Health Behaviors in Cervical Cancer Survivors and Associations with Quality of Life

Abstract

Purpose

Improvement in health behaviors following cancer diagnosis may contribute to better prognosis and well-being. This study examines the prevalence of health behaviors in cervical cancer survivors who have completed treatment, and associations between health behaviors and quality of life (QOL).

Methods

We recruited 204 women who had completed treatment for cervical cancer to participate in a randomized counseling intervention. Participants provided information on health behaviors (smoking, physical activity and alcohol consumption), QOL (FACT-Cx), and depression (PROMIS), anxiety (PROMIS) and distress (Brief Symptom Inventory) at baseline (9–30 months after diagnosis) and subsequent to the intervention. Data were analyzed using multivariate general linear models.

Findings

Participants ranged in age from 20 to 72 years at diagnosis (mean=43), 41% were Hispanic and 52% were non-Hispanic Caucasian. Three-fourths were stage 1 at diagnosis and 51% were treated with radiation with or without chemotherapy. At baseline, 15% of subjects were current smokers, 4% reported alcohol consumption of >10 drinks per week, and 63% reported exercising less than 3 hours per week. Overall, 67.4% of cervical cancer survivors did not meet recommended national guidelines for at least one of these health behaviors. QOL scores were significantly higher for subjects with greater physical activity (128 vs 118, p=0.002) and increased with the number of recommended guidelines met (p for trend = 0.030). Associations between patient-reported outcomes and smoking and alcohol consumption did not reach statistical significance. Participants who met guidelines for all health behaviors also had lower depression (p=0.008), anxiety (p=0.051) and distress (p=0.142). Participants who improved their aggregate health behaviors during the 4 month follow-up experienced a greater improvement in QOL than those who did not improve their health behaviors (10.8 vs. 4.5, p=0.026).

Implications

Results indicate that two-thirds of cervical cancer survivors are not meeting national guidelines for smoking, physical activity and alcohol consumption following completion of definitive treatment. These adverse health behaviors were associated with impaired QOL and higher levels of depression and distress. Positive changes in health behaviors are associated with significant improvement in QOL.

Background

Despite a decrease in cervical cancer incidence over the past two decades, cervical cancer represents a significant number of gynecological cancers in the United States accounting for approximately 12,900 cases in 2015.¹ Because of improved screening and diagnostic methods, nearly half of all cervical cancer cases are diagnosed at stage 1 and more than three-fourths of cases are diagnosed at stage 2 or earlier.² The overall 5-year survival rate for stage 1 disease is 91%, while among all cases, 68 percent are expected to survive 5 years.¹ Because of the potential for long-term survival, lifestyle factors which may influence prognosis and quality of life (QOL) are important for cervical cancer survivors.

Smoking, lack of physical activity, and alcohol consumption are health behaviors which have all been identified as potential risk factors for cancer. Smoking, a causal factor for cervical cancer,³ is more common in women with cervical cancer than in women with other cancers⁴ and women without cancer.⁵ Physical inactivity is prevalent in cervical and gynecologic cancer survivors, with 60%–70% identified as not meeting national recommendations,^4,6,7 and has been associated with significantly higher risk for cervical cancer.⁸ Cancer survivors also report higher alcohol consumption compared to those with no cancer history.^9,10 Furthermore, women with a history of alcohol abuse are at increased risk for cervical cancer.¹¹

These same health behaviors have been more recently examined for their contribution to recurrence and poor prognosis in cancer survivors. In prospective analyses, smokers with cervical cancer had more recurrences than non-smokers, with significantly shorter relapse-free (28.5 vs 46.8 months) and overall survival (38.7 vs 50.1 months).¹² Compared to non-smokers, women with cervical cancer who smoked had a greater likelihood of death from any cause (HR=1.35, 95% CI: 1.17–1.56) and death from cervical cancer (HR=1.21, 95% CI: 1.01–1.46).¹³ Heavy alcohol use, reported by 10% of advanced cervical cancer patients in another study, was associated with a significant decrease in disease-free (HR=10.57, 95% CI: 2.07–53.93) and overall (HR=10.80, 95%CI: 2.57–45.40) survival after adjusting for covariates.¹⁴ A number of studies have reported lower cancer and overall mortality with increased physical activity in cancer survivors, albeit mostly in patients with cancers other than cervix.^15–18

Better QOL is a strong predictor of improved prognosis and overall survival in cancer clinical trials,^19,20 thus a growing body of literature has examined associations between health behaviors and QOL in cross-sectional data on cancer survivors. While treatment and disease side effects contribute to compromised QOL in cancer patients, evidence suggests that lifestyle factors may also play a role. Adherence to physical activity guidelines was associated with higher QOL in gynecologic cancer survivors,⁴ older female cancer survivors,²¹ and in a representative cross-section of cancer survivors.^22–24 A positive association with QOL has also been observed with a combination of lifestyle behaviors including physical activity, diet, lower alcohol and/or tobacco consumption in survivors of ovarian,²⁵ endometrial cancer,²⁶ breast,²⁷ colorectal²⁸ and other cancers.^4,6 The ability of interventions to improve QOL through health behavior changes is uncertain. Improved QOL following an exercise intervention has been reported in breast cancer survivors,²⁹ however a meta-analysis including 8 studies of ovarian and endometrial cancer survivors noted only non-significant improvements in QOL following physical activity interventions.³⁰

While previous research has examined the prevalence of health behaviors and associations with quality of life in survivors of other cancers, few studies have focused specifically on the importance of lifestyle factors for cervical cancer survivors. The purpose of this study is to describe the prevalence of health behaviors (physical activity, smoking, and alcohol consumption) among cervical cancer survivors participating in a randomized clinical trial of psychosocial telephone counseling and examine potential health behavior associations with QOL. We will also examine associations between longitudinal change in health behaviors and change in QOL, depression, anxiety, and psychological distress over follow-up.

Methods

Research Design and Study sample

Following approval by the Institutional Review Boards of the University of California, Irvine and California Cancer Registry (CCR), cervical cancer survivors diagnosed between 2008 and 2012 were identified through the CCR and recruited to participate in a randomized psychosocial telephone counseling trial. Eligible subjects were diagnosed with stage I to IVa cervical cancer, had completed definitive treatment at least two months prior to contact, were free of disease and able to speak and read English or Spanish. All patients provided informed consent consistent with federal, state and local requirements and were subsequently randomized to psychosocial telephone counseling (PTC) or usual care (UC). The intervention, based on the transactional model of stress and coping, included a QOL/psychosocial interview, 3 sessions focused on managing stress and emotions, health and wellness, and relationship/sexuality concerns, a summary/integration session, and a final booster session to review progress. Further detail is provided elsewhere³¹.

Data Collected

Questionnaires were completed via mail or telephone at baseline and approximately 4 and 9 months later. Data collected included sociodemographics, patient-reported outcome (PRO) measures and health behaviors.^31,32 Age, ethnicity, marital status, education, and income data were collected on the baseline questionnaire. Comorbidities prior to cancer diagnosis were self-reported by patients using a 29-item checklist. Disease stage was derived from the CCR database. Treatment data were self-reported by patients at baseline and validated by comparison to the CCR data. At baseline and follow-up, patients were asked about amount and duration of smoking, quantity of alcohol consumed, weekly hours of physical activity, and changes in diet following cancer diagnosis. Health behaviors were subsequently classified as meeting recommended guidelines or not based on published recommendations by the American Cancer Society (ACS).^7,33 Recommendations include no smoking, consumption of ≤7 alcoholic drinks per week for women, and at least 150 minutes of moderate intensity physical activity per week.

Measures

Overall QOL was assessed by the 27-item FACT-Cx (Functional Assessment of Cancer Therapy-Cervical), a multidimensional, combined generic and disease-specific QOL questionnaire that includes the FACT-General (FACT-G) questionnaire (version 4) consisting four subscales (physical, social, emotional, and functional well-being) plus a cervical-cancer specific additional concerns subscale.³⁴ Subdomains are summed to create the FACT-Cx score (Cronbach’s alpha=0.92). High scores indicate better quality of life. Emotional distress was assessed by the Patient-Reported Outcomes Measurement Information System (PROMIS) (www.PROMIS.org) short form consisting of 8 items on depression and 7 items on anxiety. Items are rated on a 5-point Likert scale and summarized in separate summary scores for depression and anxiety. Standardized T-scores are normed to the general population with mean=50 and standard deviation (SD) =10, with higher scores indicating more distress. The Brief Symptom Inventory (BSI-18) is an 18-item measure of psychological distress including an overall total score and subscales to measure anxiety, depression and somatization^35,36 Standardized scores are normed to the general population with mean=50 and SD=10. Higher scores indicate greater psychological distress.

Statistical Methods

Participants s were classified according to whether they did or did not meet ACS recommended guidelines for smoking, alcohol consumption and exercise at baseline, individually and in aggregate. Associations between health behaviors (independent variables) and baseline QOL, depression, anxiety and distress (dependent variables) were analyzed using general linear models. Covariates included in the model were age, ethnicity, treatment, stage, comorbidities and time from diagnosis to study. The assigned study arm at randomization was not included in the analysis of baseline data. Participants were subsequently classified by change in health behaviors at the 4-month follow-up. Associations between change in health behaviors and change in PROs after adjusting for covariates were analyzed using general linear models. Effect size was calculated as the difference between subgroup means divided by the pooled SD. Effect sizes of magnitude 0.33 to 0.5 are considered clinically important differences³⁷. Because the number of participants reporting change in health behaviors was small and thus power was limited, the PTC and UC arms were combined to examine associations between change in health behavior and patient-reported outcomes.

Results

A total of 204 women completed the baseline evaluation as part of the randomized clinical trial. Demographic and clinical characteristics of the study participants are depicted in Table 1. The age of participants at diagnosis ranged from 20–72 years, with a mean of 43.1 years (SD: 9.6). The majority of participants were non-Hispanic white (52%), while 41% of participants were Hispanic. Overall, 40% had a high school education or less. Nearly three-fourths of women were diagnosed with stage I (73%) cervical cancer. Treatment included surgery only for 49% of patients, radiation alone for 7%, and both chemotherapy and radiation for 44%.

Table 1.

Baseline Characteristics of Study Participants (N=204)

Characteristic	Mean	SD
Age at baseline evaluation (yrs)	44.7	9.6
Age at cancer diagnosis (yrs)	43.2	9.6
	N	%
Race/Ethnicity
Caucasian/Non-Hispanic	105	51.7
African-American	5	2.5
Hispanic	81	39.9
Asian/Pacific Islander	11	5.4
Native American	1	0.5
Stage
I	146	71.9
II–IVa	54	26.6
Education
< High School	43	21.1
High School graduate	39	19.2
Some college	56	27.6
College graduate	29	14.3
Graduate school	34	16.7
Marital Status
Single	31	15.3
Married/living with partner	129	63.5
Separated	9	4.4
Divorced	28	13.8
Widowed	5	2.5
Income
< $15,000	51	25.1
$15,000–$24,999	17	8.3
$25,000–$34,999	15	7.4
$35,000–$44.999	15	7.4
$45,000–$54,999	10	4.9
≥ $55,000	66	32.5
Treatment
Surgery only	100	49.0
Radiation only	15	43.6
Chemotherapy +/− radiation	89	7.4
History of multiple comorbidities prior to cancer diagnosis	64	31.5

Self-reported health behavior practices at baseline (9–30 months post diagnosis) are summarized in Table 2. Fifteen percent of participants were current smokers, while 4% self-reported high alcohol consumption. National recommendations for physical activity were not met by 62.8% of participants. When the health behavior practices were aggregated, the majority of women (67.4%) reported having at least one health behavior practice below the recommended standard, while 14% reported two or more deficient health behaviors.

Table 2.

Baseline Health Behaviors of Study Participant (N=204)

Health Behavior	N	%
Smoking Status
Non Smoker	171	85.5
Current Smoker	29	14.5
Number of Alcoholic Drinks per week
None	94	46.3
3 or less	79	38.9
4–10	19	9.4
11–17	8	3.9
Number of Hours of Physical Activity per week
None	64	31.5
Less than 1	21	10.3
1–2	40	19.7
3–4	35	17.2
5–7	30	14.8
8–12	4	2.0
12+	5	2.5
Multiple Health Behavior Risks
0 Behaviors Risks	63	32.6
1 Behavior Risk	104	53.9
≥2 Behaviors Risks	26	13.5

Baseline scores for overall quality of life, depression, anxiety, and psychological distress were compared between subgroups defined by meeting or not meeting recommendations for health behaviors after adjusting for the covariates age, ethnicity, treatment, stage, comorbidities and time from diagnosis to study (Table 3). There were no significant differences in patient-reported outcomes by smoking status or alcohol use however power to detect differences was limited by the small number of smokers (n=28) and high alcohol consumers (n=8). Compared to women who exercised less than 3 hours total per week, women who met national guidelines for physical activity had significantly higher quality of life scores (128.0 vs. 117.9, p=0.002) along with significantly lower depression (50.8 vs. 55.0, p=0.003) and anxiety scores (51.7 vs. 55.5, p=0.020). Further, women who met recommended lifestyle standards (i.e., not smoking, limited alcohol and exercise) reported significantly higher QOL (p=0.013) and lower depression (p=0.008) than those reporting one or more unhealthy behaviors. Lower anxiety and distress were also observed in those who met recommended standards however differences did not reach statistical significance. For QOL, there was a significant decreasing trend with increasing numbers of health risk behaviors (p=0.030). Examination of pairwise differences showed that QOL decreased significantly with the first unhealthy behavior, however the decline in QOL with additional health risk behaviors was not significant. Trends in depression, anxiety and distress with increasing numbers of health risk behaviors did not reach statistical significance.

Table 3.

Comparison of Quality of Life, Depression and Distress by Health Behaviors at Baseline

		FACT-Cx		Depression		Anxiety		BSI
	N	Mean	SE	Mean	SE	Mean	SE	Mean	SE
Smoking Status
Non-smoker	169	122.0	2.2	53.6	0.9	54.7	1.1	51.6	1.1
Current Smoker	28	116.8	4.5	53.4	2.0	50.9	2.3	52.3	2.3
		p=0.266		p=0.893		p=0.100		p=0.765
Alcohol Consumption
≤10 drinks/week	189	121.4	2.1	53.4	0.9	54.0	1.0	51.6	1.0
>10 drinks/week	8	121.7	8.0	57.0	3.5	60.2	4.0	56.5	4.0
		p=0.976		p=0.304		p=0.130		p=0.235
Physical Activity
≤2 hours/week	123	117.9	2.2	55.0	1.0	55.5	1.1	53.0	1.2
≥3 hours/week	73	128.0	2.9	50.8	1.3	51.7	1.5	49.8	1.5
		p=0.002		p=0.003		p=0.020		p=0.056
Multiple Risk Factors
No risk factors	63	128.2	3.2	50.1	1.4	51.3	1.6	49.4	1.7
1 risk factor	101	118.7	2.4	55.0	1.1	55.7	1.2	52.6	1.3
≥2 risk factors	26	116.8	4.8	54.0	2.1	53.0	2.4	53.6	2.5
		p=0.013		p=0.008		p=0.051		p=0.142

Means adjusted for age, ethnicity, treatment, comorbidities, stage and time from diagnosis to study

At the 4-month follow-up time point, 31 of 159 returning subjects (19.5%) reported improving their health risk behaviors to meet at least one additional guideline for health behaviors, 116 (73%) reported no change and 12 (7.5%) met fewer guidelines (decreasing positive health behaviors). Longitudinal changes in patient reported outcomes from baseline to the 4-month follow-up visit were compared by change in health behavior (Figure 1) after adjusting for age, ethnicity, treatment, stage, comorbidities and time from diagnosis to study. Mean FACT-Cx scores improved for all subjects over time, however the amount of improvement differed significantly between those who improved their health behaviors (+10.8) and those who did not (+4.5, p=0.026). Changes in depression, anxiety and distress showed similar patterns, with greater improvement among those who improved their health behaviors compared to those who did not improve, however trends did not reach statistical significance. Although the intervention was not designed to change health behaviors, 25% of PTC subjects (n=20) improved health behaviors compared to 13% of controls (n=10). Subjects in both study arms who improved health behaviors had higher QOL than those who made no changes or adopted worse behaviors. Differences between arms were not statistically significant (p=0.89), however the number of subjects improving their health behaviors was small and power to detect differences between PTC and UC was limited.

Figure 1. Change in Quality of Life, Depression and Distress by Change in Health Behaviors.

This Figure illustrates change in QOL and patient-reported outcomes between subjects who improved their health behaviors over the 4 month follow-up (n=30) and those who did not improve (n=128). Increasing values for QOL (Fact-Cx) indicate improvement whereas decreasing values for depression, anxiety and distress indicate improvement.

Discussion

The high prevalence or health risk behaviors not meeting the recommended standard in our sample of cervical cancer survivors who had completed treatment was comparable to rates observed in other studies of cancer survivors. Current smoking was reported by 15% of our sample compared to 21% observed by Beesley et al in cervical cancer survivors,⁴ and 10% in healthy women participating in the 2009 California Health Interview Survey (CHIS).³⁸ Approximately 63% of our sample did not meet recommended guidelines for physical activity, similar to the 61%–70% reported in other studies of cancer survivors^4,6 and to the 68% reported for healthy California women in the 2005 CHIS.³⁹ While alcohol consumption is difficult to compare due to the categorical response, 9% of cervical cancer survivors consumed 4–10 alcoholic drinks per week while 4% reported consuming >10 per week. In other studies, the percentage of women consuming more than the recommended limit for alcohol ranged from 5–12%.^10,26,40 Overall, two thirds of cervical cancer survivors did not meet guidelines for at least one health behavior, and 13% did not meet guidelines for two or more.

Physical activity at baseline was strongly associated with baseline QOL in our study, with higher QOL on the FACT-Cx for those meeting recommended guidelines compared to those who did not (128 vs 118, p=0.002). This difference of 10 points in QOL is equal to an effect size of 0.5 SD and is considered a clinically meaningful difference.³⁷ Those who exercised at least 3 hours per week also experienced less depression (p=0.003), anxiety (p=0.020) and distress (p=0.056). These results are consistent with other evidence showing a beneficial effect on health-related QOL for moderate exercise.^4,6,22–30 Associations between QOL and both smoking and alcohol consumption at baseline were non-significant, a finding consistent with other studies of cancer survivors.^4,6,28 Current smokers reported non-significantly lower QOL at baseline (117 vs 122, p=0.266) and lower anxiety (51 vs 55, p=0.100) but no differences in depression or distress. While excessive alcohol consumption was not associated with lower QOL, cervical cancer survivors who consumed >10 drinks per week reported higher levels of depression, anxiety and distress. Differences did not reach statistical significance but effect sizes ranged from 0.4 to 0.8 suggesting these differences may be clinically meaningful.

A significant cumulative effect on QOL was observed with increasing numbers of health risk behaviors. Cervical cancer survivors with no deficient health behaviors reported the highest QOL (FACT-Cx=128) with progressively lower QOL reported with one (FACT-Cx=119) or more (FACT-Cx=117) deficient health behaviors (p for trend=0.030). Although trends in depression, anxiety and distress with increasing number of health behaviors not meeting the recommended standard did not reach statistical significance, results are consistent with observations for survivors of other cancers. Significant decreasing trends in QOL with increasing numbers of deficient health behaviors including physical inactivity, diet, obesity and/or smoking have been reported for survivors of ovarian,²⁵ endometrial,²⁶ colorectal²⁸ and other cancers.⁶ In breast cancer survivors, increasing numbers of modifiable health risk factors were associated with decreased time to recurrence and death.⁴¹ Our results are consistent with Blanchard et al.⁶ and suggest that physical activity may be more important than other lifestyle factors for improving QOL. This has clear implications for cervical cancer survivors, who may be able to enjoy a more robust post-treatment life with a moderate physical activity schedule.

A strength of our study was our ability to examine longitudinal changes in health behaviors during the post-treatment survivorship period. Women with cervical cancer who improved their health behaviors over the 4-month follow-up experienced a two-fold greater improvement in QOL than did those women who made no changes or who increased their risk-prone behaviors (p=0.026). While an intervention study focused on physical activity in breast cancer survivors did result in improved QOL,²⁹ a meta-analysis of intervention studies designed to improve lifestyle factors for ovarian and endometrial cancer survivors did not show significant improvement in QOL.³⁰ Further confirmation of the longitudinal effect of changes in health behaviors on QOL in female cancer survivors is needed.

Our results are not without limitations. Health behaviors were collected as categorical data and categories did not always correspond to recommended national guidelines which were published after design, approval and funding of the clinical study. Numbers of subjects reporting high alcohol consumption in particular were small, resulting in limited power to detect associations with QOL. Further, we were unable to examine associations between diet and QOL with our data. While some questions addressed change in diet after cancer diagnosis and treatment, baseline dietary characteristics were not collected. Thus, we could not determine if diet or dietary changes reflected national guidelines. Data on health behaviors at the 4-month follow-up were available for only a subset of our sample (n=158). Furthermore, health behaviors changed for only a small number over this 4-month period, improving for 30 and getting worse for 12. Power to test the effect of changes in health behaviors on change in QOL was thus limited, and power to test for differences between the PTC and UC arms was even more limited. While one of six counseling sessions addressed some issues related to health and wellness, modifying health behaviors was not a primary focus of the intervention. Future studies which are adequately powered to test interventions containing elements designed specifically to change health behaviors in cervical cancer survivors would be of interest.

In conclusion, our results show that poor health behaviors which may contribute to cancer recurrence or early death are prevalent in cervical cancer survivors. Large numbers of cervical cancer survivors are not meeting national guidelines for smoking, physical activity and alcohol consumption. These behaviors are associated with impaired QOL, depression and distress in cervical cancer survivors. Changes in lifestyle behaviors during cancer survivorship resulted in significant improvement in QOL. Interventions designed to moderate lifestyle and health behaviors during cancer survivorship may play an important role in improving well-being for cancer survivors.