The Association of Health Behaviors with Quality of Life in Lymphoma Survivors

Abstract

The impact of change in health behaviors [physical activity (PA), alcohol and smoking] on quality of life (QOL) in lymphoma survivors is not well understood. We evaluated the associations of health behaviors with QOL domains at diagnosis and at 3-year follow-up (FU3) in 2805 lymphoma survivors. We report clinically significant QOL score differences, defined as scores that exceeded a minimally important difference threshold and were statistically significant. Current smoking was associated with lower QOL at baseline (p<0.01) and at FU3 (p<0.01). Meeting the American Cancer Society PA guidelines was associated with better functional wellbeing and overall QOL at FU3 (p<0.01). An increase in PA from baseline to FU3 was associated with improvement in physical, functional wellbeing and overall QOL at FU3 compared to baseline (p<0.01). Thus, QOL in lymphoma survivors is associated with their health behaviors and active interventions to promote positive lifestyle changes in lymphoma survivors are needed.

Introduction

Lymphoma is a heterogeneous group of malignancies that can follow an aggressive or indolent disease course. Non-Hodgkin lymphoma (NHL) is currently the seventh most common cancer in the United States and the most common hematologic malignancy (1). With improvements in treatments, the prevalence of lymphoma survivors is expected to continue to grow. Lymphoma survivors are at risk for long term and late effects, which may be related to the disease biology or treatments, and may negatively impact quality of life (QOL)(2).

The American Cancer Society (ACS) in 2006 published guidelines for nutrition and physical activity (PA) during and after cancer treatment (3). However, very few cancer survivors meet the PA, diet, and smoking recommendations even though higher QOL is associated with meeting more health behavior recommendations(4). Although cancer survivors are more likely to adopt healthy lifestyle changes than the non-cancer population, the level of PA is often lower (5, 6). NHL survivors have also been noted to have low compliance with ACS guidelines, which is associated with lower QOL (7). Few studies have assessed the impact of change in health behaviors after a cancer diagnosis, specifically in lymphoma survivors. Therefore, we sought to measure the impact of smoking, alcohol use and levels of PA on QOL in lymphoma survivors prospectively enrolled in a cohort study, assess the change in these health behaviors and their association with change in QOL.

Patients and Methods

Study Cohort:

The study cohort comprised of patients aged ≥18 years prospectively enrolled within 9 months of lymphoma diagnosis into the University of Iowa and Mayo Clinic Lymphoma SPORE Molecular Epidemiology Resource (MER); full details of the MER have been previously published (8). This analysis only includes participants enrolled at the Mayo Clinic in Rochester, Minnesota, who were also enrolled in a companion study of lymphoma etiology from 9/1/2002 through 12/31/2012 (9). Pathology was reviewed by a hematopathologist and classified based on the WHO criteria (10, 11) (Supplementary S2). Study personnel abstracted baseline clinical data. At enrollment, participants completed a baseline health and risk-factor questionnaire (RFQ). At 3 years after diagnosis (FU3), a survivorship questionnaire was completed by participants.

Measurement of QOL and health behaviors

At baseline and FU3, QOL was measured using the Functional Assessment of Cancer Therapy - General scale (FACT-G, Version 4) which is a validated general QOL assessment tool for cancer patients measuring QOL in four domains: emotional (EWB), functional (FWB), physical (PWB) and social/family well-being (SFWB) (12). Participants answer 27 items on a 0 to 4 scale, a score of 4 reflects items that “very much” apply to them in the past 7 days while a score of 0 indicates “not at all”. The four subscales are summed to calculate overall QOL score. A lower score indicates a poorer QOL. Scores range from 0 to 24 for EWB, 0 to 28 for FWB, PWB, SFWB and 0 to 108 for total FACT-G. Patients who completed less than 80% of QOL questions (i.e., <22 questions) were excluded. In order to minimize the effect of occult lymphoma recurrences on reporting of health behaviors or QOL, patients who had events (disease recurrence, re-treatment or death) within 6 months of completing their FU3 were excluded. The minimally important difference (MID) in QOL has been defined as “the smallest difference in score in the domain of interest that patient’s perceive is important, either beneficial or harmful, and which would lead the clinician to consider a change in the patient’s management” (13). Recommended MIDs are 2 – 3 points for the FACT-G subscales and 3 – 7 points for the total score. For this study, the lower end of these recommended MID ranges (>2 points on a subscale and >3 points on the total FACT-G) were used as the thresholds for evaluating clinical significance since differences were being measured at the group level (14).

At baseline, participants were asked if they had smoked at least 100 cigarettes in their entire life. If yes, then age started and quit smoking, number of cigarettes smoked per day, and current smoking status. Responses were used to determine smoking status (current/former/never smoker) and to calculate the number of cigarettes smoked in pack years. At FU3, participants reported their smoking status and current smokers were asked how their smoking changed since the diagnosis of lymphoma.

Data on alcohol use was collected at baseline by asking the question “During your entire life, have you had 12 drinks or more of any kind of alcoholic drink?” Those who answered “yes” were asked to specify the frequency and amount of alcohol use. At FU3, participants were asked “How has your use of alcohol changed since the diagnosis of lymphoma?” Participants responded as either increase, decrease or no change. This information was used to determine alcohol use status (ever/never/former), amount in drinks/week and change in alcohol use since lymphoma diagnosis.

The Godin Leisure Score Index was calculated as a measure of PA at baseline and FU3 (15, 16). A moderate-to-strenuous Leisure Score Index score of ≥24 is used to classify cancer survivors as meeting the ACS PA guideline of ≥ 150 minutes of moderate to vigorous PA per week (17). At FU3, participants were asked “How has your level of PA changed since your diagnosis of lymphoma?” participants responded as increase, decrease or no change.

Statistical Analyses

The associations of health behaviors with QOL were evaluated using ANOVA test (categorical) and linear regression (continuous) in the four domains as well as overall QOL. Multivariable models were used to examine multiple health behaviors at once. Chi-square tests were used to compare health behaviors by clinical characteristics (age, sex, lymphoma type, and lymphoma biology). Only paired measurements of health behaviors and QOL were included in the analysis for change. A P-value of < 0.05 was considered statistically significant and QOL score differences that exceeded the MID were considered clinically significant. The same MID threshold was applied to cross-sectional group comparisons (e.g., QOL of current vs. former smokers) as well as to changes in QOL scores over time. All statistical analyses were conducted using R version 3.4.2(18).

Results

2805 participants were evaluable for QOL at baseline and 1936 at FU3 for cross-sectional group-level comparisons, and 1486 with data at both assessments were evaluable for QOL change over time. (Supplemental Figure S1 - CONSORT diagram). Table 1 lists patient characteristics of the evaluable cohort at baseline.

Table 1:

Baseline Characteristics of Patients Evaluable for QOL

	QOL assessed prior to treatment N=2005	QOL assessed after starting lymphoma treatmentN=800	All patients with QOL available at baseline N=2805
Age at diagnosis, median (range), in years	62 (18–90)	61 (18–91)	62 (18–91)
Age group  ≤ 60 years  >60 years	928 (46%) 1074 (54%)	391 (49%) 409 (51%)	1319 (47%) 1483 (53%)
Gender  Female  Male	829 (41%) 1173 (57%)	329 (41%) 471 (59%)	1158 (41%) 1644 (59%)
ECOG performance status  <2  ≥2	1927 (97%) 61 (3%)	714 (90%) 79 (10%)	2641 (95%) 140 (5%)
Subtype  CLL/SLL  Follicular  Diffuse large B-cell  Hodgkin  Marginal zone  Mantle cell  T-cell  Other NHL	604 (30%) 388 (19%) 239 (12%) 135 (7%) 178 (9%) 98 (5%) 89 (4%) 271 (13%)	41 (5%) 107 (13%) 214 (27%) 94 (12%) 49 (6%) 45 (6%) 77 (10%) 173 (22%)	645 (23%) 495 (18%) 453 (16%) 229 (8%) 227 (8%) 143 (5%) 166 (6%) 444 (16%)
Histology  Indolent  Aggressive	1282 (70%) 539 (30%)	238 (36%) 416 (64%)	1520 (61%) 955 (39%)
Ann Arbor Stage  I-II  III-IV	655 (74%) 224 (26%)	840 (78%) 234 (22%)	815 (37%) 1391 (63%)

QOL= quality of life; ECOG= Eastern Cooperative Oncology Group; CLL/SLL= chronic lymphocytic leukemia/small lymphocytic lymphoma; NHL= non-Hodgkin lymphoma

Smoking

At baseline, there were 8% (N=220) current, 29% (N=814) former and 47% (N=1310) never smokers. Smoking status was not available for 461 (16%) participants. The median number of pack years smoked among current and former smokers (N=871) was 15.5 (IQR 5.7 – 33.0). There were no differences in mean QOL scores among never vs former smokers. However, individuals who were current smokers had lower mean QOL scores for EWB (17.4 vs 18.4, p<0.01), FWB (18.2 vs 20.9, p<0.01), PWB (22.0 vs 23.3, p<0.01), SFWB (23.1 vs 24.3, p<0.01) and total FACT-G (80.7 vs 86.8, p<0.01) compared to never smokers; results were similar when compared to former smokers. While these differences met statistically significance, only the differences in FWB (>2 points) and total FACT-G (>3 points) among current smokers compared to never/former smokers exceeded the threshold for clinical significance (Figure 1A).

Figure 1:

Association of smoking status with QOL in lymphoma survivors at baseline and FU3 [*= statistically significant, p<0.05; MID= clinically meaningful difference]

At FU3, there were 5% (N=90) current, 38% (N=738) former and 55% (N=1063) never smokers. Smoking status was not available for 45 (2%) participants. There were no significant differences in QOL domains among never vs former smokers at FU3. However, QOL was lower for current smokers compared to never smokers in all domains (EWB 19.0 vs 20.1, p<0.01; FWB 21.1 vs 22.8, p<0.01; PWB 22.9 vs 24.9, p<0.01; SFWB 21.6 vs 22.9, p=0.03; total FACT-G 84.6 vs 90.8, p<0.01) and compared to former smokers for EWB (19.0 vs 20.2, p<0.01); FWB (21.1 vs 22.4, p=0.03); PWB (22.9 vs 24.6, p<0.01) and total FACT-G (84.6 vs 90, p<0.01). Total FACT-G scores for current smokers were clinically meaningfully lower (> 3 points) than scores for both never and former smokers (Figure 1B).

Of participants evaluable for QOL and smoking at both FU3 and baseline (N=1450), 2% (N=42) quit smoking, 0% (N=0) started smoking, and 73% (N=1408) reported no change in smoking status since diagnosis. Change in smoking status was not available for 25% (N=486). Of the 220 current smokers at baseline, 42 (19%) reported quitting, 67 (30%) continued to smoke and 111 (50%) did not have FU3 smoking data. Due to small numbers of participants reporting a change in smoking status and the large proportion of baseline smokers with missing FU3 data, association of change in smoking status with change in QOL could not be meaningfully analyzed.

Alcohol Use

At baseline, participants reported alcohol use as 42% (N=1189) current, 18% (N=506) former and 18% (N=518) never; 592 (21%) participants were missing data. Among current and former alcohol users, the median number of drinks per week was 2.1 (IQR 0.8 – 4.3). There were no significant differences in QOL scores between never and former alcohol users except a lower mean SFWB among former users (23.4 vs 24.3, p<0.01). Current alcohol users had significantly higher mean scores compared to those who never used alcohol for FWB (21.1 vs 20.1, p<0.01), PWB (23.8 vs 22.6, p<0.01), total FACT-G (87.4 vs 85.4, p<0.01). Current alcohol users also had higher mean FWB (21.1 vs 19.6, p<0.01), PWB (23.8 vs 22.4, p<0.01), SFWB (24.2 vs 23.4, p<0.01) and total FACT-G scores (87.4 vs 83.7, p<0.01) compared to former alcohol users. The only clinically meaningful QOL difference was the higher total FACT-G score (MID>3 points) for current alcohol users compared to former users (Figure 2A).

Figure 2:

Association of alcohol use status and QOL in lymphoma survivors at baseline [*= statistically significant, p<0.05; MID= clinically meaningful difference]

Of participants evaluable for QOL and alcohol use at both FU3 and baseline (N=1218), 25% (N=307) decreased, 3% (N=40) increased, and 71% (N=871) reported no change in alcohol use since lymphoma diagnosis. At FU3, information on amount of current alcohol use was not collected. There was no association between increasing or decreasing alcohol use and change in QOL, although the evaluation for increasing use was limited by a small sample.

Physical Activity

At baseline, 37% (N=1041) of participants met the ACS PA guidelines while 42% (N=1190) did not; 574 (20%) participants were not evaluable due to missing information. Participants meeting the PA guidelines had higher mean scores for FWB (20.8 vs 20.2, p=0.04), SFWB (24.4 vs 23.9, p<0.01) and total FACT-G (87 vs 85.3, p<0.01) compared to those not meeting the guidelines. The difference in QOL scores by meeting/not meeting PA guidelines, however, did not meet the criteria for clinical meaningfulness (Figure 3A).

Figure 3:

Association of level of physical activity and QOL in lymphoma survivors [*= statistically significant, p<0.05; MID= clinically meaningful difference]

At FU3, 26% (N=497) participants met the ACS PA guidelines, 40% (N=783) did not; 656 (34%) were not evaluable for the PA Leisure Score Index. Meeting the ACS guidelines was associated with higher mean scores for EWB (20.5 vs 19.7, p<0.01), FWB (24.0 vs 21.6, p<0.01), PWB (25.7 vs 23.9, p<0.01) and total FACT-G (93.3 vs 87.8, p<0.01) compared to not meeting the guidelines. Clinically meaningful differences were noted in FWB (>2 points) and total FACT-G (>3 points) scores based on meeting the PA guidelines at 3-year follow-up (Figure 3B).

Of participants evaluable for PA and QOL at both FU3 and baseline (N=1041), change in PA since diagnosis of lymphoma was reported as increased (10%, N=106), decreased (38%, N=391) and unchanged (47%, N=492). 25 (2%) participants did not provide a response to the change in PA question. Reporting an increase in the level of PA since lymphoma diagnosis compared to those reporting a decrease in PA was associated with a greater improvement in QOL scores from baseline to FU3 for EWB (2.5 vs 1.4, p<0.01), FWB (3.7 vs 1.2, p<0.01), PWB (2.6 vs 0.5, p<0.01) and total FACT-G (7.6 vs 1.1, p<0.01). Changes in FWB, PWB (>2 points) and total FACT-G (>3 points) were clinically meaningful. The changes in QOL were also significantly different for those reporting an increase vs no change in PA since lymphoma diagnosis for the following domains: FWB (3.7 vs 1.6, p<0.01), PWB (2.6 vs 1.1, p<0.01) and total FACT-G (7.6 vs 3.4, p<0.01). Change in FWB (>2 points) and total FACT-G (>3 points) were clinically meaningful differences. Those reporting decrease in PA had less improvements in QOL over time compared to those with no change in PA: EWB (1.4 vs 1.9, p=0.03), and total FACT-G (1.1 vs 3.4, p=0.01). SFWB decreased over time, more for those reporting decrease in PA compared to no change in PA (−2.0 vs −1.2, p=0.05). However, these differences were not clinically meaningful (Figure 3B).

Multivariate, adjusted models and sub-analyses

The distribution of current smoking status, current alcohol use and meeting ACS PA guidelines varied by age, sex and lymphoma subtype (aggressive vs indolent) (Table 2). Therefore, analysis with adjustment for these variables was performed and the association of health behaviors with QOL remained intact. These associations also remained significant in multivariate models that included smoking status, alcohol use status and PA levels. Participants with different lymphoma subtypes may have different survivorship experiences, therefore, we also performed separate sub-analyses of participants with the most common lymphoma subtypes [CLL/SLL N=645, FL N=495 and DLBCL N=453]. Overall results of the sub-analyses aligned with the results presented above. Some additional associations of health behaviors with specific QOL domains by lymphoma subtype were noted but are difficult to interpret given attenuated numbers. Since patients could be enrolled at any point within the first 9 months from diagnosis, the timing of initial QOL assessment was variable with respect to the start of lymphoma treatment. In a sensitivity analysis, QOL was modeled for health behaviors in the subset of patients with QOL assessed prior to start of treatment; results were consistent with the associations observed for all patients except association of current alcohol use with better QOL was no longer statistically significant.

Table 2:

Distribution of baseline health behaviors by age group, sex and type of lymphoma

		Age			Sex			Lymphoma type				Lymphoma biology
		<=60 N=1319	>60 N=1483	P value	Female N=1158	Male N=1164	P value	CLL/SLL N=645	DLBCL N=453	FL N=495	P value	Aggressive N=955	Indolent N=1517	P value
Smoking	Baseline smoking status
	Current	138 (13%)	82 (7%)	<0.001	100 (10%)	120 (9%)	<0.001	48 (9%)	31 (8%)	42 (10%)	0.103	68 (9%)	123 (9%)	0.083
	Former	275 (25%)	539 (43%)		297 (30%)	517 (39%)		228 (41%)	130 (34%)	142 (33%)		261 (34%)	500 (38%)
	Never	685 (62%)	623 (50%)		602 (60%)	706 (53%)		282 (50%)	216 (57%)	242 (57%)		445 (58%)	689 (53%)
	Missing	221	239		159	301		87	76	69		181	208
	Change in smoking status at FU3
	Increase	0	0	0.002	0	0	0.674	0	0	0	0.466	0	0	0.981
	Decrease	29 (4%)	13 (2%)		17 (3%)	25 (3%)		7 (2%)	7 (3%)	9 (3%)		12 (3%)	25 (3%)
	No change	634 (96%)	772 (98%)		615 (97%)	791 (97%)		363 (98%)	207 (97%)	267 (97%)		410 (97%)	847 (97%)
	Missing	656	698		526	828		275	239	219		533	648
		Age			Sex			Lymphoma type				Lymphoma biology
		<=60N=1319	>60N=1483	P value	FemaleN=1158	MaleN=1164	P value	CLL/SLLN=645	DLBCLN=453	FLN=495	P value	AggressiveN=955	IndolentN=1517	P value
Alcohol Use	Baseline alcohol use status
	Current	575 (58%)	614 (50%)	<0.001	448 (48%)	741 (58%)	<0.001	318 (58%)	163 (46%)	222 (54%)	0.005	360 (49%)	719 (57%)	0.002
	Former	231 (23%)	275 (22%)		189 (20%)	317 (25%)		108 (20%)	94 (26%)	86 (21%)		195 (26%)	266 (21%)
	Never	182 (18%)	336 (27%)		299 (32%)	219 (17%)		118 (22%)	101 (28%)	99 (24%)		182 (25%)	281 (22%)
	Missing	331	258		222	367		101	95	88		218	251
	Change in alcohol use at FU3
	Increase	29 (5%)	29 (5%)	<0.001	14 (3%)	26 (4%)	0.067	10(3%)	5 (3%)	7 (3%)	0.946	10 (3%)	27 (4%)	0.247
	Decrease	164 (29%)	142 (2%)		109 (22%)	197 (27%)		77 (25%)	49 (28%)	56 (24%)		98 (27%)	169 (23%)
	No change	382 (66%)	487 (76%)		372 (75%)	497 (69%)		222 (72%)	124 (70%)	170 (73%)		248 (70%)	533 (73%)
	Missing	744	843		663	924		336	275	262		599	791
		Age			Sex			Lymphoma type				Lymphoma biology
		<=60N=1319	>60N=1483	P value	FemaleN=1158	MaleN=1164	P value	CLL/SLLN=645	DLBCLN=453	FLN=495	P value	AggressiveN=955	IndolentN=1517	P value
Physical activity	Baseline level of physical activity
	Meeting ACS guidelines	477 (46%)	564 (47%)	0.820	446 (47%)	595 (47%)	0.846	248 (47%)	173 (49%)	188 (46%)	0.689	358 (48%)	568 (45%)	0.193
	Not meeting ACS guidelines	551 (54%)	639 (53%)		505 (53%)	685 (54%)		284 (53%)	181 (51%)	222 (54%)		382 (52%)	684 (55%)
	Missing	291	280		207	364		113	99	85		215	268
	Self-Reported Change in level of physical activity at FU3
	Increase	72 (15%)	34 (7%)	<0.001	48 (11%)	58 (11%)	0.817	24 (10%)	15 (10%)	28 (15%)	0.001	30 (10%)	66 (11%)	<0.001
	Decrease	182 (37%)	209 (42%)		175 (40%)	216 (39%)		83 (34%)	76 (51%)	61 (33%)		138 (48%)	204 (34%)
	No change	234 (48%)	258 (52%)		211 (49%)	281 (51%)		140(57%)	58(39%)	98 (52%)		122 (42%)	325 (55%)
	Missing	831	982		724	1089		398	304	308		646	880

Discussion

We found several clinically meaningful associations between health behaviors and QOL in lymphoma patients at diagnosis and in survivors at 3-year follow-up. The findings were as follows: 1) current smoking status was associated with lower FWB and total QOL at baseline and follow-up; 2) meeting the ACS physical activity guidelines at FU3 was associated with higher FWB and total QOL; and 3) reporting an increase in physical activity since lymphoma diagnosis was associated with greater increase in PWB, FWB and total QOL at FU3.

QOL is an important measure of a patient’s overall well-being and, in certain lymphoma subtypes, may be a prognostic indicator of overall survival (19). Although lymphoma survivors may experience changes in their QOL, they may not express these at oncology visits and information may need to be actively elicited in order to provide supportive interventions (20). One mediator of QOL may be health behaviors. Spector et al. in a cross-sectional study of 566 long-term NHL survivors (average 15.2 years since diagnosis) found that only 11% of survivors met all 4 of the ACS health behaviors guidelines (≥ 5 servings of fruits and vegetables per day, ≥ 150 minutes of aerobic PA per week, non-smoker and normal BMI). Meeting more of the ACS health behavior recommendations was associated with better QOL. Similar to our findings, they reported 6% prevalence of active smokers, and 52% survivors not meeting the PA guidelines. In their study, the strongest association of QOL was with non-smoking status and exercising ≥ 150 minutes/week (7).

Smoking tobacco is implicated as the cause of numerous health problems, such as pulmonary disease, cardiovascular disease, and cancer including secondary malignancies in cancer survivors (21, 22). Multiple studies in the general population have established that lower QOL is associated with higher smoking initiation and lower smoking-cessation rates. Furthermore, smoking-cessation has been shown to significantly improve QOL (23). Although our study was limited to lymphoma survivors, we found a persistent negative association of current smoking status and lower QOL at diagnosis as well as FU3. A significant negative correlation between number of pack-years smoked with QOL has also been established in the general population (23). This suggests that smoking cessation may improve QOL even in lymphoma survivors, and the diagnosis of lymphoma should be considered a teachable moment to aid in smoking cessation efforts (24). Even though smoking as an etiological agent for lymphoma is not established, there are other benefits for smoking cessation at the time of lymphoma diagnosis. Certain drugs to treat lymphoma (bleomycin, checkpoint inhibitors) have known pulmonary toxicities that could be compounded by active smoking. In addition, smoking may affect metabolism of chemotherapeutic agents (25). Although the numbers for change in smoking status were too small to analyze for change in QOL related to smoking cessation in our study, it is important to note that only 42 (19%) participants reported quitting smoking at FU3. Counselling alone without medication intervention or referrals to tobacco cessation programs is often inadequate to ensure successful smoking cessation efforts (26). Therefore, our data supports the need for more aggressive intervention and referral to tobacco cessation programs from oncology clinics when a patient is identified as an active smoker. It is also important to follow-up smoking status after cancer treatment to reinforce cessation efforts during survivorship to help patients succeed.

A majority of published literature on alcohol consumption and its association with QOL is focused on patients with alcohol use disorders and supports improvement in QOL through interventions to reduce alcohol dependence (27, 28). In our cohort, the average alcohol consumption was 2.1 drinks per week, which is “moderate” consumption. Although the literature on moderate alcohol consumption and QOL is mixed, studies including a large population-based study from South Korea have shown that the amount of alcohol consumed and QOL falls on a bell shaped curve with QOL of moderate-drinkers being higher than never-drinkers and heavy-drinkers (29, 30). At baseline, we found former alcohol users had clinically significant lower QOL compared to current users. However, this association was lost in the sensitivity analysis including only those patients who filled out the questionnaire prior to lymphoma treatment, which included fewer patients with aggressive lymphomas and worse performance status (Table 1). Decline in overall physical health has been shown to be associated with decrease in alcohol consumption (31) and this may explain our results. Due to the small number of participants in our cohort reporting a change in alcohol use from baseline to FU3, associations with change in QOL over time with regards to alcohol use could not be meaningfully interpreted.

Two randomized trials in lymphoma patients have shown that a supervised exercise intervention, compared to usual care, improved physical functioning and QOL at the end of the brief intervention period (32, 33). Other cross-sectional population-based studies have found that meeting ACS PA guidelines is associated with better QOL (7, 34). Our study confirms these findings and, more importantly, shows that an increase in the level of PA over a 3-year time period correlates with a positive change in overall QOL. The decline in SFWB over time in lymphoma survivors has been previously reported (35, 36). We have previously shown that meeting the ACS PA guidelines as well as increasing PA after lymphoma diagnosis is associated with improved overall and lymphoma-specific survival (16). Thus, there is strong evidence from these studies to support interventions to increase PA in lymphoma survivors to improve both QOL and longevity. In our cohort, we observed 37% participants at baseline and 26% at FU3 meeting the ACS PA guidelines and only 10% reported an increase in PA over 3 years. Older patients and those with aggressive lymphomas were more likely to have a decline in PA at FU3. This impairment in physical functioning in aggressive compared to indolent lymphoma survivors has been reported in one other study (37). There is significant need to raise awareness among both patients and providers regarding the benefits of PA as well as design studies to identify barriers (38) to lifestyle modifications and feasible interventions to improve health behaviors.

Our study’s strength lies in the number of participants, detailed clinical information, paired measurements of self-reported health behaviors and QOL on standardized and validated questionnaires allowing assessment of change from baseline. However, this is an observational study and self-reported data can be subject to measurement and recall bias. Surveys were completed by 860 fewer participants (30%) at FU3 compared to baseline, which may have introduced bias if those who failed to respond were systematically different than those who did, an issue inherent to the survey-based study design. It is known that QOL in lymphoma survivors is influenced to a great extent by their disease and its physical symptom burden (39{Lin, 2018 #98, 40). QOL in lymphoma patients may also improve with time from diagnosis as patients adapt to living with the diagnosis (35). Therefore, it may be difficult to attribute the changes in QOL to changes in health behaviors alone and additional confounding factors may be present.

In conclusion, our study shows that health behaviors in lymphoma survivors are significantly associated with QOL. Better QOL is associated with not smoking and being more physically active. An increase in PA after lymphoma diagnosis is associated with an incremental change in the QOL. Thus, modification of health behaviors has the potential to impact QOL in lymphoma survivors and interventions to drive positive lifestyle changes need to be designed.