Ethnic variations in diagnosis, treatment, socioeconomic status, and survival in a large population-based cohort of elderly patients with non-Hodgkin lymphoma




There is a lack of research on ethnic disparities in survival among patients with non-Hodgkin lymphoma (NHL), although these disparities have been documented for patients with many other tumors.


A retrospective cohort of 13,321 patients diagnosed with incident NHL at age ≥65 years from 1992 to 1999 were identified from the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database with 11 years of follow-up. Of these patients, 11,868 were Caucasians, 533 were African Americans, and 920 were other ethnicities. A time-to-event Cox regression model was used to examine the relative risk of all-cause and disease-specific mortality.


A larger proportion (72.2%) of African Americans were in the poorest quartile of socioeconomic status as measured by the poverty level compared with 21.8% of Caucasians, and 43.2% of African Americans received chemotherapy compared with 52.4% of Caucasians (P < .01). Hazard ratio of all-cause and NHL-specific mortality increased significantly with age, advanced stage, higher comorbidity scores, and poorer socioeconomic status. Patients receiving either chemotherapy or radiotherapy or both were significantly less likely to die. After taking into account differences in treatment, comorbidity, and socioeconomic status, there was no statistically significant difference in the risk of all-cause (hazard ratio, 0.97; 95% confidence interval, 0.88-1.08) and disease-specific mortality (hazard ratio, 1.07; 95% confidence interval, 0.92-1.25) between African Americans and Caucasians.


The risk of mortality in patients with NHL was associated with socioeconomic status and was reduced in patients receiving chemotherapy. No significant differences in the risk of mortality were observed between African Americans and Caucasians after controlling for factors such as treatment and socioeconomic status. Cancer 2008. © 2008 American Cancer Society.

Non-Hodgkin lymphoma (NHL) is a common tumor in the elderly population, with a median age at diagnosis of 67 years.1, 2 It is estimated that over 63,000 men and women will be diagnosed with NHL each year, and about 18,660 men and women will die of NHL with a median age at death of 74 years in the United States.1, 2 The incidence of NHL has been increasing over the last several decades worldwide,1–7 and Caucasians have higher incidence and mortality rates of NHL than patients in other ethnic groups.1, 2, 5 During this period of time, treatment has improved, particularly with combination chemotherapy CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone)8–16 and later with the addition of rituximab to CHOP for B-cell NHL since 1998.17–20 Although there were several studies that demonstrated ethnic disparities in receiving these treatments,21–23 none of these studies has specifically addressed ethnic disparities in survival and whether these survival differences were due to disparities in treatment, diagnosis, or socioeconomic status (SES) or to other factors in patients with NHL.


Data Sources

This study used the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked data for patients diagnosed from 1992 to 1999. The method of linking these data has been described elsewhere.24, 25 The SEER program, supported by the National Cancer Institute, consists of population-based tumor registries that collect standardized information on all cancers diagnosed within the specified geographic regions.24–27 The metropolitan areas of San Francisco/Oakland, Detroit, Atlanta, and Seattle; Los Angeles County; the San Jose–Monterey area; and the states of Connecticut, Iowa, New Mexico, Utah, and Hawaii were included in this report. The Medicare component, conversely, is the primary health insurer for the great majority of the US population ≥65 years of age. It covers hospital, physician, and other medical services for >97% of persons aged ≥65 years.24, 25 The Committee for Protection of Human Subjects at the University of Texas Health Science Center at Houston approved this study.

Study Population

The retrospective cohort consisted of 13,570 patients diagnosed with incident NHL at age ≥65 from 1992 to 1999, and identified from the SEER-Medicare linked database. A total of 249 cases were excluded because of missing information on SES, leaving 13,321 subjects in the final analysis. Of 13,321, 11,868 were Caucasians (non-Hispanic), 533 were African-Americans (non-Hispanic), and 920 were other ethnicities.

Study Variables

Survival variables

The vital status, cause of death, and date of death in SEER data were obtained from the National Center for Health Statistics by the National Cancer Institute.27 Survival time in months was calculated from the date of diagnosis to the date of death or date of last follow-up (December 31, 2002). All-cause mortality was defined as death by any underlying cause. Patients still alive at the last date of follow-up were censored. NHL-specific mortality was defined as mortality with lymphoma as the underlying cause of death; patients who died of causes other than lymphoma were censored.

Socioeconomic status

Three variables from the 1990 census available in the SEER-Medicare linked data were used to define SES at the census tract level and were categorized into quartiles: 1) education—percent of adults aged ≥25 who had <12 years of education, divided into first (0%-10.89%), second (10.91%-17.94%), third (17.95%-25.80%), and fourth quartiles (≥25.81%); 2) poverty—percent of persons living below the poverty line, divided into first (≤3.70%), second (3.71%-6.89%), third (6.90%-11.98%), and fourth quartiles (≥11.99%); 3) income—median annual household income (in US dollars) was divided into first (≥$45,427), second ($35,178-45,424), third ($26,894-35,174), and fourth quartiles (≤$26,882). In addition, a composite socioeconomic variable combining these 3 variables with equal weights was created and categorized into quartiles.28


Information on chemotherapy was identified from Medicare claims, which was described elsewhere.29, 30 In brief, the following Medicare claim codes were used for defining chemotherapy: International Classification of Disease (ICD)-9-CM procedure code 9925 for injection or infusion of cancer chemotherapeutic substance, the Common Procedure Terminology (CPT) codes 96,400-96,549, J8510, J8520, J8521, J8530-J8999, J9000-J9999, and Q0083-Q0085 for chemotherapy administration, revenue center codes 0331 (chemotherapy injected), 0332 (chemotherapy oral), and 0335 (chemotherapy intravenous), and the ICD-9-CM V codes of V58.1, V66.2, or V67.2 for follow-up examination or care after chemotherapy. Because only 23 (0.2%) patients received stem cell transplantation (ICD-9-CM codes 4100-4109, or CPT codes 38,230, 38,231, 38,240, or 38,241)31, 32 within 12 months of diagnosis, this variable was not included in the final analysis.

Radiation Therapy

Because the information on radiation therapy from SEER and Medicare complement each other,33 cases were defined as having received radiation therapy if SEER data or Medicare claims data indicated so.

Tumor Factors

Tumor stage as defined in the SEER Extension of Disease-88, third edition27 was included. The Ann Arbor staging system was used and reported as follows: stage I, involvement of a single lymph node region, or localized involvement of a single extralymphatic organ or site; stage II, involvement of 2 or more lymph node regions on the same side of the diaphragm, or localized involvement of a single extralymphatic organ or site and its regional lymph nodes on the same side of the diaphragm with or without involvement of other lymph node regions on the same side of the diaphragm; stage III, involvement of lymph node regions on both sides of the diaphragm, with or without localized involvement of an extralymphatic organ or site or spleen involvement; stage IV, disseminated or multifocal involvement of 1 or more extralymphatic organs; and unknown stage. The Extent of Disease codes from SEER data were: stage I = 00-17, II = 20-29, III = 30-78, IV = 80-90, and unstaged = 99. Because of heterogeneities of different histologic subtypes of NHL, analysis was stratified by several major types of NHL according to the methods by Morton and colleagues5, 34, 35: diffuse large B-cell lymphoma (histology codes 9675, 9678-9684), chronic lymphocytic leukemia (9670), follicular lymphoma (9690-9698), and all other subtypes or unknown.

Comorbidity Scores

Comorbidity was ascertained from Medicare claims data through diagnoses or procedures made 1 year before and 1 month after the diagnosis of NHL. The Medicare inpatient, outpatient, and physician claims were used to create a comorbidity score. The detailed method has been discussed elsewhere.36 The rationale for including diagnoses from the physician claims was that many more patients saw a physician rather than being hospitalized, thus increasing the possibility of identifying more comorbid conditions.36 This score reflects the Deyo adaptation of the Charlson comorbidity index,37, 38 with several procedure codes that reflect the Romano adaptation.39 The comorbidity score was then categorized into 0, 1, 2, or ≥3.

Other Characteristics

Other characteristics obtained from the SEER data include age at diagnosis, race/ethnicity, marital status, year of diagnosis, geographic area, tumor stage, and nodal site. Tumor variables were well validated in SEER registries,24, 25 and so was the information on marital status, year of diagnosis, and geographic area. Although race/ethnicity categories may have involved a certain degree of misclassification, 2 categories (African American and Caucasian) in this study are likely to have a minimal misclassification bias, if any.40


The differences in the distribution of baseline characteristics among the 3 ethnic groups were tested using the chi-square statistic. Cox proportional hazard regression model was used for analysis of survival using the PHREG procedure in SAS (SAS Institute Inc., Cary, NC).41 The proportionality assumption was considered to be satisfied when the log-log Kaplan-Meier curves for survival functions by race/ethnicity or SES were parallel and did not intersect. The median survival time in months was obtained from the LIFETEST procedure, which is based on the Kaplan-Meier method. The interaction between race/ethnicity and SES was tested using the product term of these 2 variables in the model. Variables that were adjusted for in the model include age, sex, race, tumor stage, nodal site, radiation, comorbidity score, geographic area, and year of diagnosis.


Table 1 presents the distribution of patient demographic and tumor characteristics across the ethnic groups of patients with NHL. A higher proportion of cases were diagnosed at age 65 years to 69 years in African Americans (24.8%) and in other ethnic groups (22.4%) than in Caucasians (19.5%). The mean age was 76 years for Caucasians, 74 years for African Americans, and 75 years for others. There were slightly more female than male patients, and a larger proportion (58.2%) of African American patients were unmarried. A slightly greater proportion of cases was diagnosed at an early stage and had a lower comorbidity score among Caucasians and other ethnic groups than among African Americans. The percentage of patients receiving chemotherapy and radiation therapy was lower among African Americans. For example, 43.2% of African Americans received chemotherapy compared with 52.4% of Caucasians and 54.7% of others (P < .001). Similarly, 18.2% of African Americans received radiation therapy compared with 24.3% of Caucasians and 26.6% of others (P < .001). The proportion of cases was stable over 10 years among these 3 ethnic groups.

Table 1. Comparison of Characteristics Among Different Ethnic Groups in Patients With Non-Hodgkin Lymphoma
CharacteristicsCaucasiansAfrican AmericansOthersP by Ethnicity
Median age [range]76 [65-104]  74 [65-100] 75 [65-100] 
Age      <.001
Sex      .011
Marital status      <.001
Tumor stage      <.001
Lymph node site      <.001
Chemotherapy      <.001
Radiation therapy      .001
Comorbidity score      <.001
Year of diagnosis      .105

Table 2 presents the distribution of SES among the 3 racial/ethnic groups of patients. A larger proportion of African Americans were in the poorest quartiles of education, poverty, income, and composite SES compared with Caucasians. For example, 72.2% of African Americans and 38.7% of others were in the poorest quartile of SES as measured by the poverty level compared with 21.8% of Caucasians (P < .001). The differences in these factors in Tables 1 and 2 between Caucasians and African Americans were all statistically significant except for sex and year of diagnosis. The tests for trend across ethnic groups by age, SES, comorbidity, and year of diagnosis were statistically significant at P < .01, but not for stage (P = .425).

Table 2. Comparison of Socioeconomic Status Among Different Racial/Ethnic Groups in Patients With Non-Hodgkin Lymphoma
Socioeconomic Status Quartile From High to LowCaucasiansAfrican AmericansOthers
Composite socioeconomic status      

Table 3 presents the 1-, 3-, and 5-year observed survival rates by racial/ethnic groups and SES, stratified by tumor stage. Since the last date of follow-up in Medicare was December 2002, all cases in our study cohort had been followed for at least 3 years. The 3-year observed survival for stage I-II from all causes was 55.8% in Caucasians, 52.6% in African Americans, and 44.7% in others. Survival increased with higher SES. For example, the 3-year observed survival was 59.9% in subjects who lived in the community with the highest quartile of composite SES, and 48.8% in those with the lowest quartile. This survival pattern in association with race/ethnicity and SES was similar at the 1- and 5-year level and for NHL-specific survival (Table 3). Median survival time was 32.5 months (95% confidence interval [CI], 31.0-34.8) for patients receiving chemotherapy and 11.9 months (95% CI, 1.4-1.8) for those who did not receive chemotherapy.

Table 3. All-Cause and Lymphoma-specific Survival in Patients With Non-Hodgkin Lymphoma, by Tumor Stage, Race/Ethnicity, and Socioeconomic Status
Race/Ethnicity and Socioeconomic Status1-Year Survival, Cases in 1992-1999, n=13,321, %3-Year Survival, Cases in 1992-1999, n=13,321, %5-Year Survival, Cases in 1992-1997, n=9962, %
All CauseNHL SpecificAll CauseNHL SpecificAll CauseNHL Specific
  1. NHL indicates non-Hodgkin lymphoma.

Tumor stage I-II      
 Racial/ethnic groups      
  African Americans68.484.752.677.936.268.1
 Socioeconomic status, quartile      
  1st, high75.686.259.979.550.170.9
  4th, low65.881.048.874.637.065.6
 Patients with stage I-II71.284.054.976.943.768.4
Tumor stage III-IV      
 Racial/ethnic groups      
  African Americans43.866.027.356.921.248.2
 Socioeconomic status, quartile      
  1st, high58.874.940.366.328.555.6
  4th, low49.769.831.361.220.950.4
 Patients with stage III-IV53.871.335.763.025.051.9
All tumor stages      
 Racial/ethnic groups      
  African Americans54.073.437.765.127.955.6
 Socioeconomic status, quartile      
  1st, high66.880.749.673.038.863.3
  4th, low56.974.839.167.127.756.9
 Patients with all stages61.677.244.669.733.859.8

Table 4 presents the percentage of patients receiving chemotherapy and radiation therapy by stage, ethnicity, and SES, and the odds ratio of receiving these therapies while controlling for other confounding factors such as age, sex, marital status, comorbidity, year of diagnosis, and SEER regions. For patients with both early stage (I-II) and late stage (III-IV), African American patients were significantly less likely to receive chemotherapy (odds ratio, 0.69; 95% CI, 0.50-0.95 for stage I-II), and other ethnic groups were slightly but not significantly more likely to receive this therapy (odds ratio, 1.16; 95% CI, 0.92-1.47), compared with Caucasians. However, there were no significant differences in receiving radiation therapy among these ethnic groups. There were also no significant variations by SES.

Table 4. Receipt of Chemotherapy and Radiation Therapy in Patients With Non-Hodgkin Lymphoma by Stage, Race/Ethnicity, and Socioeconomic Status
Race, Socioeconomic Status, and StageNo. of Patients (N=13,321)Receipt of ChemotherapyReceipt of Radiotherapy
%Odds Ratio* (95% CI)%Odds Ratio* (95% CI)
  • CI indicates confidence interval.

  • *

    Odds ratio is adjusted for sex, age, marital status, socioeconomic status, stage, site, comorbidity, year of diagnosis, and Surveillance, Epidemiology, and End Results region.

Stage I-II     
 Racial/ethnic groups     
  African Americans19040.00.69 (0.50-0.95)25.30.72 (0.50-1.02)
  Others43249.51.16 (0.92-1.47)39.61.13 (0.89-1.44)
 Socioeconomic status, quartile     
  1st, high155451.11.0036.51.00
  2nd147648.40.93 (0.80-1.09)37.71.02 (0.87-1.18)
  3rd148252.21.07 (0.92-1.26)35.80.88 (0.75-1.04)
  4th, low143350.20.98 (0.83-1.15)31.10.75 (0.63-0.89)
  II191961.21.79 (1.59-2.02)26.00.54 (0.48-0.61)
Stage III-IV     
 Racial/ethnic groups     
  African Americans29746.80.66 (0.51-0.86)14.10.85 (0.59-1.22)
  Others40962.11.47 (1.15-1.88)15.20.88 (0.64-1.22)
 Socioeconomic status, quartile     
  1st, high149156.11.0016.41.00
  2nd154458.51.13 (0.97-1.31)15.70.96 (0.79-1.18)
  3rd151956.81.06 (0.91-1.25)14.20.90 (0.72-1.11)
  4th, low157056.51.10 (0.93-1.30)14.80.96 (0.77-1.19)
  IV446456.00.86 (0.76-0.97)16.71.45 (1.22-1.73)
 Subtotal6124 15.3
All stages     
 Racial/ethnic groups     
  African Americans53343.20.68 (0.56-0.83)18.20.79 (0.62-1.00)
  Others92054.71.32 (1.12-1.55)26.61.02 (0.85-1.23)
 Socioeconomic status, quartile     
  1st, high332952.41.0025.91.00
  2nd333252.31.04 (0.94-1.15)25.40.98 (0.87-1.10)
  3rd332952.71.08 (0.97-1.20)23.80.90 (0.79-1.01)
  4th, low333151.61.04 (0.93-1.16)21.90.84 (0.74-0.95)
  II191961.21.81 (1.61-2.03)26.00.57 (0.51-0.65)
  III166059.71.61 (1.42-1.82)11.40.23 (0.20-0.27)
  IV446456.01.39 (1.26-1.52)16.70.35 (0.32-0.39)
  Unstaged125237.00.74 (0.65-0.85)15.10.31 (0.26-0.36)

Table 5 presents the effects of patient and tumor factors on the risk of all-cause and NHL-specific mortality in association with race/ethnicity for all populations and stratified by stage I-II and stage III-IV. Compared with Caucasians, African Americans did not have statistically significant risks of all cause (hazard ratio, 0.97; 95% CI, 0.88-1.08 for all stages) and disease-specific mortality (hazard ratio, 1.07; 95% CI, 0.92-1.25), whereas other ethnic patients with stage I-II were significantly more likely to die of all-cause (hazard ratio, 1.28; 95% CI, 1.12-1.46) and of NHL-specific mortality (hazard ratio, 1.43; 95% CI, 1.15-1.78), after adjusting for treatment, comorbidity, and socioeconomic factors. Women were significantly less likely to die than men, whereas the opposite was the case for unmarried patients compared with married ones. The hazard ratio of all-cause and disease-specific mortality increased significantly with age (P < .001). The hazard ratio also increased with advanced tumor stage, higher comorbidity scores, and poorer SES. Overall, patients receiving either chemotherapy or radiation therapy or both were significantly less likely to die of all-causes.

Table 5. All-Cause and Non-Hodgkin Lymphoma–specific Mortality in Association With Race/Ethnicity and Other Factors in Patients With Non-Hodgkin Lymphoma, Stratified by Stage
Race/Ethnicity and Other FactorsHazard ratio (95% CI) of Mortality*
All-Cause MortalityNHL-specific Mortality
Stage I-IIStage III-IVAll StagesStage I-IIStage III-IVAll Stages
  • CI indicates confidence interval; NHL, non-Hodgkin lymphoma; SES, socioeconomic status.

  • *

    Hazard ratio adjusted for variables listed in the tables plus the year of diagnosis and Surveillance, Epidemiology, and End Results areas.

 African Americans0.90 (0.75-1.08)1.07 (0.93-1.22)0.97 (0.88-1.08)0.84 (0.62-1.14)1.17 (0.96-1.41)1.07 (0.92-1.25)
 Others1.28 (1.12-1.46)1.02 (0.90-1.17)1.13 (1.03-1.23)1.43 (1.15-1.78)1.08 (0.90-1.30)1.19 (1.04-1.36)
 Women0.78 (0.73-0.84)0.83 (0.78-0.88)0.80 (0.77-0.84)0.83 (0.74-0.93)0.94 (0.86-1.02)0.91 (0.85-0.97)
Age, y      
 70-741.29 (1.16-1.43)1.23 (1.12-1.34)1.26 (1.18-1.35)1.22 (1.03-1.45)1.22 (1.08-1.39)1.23 (1.12-1.36)
 75-791.63 (1.47-1.81)1.52 (1.39-1.66)1.59 (1.49-1.70)1.48 (1.25-1.75)1.41 (1.24-1.61)1.46 (1.32-1.61)
 ≥802.66 (2.41-2.94)2.07 (1.90-2.26)2.39 (2.25-2.55)2.18 (1.86-2.56)1.91 (1.69-2.1602.09 (1.90-2.29)
Marital status      
 Unmarried1.21 (1.12-1.30)1.17 (1.10-1.25)1.20 (1.15-1.26)1.25 (1.12-1.41)1.09 (1.00-1.20)1.18 (110-1.26)
 Unknown0.94 (0.78-1.14)0.67 (0.57-0.80)0.73 (0.65-0.81)0.72 (0.50-1.04)0.54 (0.41-0.71)0.54 (0.45-0.66)
 II1.29 (1.20-1.39)1.28 (1.19-1.37)1.49 (1.33-1.66)1.45 (1.30-1.61)
 III1.001.52 (1.42-1.63)1.001.86 (1.67-2.08)
 IV1.20 (1.13-1.28)1.83 (1.73-1.93)1.19 (1.09-1.31)2.21 (2.03-2.40)
 Unstaged1.61 (1.50-1.74)1.86 (1.65-2.09)
Lymph node site      
 Extranodal0.91 (0.86-0.98)0.85 (0.79-0.92)0.87 (0.83-0.91)0.94 (0.84-1.04)0.80 (0.72-0.90)0.83 (0.77-0.89)
Comorbidity scores
 11.35 (1.25-1.46)1.26 (1.18-1.35)1.31 (1.25-1.37)1.23 (1.09-1.39)1.21 (1.10-1.3301.23 (1.14-1.32)
 21.59 (1.44-1.75)1.60 (1.46-1.74)1.62 (1.52-1.72)1.42 (1.21-1.67)1.37 (1.20-1.56)1.44 (1.31-1.58)
 ≥32.44 (2.19-2.72)1.86 (1.69-2.05)2.08 (1.94-2.22)2.11 (1.78-2.5101.58 (1.37-1.81)1.73 (1.56-1.92)
 Radiation only0.80 (0.72-0.88)1.11 (0.98-1.26)0.87 (0.81-0.94)0.81 (0.69-0.9601.08 (0.90-1.29)0.87 (0.78-0.98)
 Chemotherapy only1.00 (0.93-1.09)0.82 (0.77-0.88)0.88 (0.84-0.92)1.11 (0.98-1.27)0.87 (0.79-0.95)0.94 (0.87-1.00)
 Both0.84 (0.76-0.93)0.93 (0.84-1.03)0.85 (0.79-0.91)1.02 (0.88-1.19)1.07 (0.94-1.23)1.01 (0.91-1.11)
SES, quartile      
 1st, high1.
 2nd1.07 (0.98-1.17)1.05 (0.97-1.1401.04 (0.98-1.10)1.07 (0.92-1.24)1.05 (0.94-1.1901.06 (0.97-1.16)
 3rd1.12 (1.02-1.23)1.16 (1.06-1.26)1.13 (1.07-1.20)1.10 (0.94-1.2801.26 (1.12-1.43)1.20 (1.09-1.31)
 4th, low1.31 (1.19-1.44)1.22 (1.12-1.33)1.23 (1.16-1.31)1.22 (1.04-1.43)1.17 (1.03-1.33)1.18 (1.08-1.30)

When stratified by subtypes of NHL, the results were similar (Table 6). For example, racial disparities in both all-cause and NHL-specific mortality were not significant across diffuse large B-cell lymphoma, chronic lymphocytic leukemia or small lymphocytic lymphoma, follicular lymphoma, and all other subtypes.

Table 6. All-Cause and Non-Hodgkin Lymphoma–specific Mortality in Association With Race/Ethnicity, Stratified by Subtypes of Non-Hodgkin Lymphoma
Race/Ethnicity and Other FactorsHazard Ratio (95% CI) of Mortality,* by Subtypes of NHL
Diffuse Large B-Cell LymphomaChronic Lymphocytic Leukemia or Small Lymphocytic LymphomaFollicular LymphomaAll Other Subtypes or Unknown
  • CI indicates confidence interval; NHL, non-Hodgkin lymphoma.

  • *

    Hazard ratio adjusted for variables listed in the tables plus tumor stage, lymph node site, comorbidity score, therapy, socioeconomic status, the year of diagnosis, and Surveillance, Epidemiology, and End Results areas.

All-Cause Mortality
 African Americans1.05 (0.88-1.25)0.95 (0.69-1.31)1.00 (0.71-1.43)0.94 (0.80-1.10)
 Others1.11 (0.97-1.26)1.00 (0.73-1.37)0.71 (0.47-1.09)1.12 (0.97-1.28)
 Women0.88 (0.82-0.94)0.73 (0.64-0.82)0.74 (0.63-0.86)0.76 (0.71-0.82)
Age, y    
 70-741.19 (1.07-1.32)1.43 (1.20-1.70)1.53 (1.20-1.96)1.19 (1.07-1.32)
 75-791.43 (1.29-1.58)1.85 (1.54-2.21)1.77 (1.38-2.26)1.55 (1.40-1.72)
 ≥801.78 (1.61-1.97)2.94 (2.46-3.52)3.11 (2.46-3.92)2.34 (2.12-2.59)
Marital status    
 Unmarried1.06 (0.99-1.14)1.36 (1.19-1.55)1.17 (0.99-1.38)1.24 (1.15-1.33)
 Unknown0.94 (0.77-1.14)0.92 (0.65-1.30)1.09 (0.79-1.51)0.58 (0.49-0.69)
NHL-Specific Mortality
 African Americans1.19 (0.95-1.50)1.00 (0.59-1.69)0.70 (0.33-1.49)1.04 (0.82-1.33)
 Others1.14 (0.95-1.37)0.87 (0.51-1.48)0.79 (0.33-1.91)1.15 (0.92-1.43)
 Women0.92 (0.84-1.02)0.76 (0.62-0.95)0.77 (0.57-1.05)0.91 (0.82-1.02)
Age, y    
 70-741.16 (1.01-1.33)1.20 (0.91-1.59)1.67 (0.99-2.79)1.23 (1.04-1.45)
 75-791.32 (1.15-1.52)1.36 (1.01-1.84)2.44 (1.48-4.02)1.42 (1.20-1.67)
 ≥801.38 (1.20-1.58)2.31 (1.73-3.09)4.28 (2.66-6.89)2.12 (1.81-2.48)
Marital status    
 Unmarried1.09 (0.99-1.21)1.25 (1.00-1.56)1.13 (0.82-1.55)1.19 (1.06-1.34)
 Unknown0.80 (0.59-1.08)0.67 (0.34-1.32)0.54 (0.26-1.15)0.44 (0.32-0.60)


This study examined racial/ethnic disparities in diagnosis, treatment, and socioeconomic factors in association with survival in patients with NHL. Our study revealed several important findings. First, it indicates that 52.4% of elderly Caucasian patients with NHL received chemotherapy, compared with 43.2% of African Americans (P < .01). Second, although chemotherapy has been documented to be efficacious in prolonging survival in clinical trial settings, few community-based studies demonstrated its effectiveness. Our study of this large population-based cohort indicates that the receipt of chemotherapy was significantly associated with prolonged survival in patients with NHL. Third, there were substantial variations in SES within this cohort of patients, with a larger proportion of African American patients living in the lowest or poorest quartile of SES. Poor SES was significantly associated with increased risk of mortality. Finally, ethnic disparities at diagnosis, treatment, and SES played a major role in survival disparities. After adequately controlling for these factors in the analysis, there were no significant differences in all-cause and NHL-specific mortality between African American and Caucasian patients.

With advances in medical care, particularly with effective chemotherapy, patients with NHL can now live much longer. This survival benefit, documented in numerous clinical trials,8–20 has been translated into the community effectiveness according to our population-based study. Patients receiving chemotherapy had a significantly lower risk of death. However, substantial disparities in receiving these therapies were observed among ethnic groups with a particularly lower rate of chemotherapy and radiation therapy in African American patients. Furthermore, there was a strong relationship between the risk of mortality and SES, and a larger proportion of elderly African American patients lived in communities with poorer SES. Therefore, controlling for socioeconomic factor in the analysis was important when examining racial disparities in survival. This study was the first to show ethnic survival disparity in patients with NHL.

Our study also indicated that other ethnic groups had a slightly higher risk of all-cause and disease-specific death that was limited in patients with stage I-II lymphoma. However, because other ethnic groups had small numbers of cases and included numerous different ethnic populations, the results would be more vulnerable to residual confounding due to imperfect adjustment for differences in these populations. Our study also found that women with NHL were significantly less likely to die of all causes for all stages and to die of lymphoma specifically for stages I-II only, even after adjustment for other factors. This finding was consistent with reports from the SEER registries of a higher relative survival in women than in men.1, 2

This study has several strengths. First, the SEER-Medicare linked data not only provided well-validated information on tumor characteristics at diagnosis and 11 years of follow-up of the vital status, but also allowed the examination of SES at the census tract level. Second, our study cohort of elderly patients was covered by the Medicare insurance program regardless of where they received care in the country. Thus, more complete information on cancer therapy was obtained by Medicare claims. In addition, comorbid conditions can well be identified from Medicare data. Because comorbidity is a known factor associated with treatment and survival, controlling for comorbidity in analysis is important to minimize residual confounding when addressing ethnic disparities in survival.

It is also important to note several limitations of this study. First, only Medicare beneficiaries with fee-for-service for Part A and Part B coverage and without health maintenance organization (HMO) enrollment were selected in the analysis, because Medicare data do not include claims for HMO enrollees. There were potential selection biases here if those beneficiaries excluded had different characteristics or patterns of care among ethnic groups. As a result, the generalizability of the results to other populations could be affected. In addition, the study findings may not be generalizable to the population that is <65 years of age. Furthermore, this database does not have detailed clinical information on lymphoma grade, which may form the basis on which chemotherapy and radiation therapy may be recommended. Therefore, appropriateness of these therapies cannot be assessed adequately from these existing Medicare claims data. Furthermore, there are numerous histologic subtypes of NHL, but some subtypes may not be reliably differentiated. For example, agreement between cancer registry data and chart coders was 90% for diffuse large B-cell lymphoma and 89% for follicular lymphoma, but only 52% for mantle cell lymphoma.35 Therefore, certain groupings may be helpful to reduce misclassification bias.34, 35 Our analysis was stratified by several major subtypes of NHL. Finally, because patients were diagnosed with NH lymphoma from 1992 to 1999, and their treatments were followed to 2002, the use of newly approved chemotherapy agents was unknown, including rituximab, which was approved for treating NHL at the end of 1997. Further studies including cases diagnosed in more recent years would help to address the effect of these new chemotherapy agents.

In conclusion, there was a strong positive association between the use of chemotherapy and prolonged survival, and poorer SES was significantly associated with the higher risk of mortality. There were no significant differences in all-cause and lymphoma-specific mortality between African American and Caucasian patients with non-Hodgkin lymphoma after controlling for treatment, SES, and patient demographics and tumor factors. It would be helpful if further studies incorporate patients' clinical performance status and disease severity.


We acknowledge the efforts of the National Cancer Institute; Center for Medicare and Medicaid Services; Information Management Services, Inc.; and the SEER Program tumor registries in the creation of this database.