CRVAW Faculty Journal Articles


To determine predictors of cervical cancer survival by socioeconomic status (SES), urbanization, race/ethnicity, comorbid conditions, and treatment among elderly Medicare-eligible women whose conditions were diagnosed with cervical cancer in a multiethnic population.

Methods: A total of 538 women with cervical cancer aged 65 years or older were identified from 1999 to 2001 from the Texas Cancer Registry and were linked with the state Medicare data and Texas Vital Records to determine survival times. All women had similar access to care through Medicare fee-for-services insurance. A composite measure of SES was created using census tract-level data as was urbanization. Treatment and comorbid conditions were available from the Medicare data. Cox proportional hazards modeling was used for all-cause and cervical cancer-specific survival analysis.

Results: Increased age (P < 0.0001) and advanced tumor stage (P < 0.0001) were associated with poorer all-cause and cervical cancer-specific survival. Having a comorbid condition was associated with all-cause survival (P < 0.01) but not cervical cancer-specific mortality. After adjusting for confounders, women receiving some form of treatment were almost half as likely to die with cervical cancer (adjusted hazard ratio = 0.68; 95% confidence interval, 0.52-0.89). After adjustment for all confounders, Hispanic women consistently had lower all-cause and cervical cancer-specific mortality rates relative to non-Hispanic white and non-Hispanic black women.

Conclusions: Among women with similar health care coverage, Hispanic women had consistently lower all-cause and cervical cancer-specific mortality rates than other older women whose conditions were diagnosed with this disease in Texas. The presence of comorbid conditions and treatment were important predictors of survival, yet these factors do not explain the survival advantage for Hispanic women.

Document Type


Publication Date


Notes/Citation Information

This manuscript provided with permission from the publisher, and also accessible through the journal's website at

Digital Object Identifier (DOI)