The_AT Role_NN1 of_IO Poverty_NN1 Rate_NN1 and_CC Racial_JJ Distribution_NN1 in_II the_AT Geographic_JJ Clustering_NN1 of_IO Breast_NN1 Cancer_NN1 Survival_NN1 Among_II Older_JJR Women_NN2 :_: A_ZZ1 Geographic_JJ and_CC Multilevel_NP1 Analysis_NN1 About_RG 60%_NNU of_IO breast_NN1 cancer_NN1 deaths_NN2 occur_VV0 in_II women_NN2 aged_II 65_MC years_NNT2 or_CC older_JJR (_( 1_MC1 )_) ._. 
Breast_NN1 cancer_NN1 screening_NN1 and_CC more_RGR effective_JJ therapies_NN2 have_VH0 combined_VVN to_TO improve_VVI breast_NN1 cancer_NN1 survival_NN1 ,_, and_CC an_AT1 estimated_JJ 1_MC1 million_NNO women_NN2 aged_II 65_MC years_NNT2 or_CC older_JJR are_VBR currently_RR living_VVG with_IW breast_NN1 cancer_NN1 (_( 2_MC )_) ,_, a_AT1 number_NN1 that_CST is_VBZ expected_VVN to_TO increase_VVI over_II time_NNT1 as_II the_AT baby_NN1 boomer_NN1 generation_NN1 ages_NN2 ._. 
It_PPH1 is_VBZ well_RR established_VVN that_CST patient_JJ characteristics_NN2 ,_, tumor-related_JJ factors_NN2 ,_, and_CC type_NN1 of_IO treatment_NN1 received_VVD affect_NN1 breast_NN1 cancer_NN1 survival_NN1 (_( 3_MC ,_, 4_MC )_) ._. 
In_II31 addition_II32 to_II33 these_DD2 individual-level_JJ factors_NN2 ,_, there_EX has_VHZ been_VBN increasing_JJ interest_NN1 in_II the_AT extent_NN1 to_II which_DDQ area-level_JJ determinants_NN2 (_( e.g._REX ,_, racial_JJ distribution_NN1 ,_, poverty_NN1 rate_NN1 )_) influence_VV0 breast_NN1 cancer-related_NN1 behavior_NN1 and_CC outcomes_NN2 ,_, including_II breast_NN1 cancer_NN1 screening_NN1 ,_, incidence_NN1 ,_, stage_NN1 at_II diagnosis_NN1 ,_, and_CC mortality_NN1 (_( 5-9_MCMC )_) ._. 
The_AT geographic_JJ variation_NN1 in_II these_DD2 individual-level_JJ and_CC area-level_JJ characteristics_NN2 may_VM contribute_VVI to_II geographic_JJ disparities_NN2 in_II breast_NN1 cancer_NN1 survival_NN1 that_CST appear_VV0 to_TO exist_VVI in_II Europe_NP1 and_CC in_II the_AT United_NP1 States_NP1 (_( 10-16_MCMC )_) ._. 
Identification_NN1 of_IO reasons_NN2 for_IF disparities_NN2 in_II small-area_JJ variation_NN1 in_II breast_NN1 cancer_NN1 survival_NN1 will_VM allow_VVI for_IF local_JJ implementation_NN1 of_IO evidence-based_JJ approaches_NN2 according_II21 to_II22 clinical_JJ and_CC community_NN1 guidelines_NN2 (_( 17_MC ,_, 18_MC )_) in_II an_AT1 effort_NN1 to_TO reduce_VVI such_DA disparities_NN2 ._. 
The_AT purpose_NN1 of_IO this_DD1 study_NN1 was_VBDZ to_TO examine_VVI small-area_JJ geographic_JJ variation_NN1 in_II breast_NN1 cancer_NN1 survival_NN1 among_II elderly_JJ women_NN2 residing_VVG in_II 5_MC urban_JJ areas_NN2 in_II the_AT United_NP1 States_NP1 ._. 
The_AT study_NN1 of_IO the_AT effect_NN1 of_IO area-level_JJ conditions_NN2 on_II breast_NN1 cancer_NN1 survival_NN1 is_VBZ especially_RR important_JJ for_IF older_JJR populations_NN2 ,_, because_CS they_PPHS2 may_VM have_VHI had_VHN longer_JJR exposure_NN1 to_II adverse_JJ neighborhood_NN1 physical_JJ and_CC psychosocial_JJ stressors_NN2 and_CC have_VH0 a_AT1 greater_JJR need_NN1 for_IF proximity_NN1 to_II health_NN1 care_NN1 ,_, food_NN1 ,_, and_CC other_JJ resources_NN2 and_CC services_NN2 ._. 
Older_JJR adults_NN2 are_VBR vulnerable_JJ to_II adverse_JJ neighborhood_NN1 conditions_NN2 ,_, with_IW negative_JJ effects_NN2 on_II both_DB2 biologic_JJ and_CC psychologic_JJ outcomes_NN2 (_( 19_MC )_) ._. 
In_RR21 addition_RR22 ,_, we_PPIS2 examined_VVD the_AT role_NN1 that_CST patient_JJ factors_NN2 ,_, type_NN1 of_IO treatment_NN1 received_VVD ,_, tumor_NN1 characteristics_NN2 ,_, utilization_NN1 of_IO medical_JJ care_NN1 ,_, mammography_NN1 use_NN1 ,_, and_CC 2_MC area-level_JJ factors_NN2 (_( census-tract_JJ percent_NNU African_JJ American_NN1 as_II a_AT1 measure_NN1 of_IO racial_JJ segregation_NN1 and_CC census-tract_JJ poverty_NN1 rate_NN1 as_II a_AT1 measure_NN1 of_IO economic_JJ segregation_NN1 )_) played_VVD in_II explaining_VVG any_DD geographic_JJ variation_NN1 that_CST may_VM exist_VVI ._. 
MATERIALS_NN2 AND_CC METHODS_NN2 Sample_VV0 selection_NN1 The_AT sample_NN1 for_IF this_DD1 study_NN1 was_VBDZ obtained_VVN from_II a_AT1 database_NN1 that_CST links_VVZ data_NN from_II the_AT 1992-1999_MCMC National_JJ Cancer_NN1 Institute_NN1 's_GE Surveillance_NN1 ,_, Epidemiology_NP1 ,_, and_CC End_VV0 Results_NN2 (_( SEER_NN1 )_) program_NN1 with_IW 1991-1999_MCMC Medicare_NP1 claims_VVZ files_NN2 from_II the_AT Centers_NN2 for_IF Medicare_NP1 and_CC Medicaid_NP1 (_( 20_MC )_) ,_, which_DDQ allowed_VVD us_PPIO2 to_TO obtain_VVI patients_NN2 '_GE comorbidity_NN1 data_NN at_RR21 least_RR22 1_MC1 year_NNT1 prior_II21 to_II22 their_APPGE breast_NN1 cancer_NN1 diagnosis_NN1 ._. 
Ninety-four_MC percent_NNU of_IO cancer_NN1 patients_NN2 reported_VVN to_II SEER_NN1 aged_II 65_MC years_NNT2 or_CC older_JJR were_VBDR successfully_RR matched_VVN to_II the_AT Medicare_NP1 data_NN (_( 20_MC )_) ._. 
We_PPIS2 used_VVD data_NN from_II the_AT metropolitan_JJ areas_NN2 of_IO Atlanta_NP1 ,_, Georgia_NP1 ,_, Detroit_NP1 ,_, Michigan_NP1 ,_, San_NP1 Francisco-Oakland_NP1 ,_, California_NP1 ,_, Seattle-Puget_NP1 Sound_NN1 ,_, Washington_NP1 ,_, and_CC the_AT state_NN1 of_IO Connecticut_NP1 ._. 
In_II the_AT data_NN ,_, a_AT1 first_MD primary_JJ in-situ_NN1 or_CC invasive_JJ breast_NN1 cancer_NN1 was_VBDZ diagnosed_VVN in_II 37,473_MC women_NN2 from_II 1992_MC to_II 1999_MC ._. 
We_PPIS2 excluded_VVD 9,537_MC women_NN2 who_PNQS 1_MC1 )_) were_VBDR enrolled_VVN in_II a_AT1 health_NN1 maintenance_NN1 organization_NN1 at_II any_DD point_NN1 during_II the_AT 1991-1999_MCMC study_NN1 period_NN1 ,_, because_CS claims_NN2 data_NN about_II key_JJ prognostic_JJ variables_NN2 would_VM not_XX be_VBI available_JJ ;_; 2_MC )_) were_VBDR not_XX covered_VVN by_II Medicare_NP1 Parts_NN2 A_ZZ1 and_CC B_ZZ1 between_II the_AT first_MD primary_JJ breast_NN1 cancer_NN1 diagnosis_NN1 and_CC the_AT study_NN1 end_NN1 point_NN1 (_( date_NN1 of_IO death_NN1 or_CC December_NPM1 31_MC ,_, 1999_MC )_) ;_; 3_MC )_) were_VBDR identified_VVN by_II death_NN1 certificate_NN1 only_RR because_CS survival_NN1 time_NNT1 can_VM not_XX be_VBI calculated_VVN ;_; 4_MC )_) had_VHD a_AT1 bilateral_JJ mastectomy_NN1 ;_; and_CC 5_MC )_) were_VBDR aged_II 65_MC years_NNT2 at_II diagnosis_NN1 in_BCL21 order_BCL22 to_TO obtain_VVI comorbidity_NN1 data_NN from_II Medicare_NP1 during_II the_AT year_NNT1 before_II their_APPGE breast_NN1 cancer_NN1 diagnosis_NN1 because_CS Medicare_NP1 data_NN are_VBR not_XX available_JJ prior_II21 to_II22 the_AT age_NN1 of_IO 65_MC years_NNT2 ._. 
Medicare_NP1 Part_NN1 A_ZZ1 covers_VVZ inpatient_NN1 hospitalization_NN1 ,_, skilled_JJ nursing_JJ facility_NN1 care_NN1 ,_, and_CC hospice_NN1 care_NN1 ,_, while_CS Part_NN1 B_ZZ1 covers_VVZ both_RR inpatient_NN1 and_CC outpatient_NN1 medical_JJ services_NN2 ,_, as_II31 well_II32 as_II33 outpatient_NN1 therapies_NN2 ,_, limited_JJ medical_JJ supplies_NN2 and_CC medical_JJ tests_NN2 ,_, and_CC some_DD durable_JJ medical_JJ equipment_NN1 ._. 
This_DD1 left_VVD 27,936_MC patients_NN2 available_JJ for_IF the_AT remainder_NN1 of_IO the_AT study_NN1 ._. 
Women_NN2 who_PNQS were_VBDR included_VVN in_II the_AT analysis_NN1 were_VBDR statistically_RR more_RGR likely_JJ to_TO be_VBI diagnosed_VVN at_II an_AT1 earlier_JJR stage_NN1 and_CC to_TO have_VHI a_AT1 lower_JJR tumor_NN1 grade_NN1 than_CSN those_DD2 excluded_VVN ._. 
In_RR21 addition_RR22 ,_, women_NN2 who_PNQS were_VBDR included_VVN were_VBDR significantly_RR less_RGR likely_JJ to_TO be_VBI of_IO "_" other_JJ "_" race_NN1 ,_, to_TO have_VHI surgery_NN1 ,_, and_CC to_TO have_VHI radiation_NN1 therapy_NN1 ._. 
Differences_NN2 in_II 
