision therapies: primary analysis
Association of socio-economic status with biological and precision therapy receipt was reported in 40 studies [29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68]. Thirty of which were included in the meta-analysis [29, 31, 33,34,35, 37,38,39, 42, 44,45,46,47,48,49, 51,52,53,54, 56, 57, 59,60,61,62,63,64,65,66, 68]. The overall pooled OR for receipt of biological and precision therapy for patients from low socio-economic status was 0.83 (95% CI 0.75–0.91; I2 = 85%; 30 studies) (Fig. 3). Sub-group analysis suggested stronger associations with immunotherapy utilization (0.82, 95% CI 0.78–0.86; 7 studies) than other therapy classes (14 targeted therapy and 9 biological therapy studies), but the test for sub-group differences was not significant (Fig. 3). Sensitivity analyses which substituted included studies for excluded studies with overlapping sampling frames confirmed the robustness of results (0.80, 95% CI 0.72–0.88; I2 = 86%; 30 studies). Similar results were also observed in sensitivity analyses when only USA studies were considered (0.82, 95% CI 0.74–0.91, I2 = 85%, 27 studies). For full sensitivity analyses results, see Additional File 1: Fig. S2.
Biological and precision therapies: sub-group analyses
For breast cancer, 11 studies reported the association of socio-economic status with the human epidermal growth factor receptor 2 (HER2) targeting monoclonal antibody trastuzumab [29,30,31,32,33,34,35,36,37,38,39] and one with immunotherapy [40]. Eight studies were eligible for meta-analysis [29, 31, 33,34,35, 37,38,39]. The pooled OR for receipt of trastuzumab in those with low compared to high socio-economic status was 0.93 (95% CI 0.78–1.10; I2 = 68%) (Fig. 4).
Nine lung cancer studies evaluated socio-economic status with biological and precision therapy receipt [41,42,43,44,45,46,47, 67, 68]. Four of these reported bevacizumab [41,42,43,44], 2 tyrosine kinase inhibitors [67, 68], 1 both bevacizumab and tyrosine kinase inhibitors [45], 1 immunotherapy [46], and 1 biological therapies (mostly bevacizumab) [47]. Six were eligible for meta-analysis [42, 44,45,46,47, 68], and the pooled OR for receipt of biological and precision therapies in those of low compared to high socio-economic status was 0.71 (95% CI 0.51–1.00; I2 = 95%) (Fig. 5).
Twenty studies reported data of interest for 6 other cancers: hepatobiliary (4 studies) [48,49,50,51], melanoma (3 studies) [52, 53, 66], colorectal (8 studies) [40, 54,55,56,57,58,59,60], renal cell carcinoma (1 study) [61], and head and neck cancer (2 studies) [62, 63]. A further two studies reported data on more than one cancer [64, 65]. Studies referenced the following 7 treatments: immunotherapy [40, 48, 52, 53, 60], bevacizumab [54,55,56, 58, 59, 64, 65], sorafenib [49,50,51], ipilimumab [66], targeted biologics [57], IL-2 [61], and cetuximab [62, 63]. Sixteen studies could be combined into meta-analyses [48, 49, 51,52,53,54, 56, 57, 59,60,61,62,63,64,65,66], giving a pooled OR for receipt of biological and precision therapies for low socio-economic status of 0.84 (95% CI 0.76–0.94; I2 = 73%) (Additional file 1: Fig. S3). The test for sub-group differences between breast, lung, and all other cancers was not significant (Additional file 1: Fig. S4).