Education and Stage of Cancer at - PowerPoint Presentation

1. Education and Stage of Cancer at Diagnosis-A Population-Based Study in South IndiaPreethi Sara George*Aleyamma Mathew* & Christopher Booth^*Division of Cancer Registry, Epidemiology, Biostatistics, Regional Cancer Center, Trivandrum, INDIA^Professor of Medical Oncology Canada Research Chair in Population Cancer Care,Clinician- Scientist, Division of Medical Oncology, Queen's University

2. BACKGROUNDSubstantial evidence indicates that the socioeconomic status (SES) of cancer patients has a significant impact on prognosis through its associated influence on the cancer stage at diagnosis.SES is also significantly associated with education level and occupation, both of which can greatly influence patients’ perception of the tumor, thereby affecting the level of early detection, diagnosis, and treatment.Limited reports in low and middle income countries (LMIC)We hypothesized that similar findings also existed across the whole of Population Based Cancer Registry Trivandrum.

3. OBJECTIVESTo evaluate the association between socio-demographic factors especially education and stage of cancer at diagnosis in Trivandrum District in South India. To evaluate the survival of stage at diagnosis and educational level on breast and cervical cancer in women and lung and oral cavity cancer in men .

4. METHODSStudy PopulationThe Population-Based Cancer Registry (PBCR) of Trivandrum District in South India [one of 29 cancer registries operating under the National Cancer Registry Program of India]Incident cases: Breast and Cervical cancer in women and oral cavity and lung cancer in men Period : 2012-2014. Study location: Regional Cancer Centre (RCC), Trivandrum Highlights54% urban and 46% ruralKerala -highest literacy rate (94% vs 73% national rate)Greatest life expectancy (74 years vs 64 years.Lowest infant mortality rate in IndiaPopulation:3.3 million

5. MethodologyActive case finding methodology Data collected from >60 hospitals and 7 pathology laboratories. Computerized information processing includes linkage of patient data and review of duplicate records. The validity of data are monitored by conducting data quality [Microscopic confirmation, Death Certificate Only (DCO), Fatality ratio and proportion of unknown primary sites] checks periodically.The major sources for incidence data are the RCC (63% of cases) and the Government Medical College Hospital (24% of cases) both located in Trivandrum. The Trivandrum PBCR employs 14 tumour registrars who were trained in cancer registration in locally and nationally organized courses and followed by continuing in-service training.

6. Definitions of Exposures and OutcomesSocio-demographic characteristics are captured routinely by PBCR staff. In this study the association between stage of disease and the following socio-demographic characteristics were considered: age, marital status, religion, rural/urban residence, and education. Self-reported educational levels include: illiterate (No education), up to primary school (up to 4 years of education), up to middle school (up to 7 years of education), up to secondary school (up to 10 years of education), college and technical school. Date of Diagnosis, Date of Death (if died)/ Date of Last Follow-up & Vital Status.

7. Statistical analysis Comparisons of proportions between socio-demographic groups were made using the Chi-square test. The Marascuilo procedure was used to compare multiple proportions.Cumulative Survival Probability determined using the Kaplan-Meier methodP-values of <0.05 were deemed statistically significant. No adjustments were made for multiple comparisons.

8. RESULTSCharacteristics of study populationMicroscopic verification: 82% in males and 87% in females. Mean age is 58 years and 31% of cases are >65 years of age. 83% married. 74% Hindus, 26% secondary school edn.Within the 4 cancer sub-groups, patients with breast cancer are younger, more likely to live in urban areas, and more highly educatedDistribution of incident cases of breast, cervical, lung, and oral cavity cancers 2012-2014 in Trivandrum District, Kerala

9. Socio-demographic characteristics all incident cases of breast, cervical, lung, and oral cavity cancer diagnosed 2012-2014 in Trivandrum District, South India.CharacteristicAll CasesBreast*CervixOral Cavity^Lung^N=4547N=2283N=481N=797N=986Age Mean (years)5855606063 <35 120 (3)91 (4)2 (<1)18 (2)9 (<1) 35-50 1207 (26)829 (36)108 (23)164 (21)106 (11) 51-64 1811 (40)893 (39)199 (41)315 (40)404 (41) 65+ 1409 (31)470 (21)172 (36)300 (38)467 (47)Marital Status Single 118 (3)61 (3)2 (<1)31 (4)24 (2) Married 3781 (83)1805 (79)343 (71)721 (91)912 (93) Widowed355 (8)230 (10)112 (23)8 (1)5 (<1) Divorced/Separated 70 (2)27 (1)13 (3)17 (2)13 (1) Unknown223 (5)160 (7)11 (2)20 (2)32 (3)Religion Hindu 3343 (73)1678 (73)372 (77)565 (71)728 (74) Muslim 541 (12)264 (12)42 (9)81 (10)154 (16) Christian 614 (14)314 (14)66 (14)147 (18)87 (9) Unknown49 (1)27 (1)1 (<1)4 (<1)17 (1)Place of residence Rural2864 (63)1345 (59)321 (67)517 (65)681 (69) Urban 1683 (37)938 (41)160 (33)280 (35)305 (31)Education Illiterate/Primary1011 (22)347 (15)184 (38)259 (33)221 (22) Middle/ Secondary852 (19)380 (17)115 (24)195 (24)162 (16) >Secondary school1186 (26)749 (33)88 (18)180 (23)169 (17)

10. Data capture rate for incident cases in Trivandrum cancer registry 2012-2014 BreastCervix uteriLungOral CavityTotal cases in PBCR2283481986797Cases from RCC1381380490617     No. (%) with stage*1353 (98%)365 (96%)481 (98%)608 (99%)     No. (%) with education level1373 (100%)377 (99%)485 (99%)611 (99%)Cases from Medical College53057381122     No. (%) with stage*409 (77%)38 (67%)350 (92%)77 (63%)     No. (%) with education level61   (12%)4 (7%)48 (13%)12 (10%)Cases from Other Sources3724411558     No. (%) with stage*130 (35%)20 (46%)82 (71%)20 (35%)     No. (%) with education level42 (12%)6 (14%)19 (17%)11 (19%)*Stage: TNM for breast and oral cavity, FIGO for cervix, SEER CED for lung Note: Lung and oral cavity are male only, Breast is female only63% (2868/4547) of cases identified from the RCC, 24% from Government Medical College Hospital (1090/4547), and 13% (589/4547) from other sources. Rates of data capture for stage and educational level was very high for cases at RCC (range 96-99% and 99-100% respectively). Among cases at Medical College the stage and educational capture rates were 62-92% and 7-12% respectively; corresponding figures from other sources were 35-71% and 12-19%.

11. Stage of cancer at diagnosisOnly 8-14% of cases presented with early stage (i.e. stage I/localized) disease. While 10% and 13% of cases with breast and cervical cancer presented with stage IV disease, the rate of metastatic disease at diagnosis was much higher for oral cavity (40%) and lung (46%) cancers. Overall, the proportion of cases with advanced disease (i.e. stage III/IV, or regional/metastatic) was 37% for breast, 39% for cervix, 67% of oral cavity, and 88% for lung cancer. Table 2. Stage at diagnosis for all incident cases of breast, cervical, lung, and oral cavity cancer diagnosed 2012-2014 in Trivandrum District, South India.Stage at Diagnosis#Breast*CervixOral Cavity^Lung^N=2283N=481N=797N=986 Stage I (%)172 (8)65 (14)77 (10)130 (13) Stage II (%)874 (38)175 (36)101 (13) Stage III (%)620 (27)120 (26)211 (27)331 (34) Stage IV (%)227 (10)63 (13)316 (40)453 (46) Unstaged (%)390 (17)58 (12)92 (12)72 (7)*Only female breast cancer, ^Only male lung and oral cavity cancer#Note: TNM stage is used for breast and oral cavity, FIGO stage is used for cervical cancer. SEER Clinical Extent of Disease (CED) is used for lung cancer (localized, regional, metastatic).

12. Socio-demographic factors and stage of cancerDistribution of stage at diagnosis by education level for all incident cases of breast, cervical, lung, and oral cavity cancer diagnosed 2012-2014 in Trivandrum District, South IndiaA) BREAST CANCER P Value-0.001B) CERVICAL CANCER P Value-0.009C) ORAL CANCER P Value-0.001D) LUNG CANCER P Value-0.182

13. Patients with advanced disease (stage III/IV or regional/metastatic)For breast, cervix, and oral cavity cancer there is a substantial association between stage at diagnosis and educational level. Among patients with breast cancer, the proportion of patients with advanced disease is 50%, 39%, 36% (p<0.001) for illiterate/primary school, middle school, and >secondary school respectively. Corresponding figures for cervix and oral cavity cancer are 46%, 43%, 24% (p=0.002) and 77%, 76%, 59% (p<0.001) respectively. No association for lung cancer patients with advanced disease. Illiterate/PrimaryMiddle/Secondary>SecondaryPBreast50% (172/347)39% (148/380)36% (269/749)<0.001Cervix46% (84/184)43% (50/115)24% (21/88)0.002OC77% (199/259)76% (148/195)59% (106/180)<0.001Lung89% (196/221)84% (136/162)88% (149/169)0.350*Defined as stage III or IV for breast, cervix, OC and CED regional/metastatic lung

14. Proportion of cases of breast, cervical, lung, and oral cavity cancer diagnosed 2012-2014 in Trivandrum District, South India with regional/metastatic disease at time of diagnosis.Proportion of Cases with Regional/Metastatic Disease*#Factors BreastCervix uteriOral Cavity Lung (n=2283)(n=481)(n=797)(n=986)N (%)PN (%)PN (%)PN (%)PAge <50 years346/818 (42)0.68750/95 (53)0.49367/158 (42)0.03576/91 (84)0.321 >50 years607/1465 (41)188/386 (49)331/639 (52)708/895 (79)Marital Status Not married 25/61 (41)0.7751/2 (50)0.96416/31 (50)0.91422/24 (92)0.169 Married 773/1805 (43)166/343 (48)365/721 (51)734/912 (81)Religion Hindu 693/1678 (41)182/372 (49)0.475275/565 (49)585/728 (80)0.869 Muslim 116/264 (44)0.65319/42 (45)37/81 (46)0.100121/154 (79) Christian 135/314 (43)37/66 (56)85/147 (58)69/87 (79)Residence Rural588/1345 (44)0.021159/321 (50)0.974267/517 (52)0.190539/681 (79)0.671 Urban 365/938 (39)79/160 (49)131/280 (47)245/305 (80)Education Illiterate/ Primary172/347 (50)<0.00184/184 (46)0.002199/259 (77)0.001196/221(89)0.350 Middle / Secondary 148/380 (39)50/115 (44)148/195 (76)136/162 (84) > Secondary269/749 (36)21/88 (48)106/180 (59)149/169 (88)*Only female breast cancer; ^Only male lung and oral cavity cancer#Defined as stage III/IV disease for breast, cervix, oral cavity cancer and CED regional/metastatic for lung cancer

15. (A)Female Breast Cancer (B) Cervix uterine Cancer Education Breast cancer Cervix cancerIlliterate /primary57.7% (2.8%)49.5%( 3.9%)Middle65.9% (2.5%)50.4% (4.9%)College & above72.3% (1.7%)57.9%( 5.7%)P value0.00010.1282Five year Survival Probability % ( S.E%) Based on Education and Site of cancer

16. (C) Male Lung Cancer (D) Male Oral Cancer Education Male - Lung Male -Oral cavityIlliterate /primary1.0 % (0.07%)40.5% (3.3%)Middle5.0% ( 1.8%)34.2% (3.5%)College & above3.0% (1.6%)45.8% (3.8%)P value0.12090.0175Five year Survival Probability % ( S.E%) Based on Education and Site of cancer

17. SUMMARY AND CONCLUSIONThe proportion of patients diagnosed with advanced stage disease in the general population of India is far higher than HICs. A strong association between educational status and stage of breast, cervical and oral cavity cancers at diagnosis.8% of women in our study with breast cancer had stage I disease and 37% had stage III/IV disease; this is consistent with other reports from India .However, these figures are in stark contrast to HICs where 30-45% of women have stage I disease at diagnosis and 8-22% have stage III/IV disease. Likewise, we report that 14% and 39% of women with cervical cancer had stage I and stage III/IV disease respectively; these data are consistent with another study from India. Corresponding figures in Canada 34% of women have stage I and 19% of women have stage III/IV disease at diagnosis. These striking differences in stage of cancer at diagnosis between India and HICs likely reflect differences in cancer awareness/education and access to cancer care.This has important policy and public health implications.

18. The primary objective of our study was to evaluate the extent to which stage of cancer at diagnosis was associated with educational status which was chosen as a surrogate for socioeconomic status. Parallel work in HICs commonly use neighborhood median household income from census data as a measure for SES.Comparable census data in South India were lacking and we had concerns regarding the validity of self-reported household income in the Indian context. For these reasons we elected to use educational level which is commonly used in epidemiologic studies as a surrogate for SES .Public health studies in India have demonstrated that educational level in India is correlated with other measures of socioeconomic status including occupation, housing and social status.

19. Our study results show a substantial association between low educational status and advanced cancer at diagnosis for breast, cervical and oral cavity cancers. This association was not seen in lung cancer. These results have “face validity” when one considers the mechanism by which educational status may be related to stage of disease. Patients with greater education who develop early signs of breast cancer (i.e. a breast lump), cervical cancer (i.e. abnormal vaginal bleeding), and oral cancer (i.e. a mouth lesion) may have greater awareness and access to health care, thereby seeking earlier medical treatment and being diagnosed with earlier disease than patients from lower educational backgrounds. However, for patients with lung cancer, by the time symptoms manifest, the disease is often already very advanced in which case early treatment-seeking behavior would be expected to have less effect on stage at diagnosis.

20. Explored survival differences across socioeconomic groups and the relative extent to which stage and treatment delivery explain any observed survival differences. Several hospital-based studies in India have explored the association between educational level and stage of cancer at diagnosis.Two studies have used the Hospital Based Cancer Registry at the Regional Cancer Centre (RCC) in Trivandrum, Kerala to explore this issue. Among 522 patients with breast cancer seen at RCC in 2006, Ali et al found that advanced stage of disease was more common among women who were not married and who had lower education.

21. Our study does have methodologic limitations that warrant comment. Stage and educational level were more likely to be missing among those patients treated at the Government Medical College or other smaller institutions compared to the Regional Cancer Centre in Trivandrum. This may limit the generalizability of our study results. Moreover, our study cannot explain why patients with lower educational status are more likely to have advanced cancer. It is likely that this reflects delay in seeking medical attention but future qualitative work will be needed to explore this more fully in order to use this knowledge to improve care of patients from impoverished backgrounds.

22.