SURVIVAL OF U.S. BLACK AND WHITE PATIENTS: PATIENTS AND METHODS

Posted by James

Since 1992, 11 SEER areas (Atlanta; Connecticut; Detroit; Hawaii; Iowa; New Mexico; San Francisco-Oakland, CA; Los Angeles; San Jose-Monterey, CA; Seattle; and Utah) have covered about 14% of the U.S. population. All patients diagnosed in 1992-1998 with invasive esophageal cancer as the first or only reportable cancer were identified, excluding small numbers ascertained only by death certificate or autopsy; diagnoses in 1999 were excluded, due to limited follow-up (i.e., through 1999). Histologic groups were defined by ICD-O-2 morphology codes. The original sample included 1,310 black and 6,245 white patients. Because of the very small numbers of black patients with adenocarcinoma of the esophagus (N=81) (known to be uncommon in U.S. blacks) or with histologic type other than squamous cell carcinoma (N=104), only patients with squamous cell carcinoma (ICD-0-2 morphology codes 8070-8077) were included (1,125 blacks and 2,392 whites). SEER registries report site-specific extent-of-disease codes for all tumors to SEER, which uses these codes to assign SEER historical stage. Extent of disease includes a code for extension of the tumor and a separate code for involvement of lymph nodes. SEER historical stage includes: localized (confined to the esophagus, with no evidence of spread to surrounding organs/tissues or no regional lymph nodes); regional (invasion beyond the organ to surrounding organs/tissues and/or to regional lymph nodes); dis-tant/metastatic (spread to remote organs/tissues directly or by discontinuous metastasis); and unknown. Data in SEER reports have included the five-year relative survival rate (RSR)—or ratio of the observed survival rate to the expected survival rate (derived from mortality rates in the U.S. popu lation)—by SEER stage at diagnosis, showing lower RSR for blacks versus whites at localized stage for esophageal cancer; histologic category was not considered. The present study examined RSRs (at one, three, and five years after diagnosis), using a computer program that includes mortality rates for the general U.S. population (by age, sex, and race), for squamous cell carcinoma by stage and for certain subgroups within localized stage. Confidence intervals (CIs) on RSRs were estimated +/- twice the standard error.

Using Cox proportional hazards regression, relative risk for death (hazard ratio) was estimated for blacks relative to whites, controlling for the sociodemographic and clinical characteristics. Analyses of localized-stage patients also included lymph-node status (negative versus unknown, based on clinical and/or pathological assessment), and presence or absence of pathologic review of regional lymph node(s) as an indicator of potential “understaging” (i.e., underestimation of the extent or spread of cancer at diagnosis).
canadian pharmacy cialis

Survival (in months) was estimated from month of diagnosis until death, loss to follow-up, or the end of 1999, and was truncated at five years. Relative risks (hazards) for death from any cause were obtained for models, including the variables listed above; 95% confidence limits (CIs) on relative risks were based on the normal distribution. The assumption of proportional hazards was checked by graphing the log of the negative log of the survival function for each racial group, with race as the stratum; roughly parallel lines for the two races over time were obtained, supporting proportionality.