Health-related quality of life among survivors of aggressive non-Hodgkin lymphoma

Authors

  • Roxanne E. Jensen PhD,

    Corresponding author
    1. Lombardi Comprehensive Cancer Center, Georgetown University Medical Center, Washington, DC
    • Lombardi Comprehensive Cancer Center, Georgetown University Medical Center, 3300 Whitehaven Street NW, Suite 4100, Washington, DC 20007

    Search for more papers by this author
    • Fax: (202) 687-0305

  • Neeraj K. Arora PhD,

    1. Division of Cancer Control and Population Sciences, National Cancer Institute, Bethesda, Maryland
    Search for more papers by this author
  • Keith M. Bellizzi PhD, MPH,

    1. Human Development and Family Studies, University of Connecticut, Storrs, Connecticut
    Search for more papers by this author
  • Julia H. Rowland PhD,

    1. Division of Cancer Control and Population Sciences, National Cancer Institute, Bethesda, Maryland
    Search for more papers by this author
    • This article reflects the personal opinions of Drs. Arora, Rowland, and Aziz and does not convey any official position of the National Institutes of Health.

  • Ann S. Hamilton PhD,

    1. Keck School of Medicine of University of Southern California, Los Angeles, California
    Search for more papers by this author
  • Noreen M. Aziz MD, PhD, MPH,

    1. National Institute of Nursing Research, Bethesda, Maryland
    Search for more papers by this author
    • This article reflects the personal opinions of Drs. Arora, Rowland, and Aziz and does not convey any official position of the National Institutes of Health.

  • Arnold L. Potosky PhD

    1. Lombardi Comprehensive Cancer Center, Georgetown University Medical Center, Washington, DC
    Search for more papers by this author

Abstract

BACKGROUND:

Non-Hodgkin lymphoma (NHL) is the fifth most common cancer among men and women. Patients with aggressive NHL receive intense medical treatments that can significantly compromise health-related quality of life (HRQOL). However, knowledge of HRQOL and its correlates among survivors of aggressive NHL is limited.

METHODS:

Self-reported data on HRQOL (physical and mental function, anxiety, depression, and fatigue) were analyzed for 319 survivors of aggressive NHL. Survivors 2 to 5 years postdiagnosis were selected from the Los Angeles County Cancer Registry. Bivariate and multivariable methods were used to assess the influence of sociodemographic, clinical, and cognitive health-appraisal factors on survivors' HRQOL.

RESULTS:

After accounting for other covariates, marital status was associated with all HRQOL outcomes (P < .05). Younger survivors reported worse mental function and higher levels of depression, anxiety, and fatigue (P < .01). Survivors who had more comorbid conditions or lacked private health insurance reported worse physical and mental function and higher levels of depression and fatigue (P < .05). Survivors who experienced a recurrence reported worse physical function and higher levels of depression and fatigue (P < .05). With the exception of a nonsignificant association between perceived control and physical function, greater perceptions of personal control and health competence were associated significantly with more positive HRQOL outcomes (P < .01).

CONCLUSIONS:

The current results indicated that survivors of aggressive NHL who are younger, are unmarried, lack private insurance, or experience greater illness burden may be at risk for poorer HRQOL. Cognitive health-appraisal factors were strongly related to HRQOL, suggesting potential benefits of interventions focused on these mutable factors for this population. Cancer 2013. © 2012 American Cancer Society.

INTRODUCTION

Non-Hodgkin lymphoma (NHL) is the fifth most common cancer for both men and women1 and accounts for 3.8% of the 11.9 million cancer survivors in the United States.2 Adult patients with NHL are classified based on histology into 2 groups that have approximately similar rates of incidence: indolent lymphomas (low-grade), which grow slowly, and aggressive lymphomas (intermediate and high-grade), which grow more quickly and can be fatal if not treated aggressively. Treatment options for aggressive NHL include multiagent chemotherapy, radiation, and bone marrow or stem cell transplantation, all of which can significantly compromise patients' health-related quality of life (HRQOL). Although negative changes in HRQOL have been identified after NHL diagnosis and during treatment,3, 4 limited research has been conducted on the impact of a diagnosis of aggressive NHL and its related treatments on HRQOL outcomes among survivors who have completed active treatment.

The majority of HRQOL studies in cancer survivorship have examined breast, colorectal, or prostate cancer5-7 or analyzed mixed cancer survivor populations8, 9 and have demonstrated increased rates of psychological effects, such as depression and anxiety.10-12 The few studies of NHL survivors have indicated that they experienced lower physical and mental HRQOL13, 14 and increased rates of psychosocial distress.15 Although many of the survivorship studies have focused on the association of patient sociodemographic and clinical characteristics with HRQOL, relatively few have examined more intervenable determinants of HRQOL. The existing literature suggests that deficits in HRQOL may be reduced by enhancing patients' cognitive health appraisal, such as their perceptions of personal control over their illness and of health competence and self-efficacy.16-19 However, to our knowledge, the association between cognitive health appraisal and HRQOL has not been studied during post-treatment cancer survivorship among NHL survivors.

To add further to the very limited research on NHL survivors, we conducted a population-based study of HRQOL among survivors of aggressive NHL who were diagnosed 2 to 5 years before the study. This period represents an important point in cancer care for this population, because, during this time, patients transition to follow-up surveillance and still may experience treatment-related HRQOL deficits. We examined survivors' HRQOL using multiple outcome measures: physical function, mental function, anxiety, depression, and fatigue. We also evaluated the association of several patient sociodemographic, clinical, and cognitive health-appraisal factors with HRQOL outcomes.

MATERIALS AND METHODS

Participants and Procedures

We analyzed data from the Experience of Care and Health Outcomes of Survivors of Non-Hodgkin Lymphoma (ECHOS-NHL) study. The patients were adult survivors of aggressive NHL who were selected in 2003 from the Cancer Surveillance Program (Surveillance Epidemiology, and End Results [SEER] registry) for Los Angeles County who were diagnosed 2 to 5 years before the study (between June 1, 1998 and August 31, 2001). The selection criteria included individuals aged ≥20 years with a first-time diagnosis of either intermediate-grade or high-grade NHL. Survivors were excluded if they were diagnosed with another cancer within a year before their NHL diagnosis or subsequently.

We used a mailed questionnaire with telephone follow-up for nonresponders to the mailing, and we also conducted a medical record abstraction to augment clinical data available from the SEER registry. Of the 744 eligible NHL survivors who were identified and mailed a questionnaire, 408 completed it, for a response rate of 55%. Among these 408 survivors, 319 completed the full mailed survey, and 89 completed a briefer version by telephone. Because the brief survey did not include HRQOL measures, analyses in this study are based only on the 319 full-survey respondents. Survivors who completed the brief telephone survey did not differ from those who responded to the full survey on any of the sociodemographic or clinical variables. The only exception was that telephone responders were more likely to have been diagnosed earlier (3.8 years vs 3.5 years; P < .01). The study protocol was approved by the Institutional Review Board at the University of Southern California. Further information about the population, study design, data-collection procedures, differences between responders and nonresponders, and conceptual framework used to guide dependent and independent variable selection for this study has been reported previously in detail.20

Measures

Health-related quality-of-life outcomes

We measured NHL survivors' HRQOL using version 2 of the Medical Outcomes Study 36-item short-form health survey (SF-36),21, 22 a standardized measure that has been used extensively in both general and disease-specific populations, including NHL.13 The SF-36 has 8 subscales: physical function, role limitations because of physical health, mental health, role limitations because of emotional problems, social function, bodily pain, vitality, and general health perceptions. Scores from these subscales were used to calculate 2 summary health scores, the Physical Component Summary (PCS) and the Mental Component Summary (MCS). On the basis of norms from the 1999 general US population, we standardized all scores on a T-score metric, such that a score of 50 represented the average score in the US general population with a standard deviation of 10, with higher scores reflecting better HRQOL. We evaluated anxiety and depression using the Hospital Anxiety and Depression Scale (HADS),23 a 14-item self-report scale that has demonstrated good reliability and validity in cancer populations.24 Seven items each assessed anxiety and depression on a 4-point (0-3) response format. We created separate indicators for anxiety and depression with scores ranging from 0 to 21 (in which scores of 0-7 are considered normal, scores of 8-10 are considered mild anxiety/depression, scores of 11-14 are considered moderate, and scores of 15-21 are considered severe).

We evaluated fatigue using the 7-item fatigue interference subscale of the Fatigue Symptom Inventory (FSI), which has been validated across different cancer patient populations.25, 26 These 7 items ask respondents to indicate the extent to which fatigue interfered during the past week with their general activity, ability to bathe and dress, normal work activity, ability to concentrate, relations with others, enjoyment of life, and mood. Items are scored from 0 (no interference) to 10 (extreme interference). An overall score was created by taking the average of the scores on the 7 items; then, the raw score was transformed linearly to a scale from 0 to 100 to facilitate interpretation, with a higher score representing greater fatigue interference.

Cognitive health appraisal

We measured NHL survivors' perception of their self-efficacy in taking care of their health using a 4-item short form of the Perceived Health Competence Scale (PHCS),27 which has been used in prior cancer studies.19 The PHCS asks responders to indicate their agreement or disagreement, on a 5-point response scale (from “strongly agree” to “strongly disagree”), with the following items: “No matter how hard I try, my health does not turn out the way I would like;” “I am usually unable to find effective solutions for my health problems;” “My efforts to change things about my health are usually ineffective;” and “Typically, my plans for my health do not work out well.”

We measured NHL survivors' perceptions of personal control using a 4-item Perceived Personal Control (PPC) scale that was developed in a prior study on follow-up care experiences of cancer survivors.28 The 4 PPC items assess, on a 5-point response scale (from “no control at all” to “complete control”), survivors' perceptions of control over their emotional responses to their cancer, the physical side effects of their cancer and its treatment, the kind of follow-up care they receive, and the course of their cancer. Exact wordings of items are published elsewhere.28

Scale scores were created for the PHCS and the PPC by taking an average of all the item scores. Raw scale scores were linearly transformed to a scale from 0 to 100, with higher scores representing greater perceptions of health competence and control. For this study, we divided PHCS and PPC scores into 3 approximately equal groups labeled as low, medium, and high.

Patient-level variables

We collected data on several patient sociodemographic variables (age at the time of study, sex, race/ethnicity, education, household income, insurance, and marital status) and clinical characteristics (type of treatment, NHL grade, recency of treatment, number of comorbid health conditions, time since diagnosis, and experience of recurrence or disease progression). Information on NHL grade was obtained from the SEER registry, and all other variables were collected as part of the survey. Survivors' self-reports of recurrence/disease progression and treatment were validated against medical record data.20 Comorbid health conditions were based on a self-reported count of being diagnosed with 1 or more of the following 12 medical conditions used in a prior cancer study: congestive heart failure; heart attack/myocardial infarction; chest pain or angina; high blood pressure; blood clots in the veins of legs or in the lungs; stroke or brain hemorrhage; chronic lung disease, or bronchitis, or emphysema; liver disease or cirrhosis; inflammatory bowel disease, or colitis, or Crohn disease; diabetes; osteoporosis or brittle bones; and arthritis or rheumatism.29

Analytic Methods

We computed descriptive statistics for all independent variables and outcome measures. Our analytic models were limited to 5 HRQOL outcome variables: SF-36 (PCS and MCS), HADS depression and anxiety subscales, and FSI. We conducted bivariate analyses using t tests to evaluate the association of the various patient-level variables (sociodemographic and clinical characteristics) with the outcome variables. Patient variables that were associated with at least 1 of the 5 HRQOL outcomes at P ≤ .1 were included in the multivariable models. NHL grade, time since diagnosis, and treatment type were not associated with any of the outcomes (P > .1) and were not considered for the multivariable models. We included insurance status in the multivariable models but excluded income, because the 2 variables were significantly correlated (ρ = .5; P < .0001), both had similar associations with the outcome variables, and income had a higher percentage of missing data (9% vs 5%). On the basis of findings from the bivariate analyses, we examined the adjusted associations of patient sociodemographic, clinical, and cognitive health-appraisal factors with NHL survivors' HRQOL by estimating 2 sets of hierarchically built analysis of covariance models. The first set (1 model for each of the 5 outcome variables) examined the association of patient sociodemographic and clinical variables (age, sex, race/ethnicity, marital status, education, insurance status, comorbidities, recurrence, and recency of treatment) with each HRQOL outcome. The second set of analysis of covariance models examined the association of the 2 cognitive health-appraisal variables (perceived health competency and personal control) with each HRQOL outcome, adjusting for the sociodemographic and clinical variables that were included in the first set of models. All analyses were conducted using the GLM procedure in SPSS version 14.0 (SPSS, Inc., Chicago, Ill). A P value ≤ .05 was considered statistically significant. In addition to statistical significance, we also highlighted subgroup differences in HRQOL that could be clinically meaningful. HRQOL studies often use the Cohen criteria of effect sizes (small effect: 0.2 of a standard deviation [SD]; medium effect: 0.5 SD; large effect 0.8 SD) to characterize minimally important differences.30, 31 It has also been demonstrated that a difference of 3 to 5 points on the SF-36 scale (a small to medium effect size of 0.3–0.5) is clinically meaningful.32 In the current study, we took a conservative approach and considered subgroup differences equal to or larger than 0.5 of a standard deviation (ie, a medium effect size or larger) to be clinically meaningful.

RESULTS

Sample Characteristics

The sample consisted of equal proportions of men and women, and 42% of survivors were aged ≥65 years, 30% belonged to minority racial/ethnic subgroups, and 65% had private health insurance. Clinically, 16% of respondents had experienced a recurrence or progression of their cancer, and 69% were living with 1 or more comorbid health conditions (Table 1). Of those survivors who reported a recurrence, 46% reported receiving treatment in the past 6 months.

Table 1. Sociodemographic and Clinical Characteristics, n = 319
CharacteristicNo. of Patients (%)
  1. Abbreviations: NHL, non-Hodgkin lymphoma.

Sociodemographics
 Age, y
  <5084 (26.3)
  50-64101 (31.7)
  ≥65134 (42.0)
 Sex
  Women156 (48.9)
  Men163 (51.1)
 Race/ethnicity
  Hispanic68 (21.3)
  Non-Hispanic white223 (69.9)
  Other28 (8.8)
 Marital status
  Married/living as married205 (64.3)
  Other110 (34.5)
  Missing4 (1.3)
 Education
  ≤High school93 (29.2)
  Some college106 (33.2)
  ≥College degree117 (36.7)
  Missing3 (0.9)
 Household income
  <$20,00065 (20.4)
  $20,000-$60,000101 (31.7)
  ≥$60,000125 (39.2)
  Missing28 (8.8)
 Health insurance status
  Private208 (65.2)
  Public or none95 (29.8)
  Missing16 (5.0)
Clinical characteristics
 NHL grade
  Intermediate283 (88.7)
  High36 (11.3)
 Time since diagnosis, y
  Mean3.5
  Range2.1-5.5
 Recurrence/disease progression
  Yes50 (15.7)
  No264 (82.8)
  Missing5 (1.6)
 Treatment in past 6 mo
  Yes35 (11)
  No279 (87.5)
  Missing5 (1.6)
 Treatment type
  Chemotherapy and radiation108 (33.9)
  Chemotherapy only156 (48.9)
  Transplantation34 (10.7)
  Missing21 (6.6)
 No. of comorbid conditions
  None99 (31)
  1 or 2156 (48.9)
  ≥363 (19.7)
  Missing1 (0.3)

Health-Related Quality-of-Life Outcomes Distribution

Among the SF-36 sub-scales, mean scores on the PCS, physical function, and role limitations because of physical health were at least 5 points lower than the mean score of 50 observed in the general US population, a clinically meaningful difference. MCS scores were similar to US population norms (Table 2). The mean HADS scores suggest that the NHL survivors had normal levels of anxiety and depression with respect to US population norms. However, 20% and 16% of survivors did report mild-to-severe levels of anxiety or depression, respectively (data not shown). The mean score on the fatigue interference scale was 22 (SD, 22.8).

Table 2. Distribution of Outcome Variables
OutcomesScore: Mean ± SDMissing, %
  1. Abbreviations: FSI, Fatigue Symptom Inventory; HADS, Hospital Anxiety and Depression Scale; MCS, Mental Component Summary; PCS, Physical Component Summary; SD, standard deviation; SF-36, Medical Outcomes Survey 36-item short form health survey.

Health-related quality of life
 PCS44.8 ± 11.95.3
 MCS49.8 ± 11.05.3
 Anxiety (HADS)4.7 ± 3.81.9
 Depression (HADS)3.9 ± 3.71.9
 Fatigue (FSI)21.6 ± 22.83.1
SF-36 subscale
 General Health46.6 ± 11.51.3
 Mental Health49.6 ± 10.71.3
 Physical Function43.0 ± 12.42.2
 Role Emotional46.4 ± 12.92.8
 Role Physical44.1 ± 13.01.9
 Social Function46.6 ± 12.11.0
 Vitality49.5 ± 11.41.3
 Bodily Pain49.2 ± 11.31.6

Patient Characteristics and Health-Related Quality of Life

Age, marital status, health insurance, number of comorbidities, and NHL recurrence were associated significantly with several of the HRQOL outcomes after adjustment for all other patient characteristics (Table 3). With the exception of the PCS score, younger survivors reported significantly lower HRQOL compared with older survivors (P < .01). Differences in MCS, depression, anxiety, and fatigue scores among survivors aged <50 years and those aged >65 years were large enough (effect sizes ≥ 0.5) to be considered clinically meaningful. Being married or living as a married couple was associated with better HRQOL outcomes across all domains (P ≤ .05).

Table 3. Adjusted Mean Scores of Survivor Health Outcomes by Sociodemographics and Clinical Characteristicsa
 PCSMCSAnxietyDepressionFatigue
VariableMean ScorePMean ScorePMean ScorePMean ScorePMean ScoreP
  • Abbreviations: MCS, Mental Component Summary; PCS, Physical Component Summary; R2, the square of the correlation coefficient.

  • a

    The model was controlled for age, sex, race/ethnicity, marital status, education, health insurance status, recurrence/disease progression, treatment in the past 6 months, and the number of comorbid conditions.

Age, y .28 < .001 < .001 < .01 < .001
 <5042.21 41.95 6.56 5.65 37.69 
 50-6439.51 46.07 5.27 5.12 27.52 
 ≥6539.93 50.57 3.96 3.87 20.93 
Sex .12 .58 .29 .99 .32
 Women39.55 46.55 5.51 4.88 30.06 
 Men41.55 45.85 5.02 4.88 27.36 
Race/ethnicity .15 .87 .17 .48 .24
 Hispanic/other41.68 46.07 5.65 4.70 26.77 
 Non-Hispanic white39.41 46.33 4.88 5.06 30.66 
Marital status < .01 < .05 .05 < .01 < .05
 Married/living as married42.71 48.01 4.77 4.18 24.87 
 Other38.38 44.39 5.76 5.58 32.56 
Education .26 .71 .93 .61 .82
 ≤High school39.65 46.93 5.39 5.04 27.64 
 Some college39.94 45.54 5.23 5.03 29.83 
 ≥College degree42.05 46.12 5.17 4.57 28.67 
Health insurance status < .05 < .01 .64 < .05 < .05
 Private42.52 48.55 5.13 4.26 24.81 
 Public or none38.58 43.84 5.40 5.49 32.62 
Recurrence/progression < .01 .10 .53 .05 < .05
 Yes37.62 44.53 5.49 5.54 33.04 
 No43.47 47.86 5.04 4.22 24.39 
Treatment in past 6 mo .47 .87 .30 .85 .84
 Yes39.73 46.39 4.84 4.80 28.22 
 No41.37 46.01 5.69 4.95 29.21 
No. of comorbid conditions < .001 < .05 .13 < .01 < .01
 None46.24 49.27 4.57 3.60 20.78 
 1 or 240.09 45.79 5.21 4.97 28.84 
 ≥335.32 43.54 6.01 6.06 36.51 
Adjusted R2 .28 .12 .08 .11 12

Survivors with private health insurance reported higher PCS (P < .05) and MCS (P < .01) scores compared with survivors with public insurance or no insurance, even after adjusting for other factors, such as education, race, and age, that are known to be associated with insurance status. The difference in depression and fatigue scores between the 2 insurance groups also was statistically significant (P < .05).

Living with any comorbid health condition in addition to cancer was associated with significantly lower HRQOL scores for PCS (P < .001), MCS (P < .05), depression (P < .01), and fatigue (P < .01). Differences in HRQOL among survivors with no comorbidities and those with ≥3 comorbid health conditions suggested clinically meaningful differences between these subgroups for these outcomes (effect sizes between 0.6 and 1.1). Differences in PCS scores were also clinically meaningful between survivors with no comorbidities and those with 1 or 2 comorbidities (effect size of 0.6). NHL survivors who experienced a cancer recurrence also reported lower PCS scores (P < .01) and higher levels of fatigue (P < .05). The difference in the PCS score was 5.9 points, suggesting clinically meaningful deficits in physical health were also present among survivors who had recurrent NHL compared with those who did not.

Cognitive Health Appraisal and Health-Related Quality of Life

After adjusting for key patient characteristics, we observed significant increases in HRQOL scores with increasing levels of health competence across all 5 HRQOL domains (P < .001) (Figs. 1, 2, 3, 1-3). HRQOL differences among NHL survivors who reported low levels and high levels of health competence far exceeded the threshold of clinically meaningful differences, with effect sizes between 0.8 and 1.0. Meaningful differences also were identified between survivors with low and medium health competence across all HRQOL outcomes except MCS.

Figure 1.

Physical Component Summary (PCS) and Mental Component Summary (MCS) scores are illustrated according to cognitive health appraisal level. The model was controlled for age, sex, race/ethnicity, marital status, education, health insurance status, recurrence/disease progression, treatment in the past 6 months, and the number of comorbid conditions. Adjusted square of the correlation coefficient (R2) = 0.39 for PCS, R2 = 0.29 for MCS.

Figure 2.

Depression and anxiety are illustrated by according to cognitive health appraisal level. The model was controlled for age, sex, race/ethnicity, marital status, education, health insurance status, recurrence/disease progression, treatment in the past 6 months, and the number of comorbid conditions. Adjusted square of the correlation coefficient (R2) = 0.25 for anxiety, and R2 = 0.36 for depression.

Figure 3.

Fatigue is illustrated according to cognitive health appraisal level. The model was controlled for age, sex, race/ethnicity, marital status, education, health insurance status, recurrence/disease progression, treatment in the past 6 months, and the number of comorbid conditions. Adjusted square of the correlation coefficient (R2) = .30 for fatigue.

With the exception of PCS scores, greater perceptions of personal control also were associated with significantly better HRQOL outcomes (P < .01) (Figs. 1, 2, 3, 1-3). Differences in the range of medium to large effect sizes (≥0.5) were identified between survivors with low and high levels of control.

Sensitivity Analyses

Because NHL survivors who are disease free are distinct both conceptually and clinically from survivors whose cancer has recurred, we conducted sensitivity analyses in which we constructed our regression models for only those survivors whose cancer did not recur and who were in remission at the time of the study. Overall, we observed a similar pattern of association for all sociodemographic, clinical, and cognitive appraisal variables. The only exception was that, in our disease-free sample, sex also was associated significantly with PCS scores, with men reporting significantly better physical functioning than women (P = .01).

DISCUSSION

Our findings provide a snapshot of HRQOL for survivors of aggressive NHL 2 to 5 years after diagnosis. Although NHL survivors had MCS scores similar to those of the general US population, their PCS scores were significantly lower, suggesting that many individuals in this survivor population report deficits in physical functioning. Similar findings have been reported in another recent study of NHL survivors that examined all NHL types across a much larger window of time since diagnosis (mean, 11 years).14, 33 These findings suggest that persistent disease- or treatment-related adverse outcomes related to physical function may be present for survivors of aggressive NHL during this period.

The current study also provides further details on NHL survivors' HRQOL by reporting rates of depression, anxiety, and fatigue. Although the anxiety and depression scores suggest that, on average, these are healthy survivors who report normal mental health, approximately 1 in 5 survivors reported scores above the clinical threshold for depression or an anxiety disorder. The overall rates of anxiety and depression in our NHL survivor cohort were slightly higher than those reported in the general US population34 and were consistent with the rates previously reported among lymphoma survivors.35-37 One recent study indicated a higher rate of anxiety (37%) than we did among survivors of aggressive NHL in a clinical setting.38 Fatigue levels were higher than the scores reported (using the same scale that we used) both for the general population26 and for cancer survivors more than 1 year after treatment.25 This suggests that fatigue may be greater among survivors of aggressive NHL than among other cancer survivors; however, further research will be necessary to determine the clinical relevance of this difference.

We identified specific patient-level characteristics that were associated with HRQOL. Younger survivors, those with limited access to health care, those without a partner, and those with greater illness burden reported poorer HRQOL. Subgroup differences often exceeded an effect size of 0.5, suggesting that these HRQOL deficits probably are clinically meaningful. Some patient characteristics were identified as specific to the physical and mental health domains (PCS and MCS): Younger age was associated with poorer mental health, and disease recurrence or progression was associated with poorer physical health. Other characteristics, such as having a partner, private insurance, or having a greater comorbidity burden, were associated with both lower physical health and mental health scores. This supports previous findings reported in survivors of NHL14, 39 and other cancers40, 41 and in the general US population.42 Our current study did not replicate some other relationships identified in previous studies. Although older age has been associated with poorer physical health in cancer survivors and in the general population, we observed no age difference in our study for physical health. Instead, the PCS scores reported here were significantly lower across all age subgroups of NHL survivors than the US general population mean, suggesting a strong physical health burden across this entire survivor population. In addition, sociodemographic characteristics (education, race/ethnicity) and clinical variables (NHL grade and treatment history) were not associated with any HRQOL domains. This may be because of the study's smaller survivorship window of 2 to 5 years and focus on aggressive NHL. Although further research will be needed to investigate these differences, our findings provide important information about specific subgroups of NHL survivors who may be at greater risk for decrements in physical and mental health 2 to 5 years after diagnosis.

Perceptions of greater health competence and personal control were consistently associated with higher HRQOL. Moreover, differences between survivors with low levels of competence and control and those with high levels of competence and control exceeded thresholds for clinical significance. The 1 exception was that physical health was positively associated with perceived health competence but not with personal control. This finding illustrates the subtle, distinct impact that these 2 cognitive factors may have on NHL survivors' perceptions of physical and mental aspects of their HRQOL. In addition, the cognitive appraisal variables explained more variance in MCS, anxiety, depression, and fatigue outcomes than the sociodemographic and clinical factors, indicating the distinctive associations that static patient characteristics and more mutable cognitive factors may have with respect to HRQOL for this population. Overall, our findings suggest that different aspects of survivors' cognitive health appraisal may be associated not only with their perceptions of overall physical or mental health but also with specific long-term symptoms, such as anxiety, depression, and fatigue.

Existing studies report that facilitation of cancer survivors' information-seeking efforts and facilitation of shared decision-making between clinicians and survivors during follow-up care visits may improve cancer survivors' perceptions of health competence and control.19, 28, 43 Our findings suggest that such efforts directed at improving survivors' cognitive health appraisal also may facilitate better adjustment among survivors of aggressive NHL. However, given the cross-sectional nature of our study, we encourage future longitudinal studies to replicate our finding of a positive association of perceived health competence and personal control with HRQOL.

Our findings should be interpreted in light of 3 potential limitations. First, our response rate was modest at 55%, and respondents may have differed systematically from nonresponders. We compared study responders and nonresponders on a limited number of indicators available from the SEER registry (age, sex, race/ethnicity, NHL grade, and time since diagnosis), as published elsewhere.20 Among NHL survivors who we were able to contact, we did not observe any significant differences between those who participated in the study and those who refused to participate. Also, as noted above, we observed little bias in our respondent sample between those who completed the abbreviated telephone interview and those who filled out the mailed questionnaire. Second, this cohort was diagnosed between 1998 and 2001, and medical treatments for aggressive NHL have changed over the past 10 years, especially with the introduction of rituximab. However, treatment for this type of NHL remains aggressive and has the potential for a similar impact on HRQOL, limiting the likelihood of any major differences from what is reported here. Nevertheless, we encourage others to conduct additional studies with more recent cohorts of NHL survivors to replicate our findings and further examine specific treatment protocols. Finally, NHL survivors from Los Angeles County may not be representative of all US survivors, and our findings may not generalize to the experiences of longer-term survivors, thereby limiting the generalizability of our findings. Despite these limitations, the current study adds to the very limited literature on the HRQOL of NHL survivors, providing a more comprehensive assessment of HRQOL outcomes in this population than has been previously reported. The recruitment of survivors from a population-based SEER registry, rather than a clinically recruited sample, further adds to the strength and validity of our findings.

In conclusion, in the current study, we identified key post-treatment HRQOL issues faced by NHL survivors. The largest deficits occurred in physical function. Our study indicated that younger NHL survivors, those who may lack support from a partner or a spouse, those with limited access to care, and those who experience a greater illness burden are likely to experience poorer HRQOL and may warrant special attention during follow-up care visits from health care providers. Our findings also provide a strong foundation for future longitudinal and intervention research that should explore the potential for improving HRQOL among NHL survivors by enhancing their perceptions of health competence and personal control.

FUNDING SOURCES

The collection of cancer incidence data used in this study was supported by the California Department of Health Services as part of the Surveillance, Epidemiology, and End Results Program (under contract N01-PC-35,139 awarded to the University of Southern California and contract N02-PC-15,105 awarded to the Public Health Institute) and by the Centers for Disease Control and Prevention's National Program of Cancer Registries (under agreement U55/CCR921930-02 awarded to the Public Health Institute).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary