- Email: [email protected]
Regional variation in lung and bronchus cancer survival in the US using mortality-to-incidence ratios
Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
Contents lists available at ScienceDirect
Spatial and Spatio-temporal Epidemiology journal homepage: www.elsevier.com/locate/sste
Original Research
Regional variation in lung and bronchus cancer survival in the US using mortality-to-incidence ratios Cassie L. Odahowski a,b,c, James R. Hébert a,b, Jan M. Eberth a,b,c,∗ a
Cancer Prevention and Control Program, University of South Carolina, Columbia, SC, United States Department of Epidemiology and Biostatistics, University of South Carolina, Columbia, SC, United States c SC Rural Health Research Center, University of South Carolina, Columbia, SC, United States b
a r t i c l e
i n f o
Article history: Received 6 July 2017 Revised 30 May 2018 Accepted 8 June 2018 Available online 18 June 2018 Keywords: Geographic information systems Lung diseases Cancer Regional planning Healthcare disparities
a b s t r a c t Despite major achievements aimed at reducing smoking over the last 50 years in the U.S., lung cancer remains the leading cause of cancer death. This study used mortality-toincidence rate ratios (MIR) calculated from 2008 to 2012 National Cancer Institute data to highlight state-level variations in relative lung and bronchus cancer survival. In an ad hoc sensitivity analysis, we calculated a correlation between our state-level MIRs and five-year 1-survival rates for states reporting incident lung and bronchus cancer cases (20 04–20 08) to the Surveillance, Epidemiology, and End Results (SEER) Program database. Differences were observed in state lung and bronchus cancer MIRs, with the highest MIR values (poor relative survival) in southern states and the lowest MIRs primarily in northeastern states. In our sensitivity analysis, state-level MIRs were highly correlated with 1-survival rates. Examining regional variation in survival using MIRs can be a useful tool for identifying areas of health disparities and conducting surveillance activities. © 2018 Elsevier Ltd. All rights reserved.
1. Introduction Although smoking prevalence has steadily decreased since the Surgeon General’s Report on Smoking and Health of 1964 (Islami et al., 2015), lung cancer remains the leading cause of cancer death in the US and worldwide (Siegel et al., 2017; Hébert et al., 2009), with over 70% of cases being diagnosed at late stage (Hébert et al., 2009). Early detection of the disease is critical for improving prognosis, as more than 50% of patients diagnosed at early stages survive at least 5 years (Wagner et al., 2012). Previous studies have shown regional disparities in rates of lung cancer incidence and mortality rates; however, less is known about regional variation in lung cancer survival.
Abbreviations: MIR, mortality-to-incidence rate ratios. Correspondence to: Arnold School of Public Health, University of South Carolina, 915 Greene St., Columbia, SC 29208, United States. E-mail address: [email protected] (J.M. Eberth). ∗
https://doi.org/10.1016/j.sste.2018.06.004 1877-5845/© 2018 Elsevier Ltd. All rights reserved.
(Siegel et al., 2017; Hébert et al., 2009; Wagner et al., 2012) The mortality-to-incidence rate ratio (MIR) has been used to quantify cancer mortality in relation to incidence for a specified geographic region or population. (Hébert et al., 2009; Asadzadeh Vostakolaei et al., 2011; Sunkara and Hébert, 2016; Sunkara and Hébert, 2015) The objective of this study was to construct and compare state-level MIRs to evaluate geographic disparities in lung and bronchus cancer survival the United States. National MIRs were also stratified by sex, race and ethnicity to further explore existing disparities. 2. Materials and methods MIRs were calculated for all U.S. states (including D.C.) and counties for which data were available in the National Cancer Institute (NCI) State Cancer Profiles website. (State Cancer Profiles, 2016) Calculations were based upon 5-year averages of incidence and mortality rates of lung and bronchus cancer from 2008 to 2012. Incidence and
108
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
Fig. 1. State and county level variation in lung and bronchus cancer MIRs, 2008–2012. The bars surrounding each state’s MIR value represent the range of MIR values for the state’s respective counties with the far left and far right endpoints representing the minimum and maximum county MIR value within each respective state (n = counties with a population ≥20,0 0 0, as reported in the U.S. 2010 Decennial Census). Counties with populations <20,0 0 0 were excluded to ensure rate stability.
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
109
Fig. 2. Quintiles of lung and bronchus cancer MIRs in the U.S., 2008–2012.
mortality rates were age-adjusted per 10 0,0 0 0 population per year to the 20 0 0 US standard population. (State Cancer Profiles, 2016) Data extraction from the NCI State Cancer Profiles was conducted in July 2016. MIR calculations were performed by dividing the ageadjusted mortality rate for a particular geographic area by the age-adjusted incidence rate per 10 0,0 0 0 for that same given area. Higher MIR values are indicative of worse relative survival per unit incidence in a given population. National incidence and mortality data were used to make additional MIR comparisons by sex, race and ethnicity. Race and ethnicity categories included non-Hispanic White, Hispanic White, Black, American Indian or Alaskan native, Asian or Pacific Islander, and all Hispanic. States were ranked by ascending MIR and divided into quintiles for mapping purposes to highlight regional variation in lung and bronchus cancer MIRs. A second map was created using a continuous scale in order to visualize one-unit changes in MIR. Kansas and Minnesota county-level estimates were not available due to state regulations limiting the release of county-level data to protect patient confidentiality. Data from the Surveillance, Epidemiology, and End Results Program (SEER) were used to compare MIR values with the five-year 1-survival rates for all lung and bronchus cancer cases newly diagnosed between 2004 and 2008 for California, Connecticut, Georgia, Hawaii, Iowa, Kentucky, Louisiana, New Jersey, New Mexico, and Utah (i.e., SEER states with entire state coverage). Incident cases in the 20 04–20 08 timeframe were used to capture a full 5-year survival time through the end of 2012, overlapping with the timeframe comprising our MIR. We calculated the
correlation between the MIR values and 1-survival rates for theses selected states using Fisher’s z-transformation and then performed a linear regression of the relationship. All data management and descriptive statistics were performed in Microsoft Excel and SAS® Version 9.3. Maps were created using ArcGIS® Version 10.2. 3. Results A mean MIR of 0.75 for lung and bronchus cancer was observed across all U.S. states. MIRs ranged from 0.83 (Arkansas) to 0.65 (Connecticut; Fig. 1). The five states with the lowest relative survival from lung and bronchus cancer, as evidenced by the highest MIR values, were all in the South: Arkansas, Oklahoma, Tennessee, Alabama, and Louisiana. The five states with the lowest MIR values, indicating the best relative survival, were primarily located in the Northeast: Connecticut, New York, Massachusetts, New Jersey, and Hawaii. Interestingly, the state with the highest lung and bronchus incidence and mortality rate in the U.S. – Kentucky – had an MIR value higher than the national average and it fell into the top third of states for relative survival. In contrast, a contiguous cluster of low-performing states was found in the upper mid-Atlantic and Rust Belt portion of the U.S., including Michigan, Indiana, Ohio, West Virginia, and Virginia (Figs. 2 and 3). When national MIRs were stratified by sex, males had a higher overall MIR than females (0.83 vs. 0.74), and when stratified by race and ethnicity, all Hispanics had the lowest MIR at 0.60, while non-Hispanic Whites had the highest at 0.75. Non-Hispanic Whites were closely followed by
110
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
Fig. 3. Continuous scale of lung and bronchus cancer MIRs in the U.S., 2008–2012.
Fig. 4. Differences in lung and bronchus cancer MIRs in the U.S. by race and ethnicity, 2008–2012.
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
111
Fig. 5. Comparison of SEER five year 1-survival rates versus MIRs for lung and bronchus cancer by state.
Blacks at 0.74, American Indian and Alaskan Natives at 0.69, and Asian and Pacific Islanders at 0.67 (Fig. 4). The state-level five-year 1-survival rates from SEER were highly and significantly correlated with state-level MIRs (Pearson r = 0.88 (p = 0.002)). Fig. 5 shows the fitted regression line, with a R2 value of 0.78 and corresponding correlation coefficient of 0.88. For comparative purposes, the dashed line in Fig. 5 illustrates a line of equality where MIR would be exactly equal to 1-survival. 4. Discussion and conclusions This study highlights the utility of using the MIR to describe regional variations in cancer survival across areas or populations. The calculation and mapping of state-level MIRs for lung and bronchus cancer revealed that many southern states have low relative lung and bronchus cancer survival rates. The proportion of cancer deaths attributable to cigarette smoking in each state mimics this same geographic trend. (Lortet-Tieulent et al., 2016) Despite the high incidence and mortality rates of lung and bronchus cancer observed in Kentucky, and their high smoking prevalence, (Siegel et al., 2017) their low MIR suggests that other factors, such as clinical interventions and treatment programs,
are leading to a better prognosis for lung cancer patients in comparison to other southern states. Our findings showed that Hispanics had the lowest mortality-to-incidence ratio for lung and bronchus cancer in comparison to any other ethnicity category in our data. This type of “Hispanic Paradox” has been documented in other studies on ethnic differences in the survival of lung cancer. (Markides and Eschbach, 2011; Lariscy et al., 2015) Additional studies have documented social and biological factors affecting lung cancer survival, (Arrieta et al., 2015; Patel et al., 2013) including social support structures, tumor characteristics, and histologic types which may account for improved survival in Hispanic lung cancer patients vs. non-Hispanic White patients. Alternatively, because Hispanic lung cancer patients are more likely to be diagnosed at a distant stage, (Siegel et al., 2015) some researchers caution that their survival rates should be interpreted carefully due to possible loss to follow-up. (Pinheiro et al., 2011) The lower MIRs among Hispanics observed in our study could indicate better relative survival or could be an artifact of migration of Hispanic lung cancer patients following diagnosis but prior to death. Our findings also showed that, compared to men, women had better lung and bronchus cancer survival. This also has
112
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
been documented previously in the literature, pointing to differences in smoking history and tumor types (Belani et al., 2007; Ferguson et al., 1990; Ringer et al., 2005). Our examination of the correlation of our MIR values with five-year 1-survival aligns with results from previous literature showing that the MIR is highly correlated with 1-survival from SEER data, (Asadzadeh Vostakolaei et al., 2011) and provides additional evidence supporting the use of the MIR for approximating 1 minus relative survival for lung and bronchus cancer. This has important implications for future work in this arena because of the vast difference in resources needed to compute survival compared to what is required to compute the MIR. Limitations of this study include its ecological design and our inability to include data for Kansas or Minnesota counties due to patient confidentiality regulations. This study also is limited by the lack of publicly available stagespecific incidence and mortality data, which could help explain the observed regional variation in relative survival. Mortality-to-incidence ratios are most accurate in approximating relative survival in environments with stable incidence and mortality rates along with meticulous and timely case reporting. MIRs are computed from available cancer incidence and mortality data and presented as a point estimate (just as for survival). As with any measure, there may be errors in both the numerator and denominator. However, cancer registries rely on the legal mandate to report cancers centrally and strict requirements for data quality. Likewise, data entered into the national death index and local registries are subject to strict reporting requirements and quality control. Future studies on stage-specific relative survival are warranted to further explore regional variation in MIRs across and within states. Such an analysis would allow one to investigate whether observed MIR variations are due to underlying differences in the stage distribution of lung and bronchus cancer patients, or other factors such as access to high quality treatment (e.g., Commission on Cancer-accredited or NCI-designated Cancer Centers). Additionally, future research should focus on identifying the underlying risk factors for poor lung and bronchus cancer outcomes, controlling for known covariates such as comorbidities, tumor histology and stage at diagnosis. This may entail exploring variation at smaller scales and examining the impact of factors such as access to healthcare, receipt of guideline-concordant/quality treatment, lifestyle behaviors (e.g., prevalence of menthol cigarette smoking), and socioeconomic status on patient outcomes. Funding No external funding was used for this analysis.
References Arrieta O, Ramírez-Tirado L-A, Báez-Saldaña R, Peña-Curiel O, SocaChafre G, Macedo-Perez E-O. Different mutation profiles and clinical characteristics among Hispanic patients with non-small cell lung cancer could explain the “Hispanic paradox”. Lung Cancer 2015;90(2):161–6. doi:10.1016/j.lungcan.2015.08.010. Asadzadeh Vostakolaei F, Karim-Kos HE, Janssen-Heijnen MLG, Visser O, Verbeek ALM, Kiemeney LALM. The validity of the mortality to incidence ratio as a proxy for site-specific cancer survival. Eur J Public Health 2011;21(5):573–7. doi:10.1093/eurpub/ckq120. Belani CP, Marts S, Schiller J, Socinski MA. Women and lung cancer: epidemiology, tumor biology, and emerging trends in clinical research. Lung Cancer 2007;55(1):15–23. doi:10.1016/j.lungcan.20 06.09.0 08. Ferguson MK, Skosey C, Hoffman PC, Golomb HM. Sex-associated differences in presentation and survival in patients with lung cancer. J Clin Oncol 1990;8(8):1402–7 http://www.ncbi.nlm.nih.gov/pubmed/ 2166143. Accessed July 182016. Hébert JR, Daguise VG, Hurley DM, et al. Mapping cancer mortality-toincidence ratios to illustrate racial and sex disparities in a high-risk population. Cancer 2009;115(11):2539–52. doi:10.1002/cncr.24270. Islami F, Torre LA, Jemal A. Global trends of lung cancer mortality and smoking prevalence. Transl Lung Cancer Res 2015;4(4):327–38. doi:10.3978/j.issn.2218-6751.2015.08.04. Lariscy JT, Hummer RA, Hayward MD. Hispanic older adult mortality in the United States: new estimates and an assessment of factors shaping the hispanic paradox. Demography 2015;52(1):1–14. doi:10.1007/s13524-014-0357-y. Lortet-Tieulent J, Goding Sauer A, Siegel RL, et al. State-level cancer mortality attributable to cigarette smoking in the United States. JAMA Int Med 2016;176(12):1792. doi:10.1001/jamainternmed.2016.6530. Markides KS, Eschbach K. Hispanic paradox in adult mortality in the United States. Netherlands: Springer; 2011. p. 227–40. doi: 10.1007/ 978- 90- 481- 9996- 9_11 Patel MI, Schupp CW, Gomez SL, Chang ET, Wakelee HA. How do social factors explain outcomes in non-small-cell lung cancer among Hispanics in California? Explaining the Hispanic paradox. J Clin Oncol 2013;31(28):3572–8. doi:10.1200/JCO.2012.48.6217. Pinheiro PS, Williams M, Miller EA, Easterday S, Moonie S, Trapido EJ. Cancer survival among latinos and the hispanic paradox. Cancer Causes Control 2011;22(4):553–61. doi:10.1007/s10552-011-9727-6. Ringer G, Smith JM, Engel AM, Pat Hendy M, Lang J. Influence of sex on lung cancer histology, stage, and survival in a Midwestern United States tumor registry. Clin Lung Cancer 2005;7(3):180–2. doi:10.3816/CLC.2005.n.033. Siegel RL, Fedewa SA, Miller KD, et al. Cancer statistics for Hispanics/Latinos, 2015. CA Cancer J Clin 2015;65(6):457–80. doi:10.3322/caac.21314. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin 2017. doi:10.3322/caac.21387. State Cancer Profiles. http://statecancerprofiles.cancer.gov/map/map. noimage.php. Accessed June 28, 2016. Sunkara V, Hébert JR. The colorectal cancer mortality-to-incidence ratio as an indicator of global cancer screening and care. Cancer 2015;121(10):1563–9. doi:10.1002/cncr.29228. Sunkara V, Hébert JR. The application of the mortality-to-incidence ratio for the evaluation of cancer care disparities globally. Cancer 2016;122(3):487–8. doi:10.1002/cncr.29746. Wagner SE, Hurley DM, Hébert JR, McNamara C, Bayakly AR, Vena JE. Cancer mortality-to-incidence ratios in Georgia: describing racial cancer disparities and potential geographic determinants. Cancer 2012;118(16):4032–45. doi:10.1002/cncr.26728.
Contents lists available at ScienceDirect
Spatial and Spatio-temporal Epidemiology journal homepage: www.elsevier.com/locate/sste
Original Research
Regional variation in lung and bronchus cancer survival in the US using mortality-to-incidence ratios Cassie L. Odahowski a,b,c, James R. Hébert a,b, Jan M. Eberth a,b,c,∗ a
Cancer Prevention and Control Program, University of South Carolina, Columbia, SC, United States Department of Epidemiology and Biostatistics, University of South Carolina, Columbia, SC, United States c SC Rural Health Research Center, University of South Carolina, Columbia, SC, United States b
a r t i c l e
i n f o
Article history: Received 6 July 2017 Revised 30 May 2018 Accepted 8 June 2018 Available online 18 June 2018 Keywords: Geographic information systems Lung diseases Cancer Regional planning Healthcare disparities
a b s t r a c t Despite major achievements aimed at reducing smoking over the last 50 years in the U.S., lung cancer remains the leading cause of cancer death. This study used mortality-toincidence rate ratios (MIR) calculated from 2008 to 2012 National Cancer Institute data to highlight state-level variations in relative lung and bronchus cancer survival. In an ad hoc sensitivity analysis, we calculated a correlation between our state-level MIRs and five-year 1-survival rates for states reporting incident lung and bronchus cancer cases (20 04–20 08) to the Surveillance, Epidemiology, and End Results (SEER) Program database. Differences were observed in state lung and bronchus cancer MIRs, with the highest MIR values (poor relative survival) in southern states and the lowest MIRs primarily in northeastern states. In our sensitivity analysis, state-level MIRs were highly correlated with 1-survival rates. Examining regional variation in survival using MIRs can be a useful tool for identifying areas of health disparities and conducting surveillance activities. © 2018 Elsevier Ltd. All rights reserved.
1. Introduction Although smoking prevalence has steadily decreased since the Surgeon General’s Report on Smoking and Health of 1964 (Islami et al., 2015), lung cancer remains the leading cause of cancer death in the US and worldwide (Siegel et al., 2017; Hébert et al., 2009), with over 70% of cases being diagnosed at late stage (Hébert et al., 2009). Early detection of the disease is critical for improving prognosis, as more than 50% of patients diagnosed at early stages survive at least 5 years (Wagner et al., 2012). Previous studies have shown regional disparities in rates of lung cancer incidence and mortality rates; however, less is known about regional variation in lung cancer survival.
Abbreviations: MIR, mortality-to-incidence rate ratios. Correspondence to: Arnold School of Public Health, University of South Carolina, 915 Greene St., Columbia, SC 29208, United States. E-mail address: [email protected] (J.M. Eberth). ∗
https://doi.org/10.1016/j.sste.2018.06.004 1877-5845/© 2018 Elsevier Ltd. All rights reserved.
(Siegel et al., 2017; Hébert et al., 2009; Wagner et al., 2012) The mortality-to-incidence rate ratio (MIR) has been used to quantify cancer mortality in relation to incidence for a specified geographic region or population. (Hébert et al., 2009; Asadzadeh Vostakolaei et al., 2011; Sunkara and Hébert, 2016; Sunkara and Hébert, 2015) The objective of this study was to construct and compare state-level MIRs to evaluate geographic disparities in lung and bronchus cancer survival the United States. National MIRs were also stratified by sex, race and ethnicity to further explore existing disparities. 2. Materials and methods MIRs were calculated for all U.S. states (including D.C.) and counties for which data were available in the National Cancer Institute (NCI) State Cancer Profiles website. (State Cancer Profiles, 2016) Calculations were based upon 5-year averages of incidence and mortality rates of lung and bronchus cancer from 2008 to 2012. Incidence and
108
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
Fig. 1. State and county level variation in lung and bronchus cancer MIRs, 2008–2012. The bars surrounding each state’s MIR value represent the range of MIR values for the state’s respective counties with the far left and far right endpoints representing the minimum and maximum county MIR value within each respective state (n = counties with a population ≥20,0 0 0, as reported in the U.S. 2010 Decennial Census). Counties with populations <20,0 0 0 were excluded to ensure rate stability.
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
109
Fig. 2. Quintiles of lung and bronchus cancer MIRs in the U.S., 2008–2012.
mortality rates were age-adjusted per 10 0,0 0 0 population per year to the 20 0 0 US standard population. (State Cancer Profiles, 2016) Data extraction from the NCI State Cancer Profiles was conducted in July 2016. MIR calculations were performed by dividing the ageadjusted mortality rate for a particular geographic area by the age-adjusted incidence rate per 10 0,0 0 0 for that same given area. Higher MIR values are indicative of worse relative survival per unit incidence in a given population. National incidence and mortality data were used to make additional MIR comparisons by sex, race and ethnicity. Race and ethnicity categories included non-Hispanic White, Hispanic White, Black, American Indian or Alaskan native, Asian or Pacific Islander, and all Hispanic. States were ranked by ascending MIR and divided into quintiles for mapping purposes to highlight regional variation in lung and bronchus cancer MIRs. A second map was created using a continuous scale in order to visualize one-unit changes in MIR. Kansas and Minnesota county-level estimates were not available due to state regulations limiting the release of county-level data to protect patient confidentiality. Data from the Surveillance, Epidemiology, and End Results Program (SEER) were used to compare MIR values with the five-year 1-survival rates for all lung and bronchus cancer cases newly diagnosed between 2004 and 2008 for California, Connecticut, Georgia, Hawaii, Iowa, Kentucky, Louisiana, New Jersey, New Mexico, and Utah (i.e., SEER states with entire state coverage). Incident cases in the 20 04–20 08 timeframe were used to capture a full 5-year survival time through the end of 2012, overlapping with the timeframe comprising our MIR. We calculated the
correlation between the MIR values and 1-survival rates for theses selected states using Fisher’s z-transformation and then performed a linear regression of the relationship. All data management and descriptive statistics were performed in Microsoft Excel and SAS® Version 9.3. Maps were created using ArcGIS® Version 10.2. 3. Results A mean MIR of 0.75 for lung and bronchus cancer was observed across all U.S. states. MIRs ranged from 0.83 (Arkansas) to 0.65 (Connecticut; Fig. 1). The five states with the lowest relative survival from lung and bronchus cancer, as evidenced by the highest MIR values, were all in the South: Arkansas, Oklahoma, Tennessee, Alabama, and Louisiana. The five states with the lowest MIR values, indicating the best relative survival, were primarily located in the Northeast: Connecticut, New York, Massachusetts, New Jersey, and Hawaii. Interestingly, the state with the highest lung and bronchus incidence and mortality rate in the U.S. – Kentucky – had an MIR value higher than the national average and it fell into the top third of states for relative survival. In contrast, a contiguous cluster of low-performing states was found in the upper mid-Atlantic and Rust Belt portion of the U.S., including Michigan, Indiana, Ohio, West Virginia, and Virginia (Figs. 2 and 3). When national MIRs were stratified by sex, males had a higher overall MIR than females (0.83 vs. 0.74), and when stratified by race and ethnicity, all Hispanics had the lowest MIR at 0.60, while non-Hispanic Whites had the highest at 0.75. Non-Hispanic Whites were closely followed by
110
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
Fig. 3. Continuous scale of lung and bronchus cancer MIRs in the U.S., 2008–2012.
Fig. 4. Differences in lung and bronchus cancer MIRs in the U.S. by race and ethnicity, 2008–2012.
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
111
Fig. 5. Comparison of SEER five year 1-survival rates versus MIRs for lung and bronchus cancer by state.
Blacks at 0.74, American Indian and Alaskan Natives at 0.69, and Asian and Pacific Islanders at 0.67 (Fig. 4). The state-level five-year 1-survival rates from SEER were highly and significantly correlated with state-level MIRs (Pearson r = 0.88 (p = 0.002)). Fig. 5 shows the fitted regression line, with a R2 value of 0.78 and corresponding correlation coefficient of 0.88. For comparative purposes, the dashed line in Fig. 5 illustrates a line of equality where MIR would be exactly equal to 1-survival. 4. Discussion and conclusions This study highlights the utility of using the MIR to describe regional variations in cancer survival across areas or populations. The calculation and mapping of state-level MIRs for lung and bronchus cancer revealed that many southern states have low relative lung and bronchus cancer survival rates. The proportion of cancer deaths attributable to cigarette smoking in each state mimics this same geographic trend. (Lortet-Tieulent et al., 2016) Despite the high incidence and mortality rates of lung and bronchus cancer observed in Kentucky, and their high smoking prevalence, (Siegel et al., 2017) their low MIR suggests that other factors, such as clinical interventions and treatment programs,
are leading to a better prognosis for lung cancer patients in comparison to other southern states. Our findings showed that Hispanics had the lowest mortality-to-incidence ratio for lung and bronchus cancer in comparison to any other ethnicity category in our data. This type of “Hispanic Paradox” has been documented in other studies on ethnic differences in the survival of lung cancer. (Markides and Eschbach, 2011; Lariscy et al., 2015) Additional studies have documented social and biological factors affecting lung cancer survival, (Arrieta et al., 2015; Patel et al., 2013) including social support structures, tumor characteristics, and histologic types which may account for improved survival in Hispanic lung cancer patients vs. non-Hispanic White patients. Alternatively, because Hispanic lung cancer patients are more likely to be diagnosed at a distant stage, (Siegel et al., 2015) some researchers caution that their survival rates should be interpreted carefully due to possible loss to follow-up. (Pinheiro et al., 2011) The lower MIRs among Hispanics observed in our study could indicate better relative survival or could be an artifact of migration of Hispanic lung cancer patients following diagnosis but prior to death. Our findings also showed that, compared to men, women had better lung and bronchus cancer survival. This also has
112
C.L. Odahowski et al. / Spatial and Spatio-temporal Epidemiology 26 (2018) 107–112
been documented previously in the literature, pointing to differences in smoking history and tumor types (Belani et al., 2007; Ferguson et al., 1990; Ringer et al., 2005). Our examination of the correlation of our MIR values with five-year 1-survival aligns with results from previous literature showing that the MIR is highly correlated with 1-survival from SEER data, (Asadzadeh Vostakolaei et al., 2011) and provides additional evidence supporting the use of the MIR for approximating 1 minus relative survival for lung and bronchus cancer. This has important implications for future work in this arena because of the vast difference in resources needed to compute survival compared to what is required to compute the MIR. Limitations of this study include its ecological design and our inability to include data for Kansas or Minnesota counties due to patient confidentiality regulations. This study also is limited by the lack of publicly available stagespecific incidence and mortality data, which could help explain the observed regional variation in relative survival. Mortality-to-incidence ratios are most accurate in approximating relative survival in environments with stable incidence and mortality rates along with meticulous and timely case reporting. MIRs are computed from available cancer incidence and mortality data and presented as a point estimate (just as for survival). As with any measure, there may be errors in both the numerator and denominator. However, cancer registries rely on the legal mandate to report cancers centrally and strict requirements for data quality. Likewise, data entered into the national death index and local registries are subject to strict reporting requirements and quality control. Future studies on stage-specific relative survival are warranted to further explore regional variation in MIRs across and within states. Such an analysis would allow one to investigate whether observed MIR variations are due to underlying differences in the stage distribution of lung and bronchus cancer patients, or other factors such as access to high quality treatment (e.g., Commission on Cancer-accredited or NCI-designated Cancer Centers). Additionally, future research should focus on identifying the underlying risk factors for poor lung and bronchus cancer outcomes, controlling for known covariates such as comorbidities, tumor histology and stage at diagnosis. This may entail exploring variation at smaller scales and examining the impact of factors such as access to healthcare, receipt of guideline-concordant/quality treatment, lifestyle behaviors (e.g., prevalence of menthol cigarette smoking), and socioeconomic status on patient outcomes. Funding No external funding was used for this analysis.
References Arrieta O, Ramírez-Tirado L-A, Báez-Saldaña R, Peña-Curiel O, SocaChafre G, Macedo-Perez E-O. Different mutation profiles and clinical characteristics among Hispanic patients with non-small cell lung cancer could explain the “Hispanic paradox”. Lung Cancer 2015;90(2):161–6. doi:10.1016/j.lungcan.2015.08.010. Asadzadeh Vostakolaei F, Karim-Kos HE, Janssen-Heijnen MLG, Visser O, Verbeek ALM, Kiemeney LALM. The validity of the mortality to incidence ratio as a proxy for site-specific cancer survival. Eur J Public Health 2011;21(5):573–7. doi:10.1093/eurpub/ckq120. Belani CP, Marts S, Schiller J, Socinski MA. Women and lung cancer: epidemiology, tumor biology, and emerging trends in clinical research. Lung Cancer 2007;55(1):15–23. doi:10.1016/j.lungcan.20 06.09.0 08. Ferguson MK, Skosey C, Hoffman PC, Golomb HM. Sex-associated differences in presentation and survival in patients with lung cancer. J Clin Oncol 1990;8(8):1402–7 http://www.ncbi.nlm.nih.gov/pubmed/ 2166143. Accessed July 182016. Hébert JR, Daguise VG, Hurley DM, et al. Mapping cancer mortality-toincidence ratios to illustrate racial and sex disparities in a high-risk population. Cancer 2009;115(11):2539–52. doi:10.1002/cncr.24270. Islami F, Torre LA, Jemal A. Global trends of lung cancer mortality and smoking prevalence. Transl Lung Cancer Res 2015;4(4):327–38. doi:10.3978/j.issn.2218-6751.2015.08.04. Lariscy JT, Hummer RA, Hayward MD. Hispanic older adult mortality in the United States: new estimates and an assessment of factors shaping the hispanic paradox. Demography 2015;52(1):1–14. doi:10.1007/s13524-014-0357-y. Lortet-Tieulent J, Goding Sauer A, Siegel RL, et al. State-level cancer mortality attributable to cigarette smoking in the United States. JAMA Int Med 2016;176(12):1792. doi:10.1001/jamainternmed.2016.6530. Markides KS, Eschbach K. Hispanic paradox in adult mortality in the United States. Netherlands: Springer; 2011. p. 227–40. doi: 10.1007/ 978- 90- 481- 9996- 9_11 Patel MI, Schupp CW, Gomez SL, Chang ET, Wakelee HA. How do social factors explain outcomes in non-small-cell lung cancer among Hispanics in California? Explaining the Hispanic paradox. J Clin Oncol 2013;31(28):3572–8. doi:10.1200/JCO.2012.48.6217. Pinheiro PS, Williams M, Miller EA, Easterday S, Moonie S, Trapido EJ. Cancer survival among latinos and the hispanic paradox. Cancer Causes Control 2011;22(4):553–61. doi:10.1007/s10552-011-9727-6. Ringer G, Smith JM, Engel AM, Pat Hendy M, Lang J. Influence of sex on lung cancer histology, stage, and survival in a Midwestern United States tumor registry. Clin Lung Cancer 2005;7(3):180–2. doi:10.3816/CLC.2005.n.033. Siegel RL, Fedewa SA, Miller KD, et al. Cancer statistics for Hispanics/Latinos, 2015. CA Cancer J Clin 2015;65(6):457–80. doi:10.3322/caac.21314. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin 2017. doi:10.3322/caac.21387. State Cancer Profiles. http://statecancerprofiles.cancer.gov/map/map. noimage.php. Accessed June 28, 2016. Sunkara V, Hébert JR. The colorectal cancer mortality-to-incidence ratio as an indicator of global cancer screening and care. Cancer 2015;121(10):1563–9. doi:10.1002/cncr.29228. Sunkara V, Hébert JR. The application of the mortality-to-incidence ratio for the evaluation of cancer care disparities globally. Cancer 2016;122(3):487–8. doi:10.1002/cncr.29746. Wagner SE, Hurley DM, Hébert JR, McNamara C, Bayakly AR, Vena JE. Cancer mortality-to-incidence ratios in Georgia: describing racial cancer disparities and potential geographic determinants. Cancer 2012;118(16):4032–45. doi:10.1002/cncr.26728.