Demographic and Clinical Factors Associated with Health-Related Quality of Life Profiles among Prostate Cancer Survivors

Abstract

PURPOSE

Our purpose was to describe the prevalence and predictors of symptom and function clusters relating to physical, emotional, and social components of general health-related quality of life (HRQOL) in a population-based sample of prostate cancer survivors.

METHODS

Participants (n=1162) completed a baseline survey at a median of nine months after diagnosis to ascertain the co-occurrence of eight symptom and functional domains that are common across all cancers and not treatment-specific. We used latent profile analysis (LPA) to identify subgroup profiles of survivors with low, moderate, or high HRQOL levels. Multinomial logistic regression models were used to identify clinical and sociodemographic factors associated with survivors’ membership in the low vs. moderate or high HRQOL profile.

RESULTS

The LPA identified 16% of survivors who were categorized in the low HRQOL profile at baseline, indicative of the highest symptom burden and lowest functioning. Factors related to survivors’ membership in the low vs. higher HRQOL profile groups included less than age 65 at diagnosis, identifying as non-Hispanic Black race, not working, being a former vs. never smoker, systemic therapy, less companionship, more comorbidities, lower healthcare financial well-being, or less spirituality. Several factors remained associated with remaining in the low vs. higher HRQOL profiles on the follow-up survey (n=699), including younger age, Black race, comorbidity, and lower financial and spiritual well-being.

CONCLUSION

About one of six prostate cancer survivors experienced elevated physical and psychosocial symptoms that were independent of local curative therapy, but with younger age, race, comorbidity, and lower financial and spiritual well-being as stable risk factors for poor HRQOL over time.

Introduction

There are more than 3.5 million prostate cancer (PCa) survivors in the United States,¹ and ensuring their health-related quality of life (HRQOL) following diagnosis and treatment is essential. There has been extensive research on PCa treatment-related symptom and functional deficits specific to the urinary, bowel, and sexual domains.^2–4 Prior studies of PCa survivors’ physical, cognitive, emotional, and social well-being have analyzed these health domains separately using older legacy health measures.^4–7

Our conceptual approach defines HRQOL as the presence of multiple symptom and function domains that are highly prevalent and impactful in persons with any chronic condition, including cancer.^{8, 9} In contrast to prior studies, we used latent profile analysis (LPA) to simultaneously assess eight HRQOL domains, using contemporary, state-of-the-art health measures derived using modern measurement theory.^{10, 11} We categorized PCa survivors into sub-groups (i.e., profiles) based on their scores on these eight HRQOL domains. This methodology allowed us to identify demographic, clinical, and treatment factors associated with membership in HRQOL holistically. In addition, we evaluated the stability of HRQOL profiles over time among survivors who completed a follow-up survey six months after the baseline survey.

Because most PCa survivors have a prolonged survival period, studies of common and impactful HRQOL domains during the early phase of survivorship are necessary to identify survivors who are at risk for poor outcomes for earlier management and supportive care. Thus, our goal is to inform new and better strategies to identify at-risk PCa survivors to help mitigate the potential adverse impacts of symptoms and functional deficits during the survivorship period.

Methods

Participants and Data Collection Procedures

This is a secondary analysis of previously collected data. Adult participants were enrolled in the Measuring Your Health (MY-Health) study.¹² During 2010–12, four population-based cancer registries (New Jersey State Cancer Registry, Louisiana Tumor Registry, Cancer Registry of Greater California, Greater Bay Area Cancer Registry), which are part of the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program, enrolled 5506 participants within 6 to 13 months (median 9 months) after they were diagnosed with, and initially treated for, one of seven different cancer types. There were 1162 participants with primary invasive cancer of the prostate gland (all stages) who formed the cohort for this analysis. The MY-Health study oversampled younger age groups and individuals from different racial and ethnic groups to ensure a heterogeneous sample. Enrolled participants completed a baseline survey via mail, ranging from 6–13 months post-diagnosis. A follow-up survey was conducted six months after the completion of the baseline survey. More details on the study design, response rates, and representativeness of the MY-Health study cohort are reported elsewhere.¹² The study was approved by all participating sites’ IRBs.

Patient Characteristics

We extracted clinical data including the date of cancer diagnosis, cancer type, cancer stage, and initial locally curative therapy (surgery and radiation therapy) from collaborating cancer registry databases. The survey obtained information on patient-reported sociodemographics (e.g., education, race, ethnicity), financial information (e.g., insurance coverage, working status, healthcare related financial well-being¹³), health behaviors (e.g., smoking status, BMI), social support (marital status, ability to find companionship when needed), spirituality,¹⁴ and comorbid conditions. Participants also reported on receipt of systemic hormonal and chemotherapies. In the follow-up survey, participants were asked if they had experienced a recurrence or progression of their cancer.

Outcomes: Symptoms and Functioning

The Patient-Reported Outcomes Measurement Information System^® (PROMIS^®) measures included the short-form versions of fatigue, pain interference, anxiety, depression, sleep disturbance, physical function, ability to participate in social roles (social function), and cognitive function.^{12, 15} All PROMIS T-scores have a mean of 50 and a standard deviation (SD) of 10 points. Higher PROMIS function T-scores represent better functioning, and higher PROMIS symptom T-scores reflect greater symptom burden. PROMIS score differences of 3–6 points (close to 1/2 SD of 5 points) or more have been found to be a minimally important clinical difference.^{16, 17}

Latent Profile Analysis

We used latent profile analysis (LPA) to identify subgroups (profiles) of PCa survivors responding to the baseline survey (n=1162) with similar scores across the eight PROMIS symptom or function indicators. LPA does not require that the indicators included in the analyses be correlated, so it is possible for the model to identify a profile subgroup with poor levels or high levels across all 8 domains, or a subgroup with low physical health scores but “normal” mental health scores. To determine the ideal number of profiles, we compared hierarchically-nested models with increasing number of survivor group profiles starting with two profiles. The ideal number of profiles was based on the models’ goodness-of-fit, having at least 10% survivors in each profile group, and clinical interpretability.^{18, 19} Clinical evaluations of interpretability were provided by a radiation oncologist (RC), an oncology clinical nurse specialist (JFD), and a clinical health psychologist (KDG). Subsequently, using a subset of participants who completed both baseline and follow-up surveys (n=699), we used latent profile transition analysis (LPTA) to identify three subgroup profiles on the baseline and follow-up surveys.²⁰ This LPTA profile identification is equivalent to conducting two separate LPA on the baseline and follow-up surveys, but imposes the condition that between the two timepoints the LPA profiles have similar mean scores across all eight PROMIS domains. LPA and LPTA were performed with M-plus (Ver. 8.2).

Multivariable Analysis

Multivariable adjusted multinomial logistic regression models were used to determine the demographic and clinical factors associated with HRQOL profile membership on both the initial survey (median 9-months post-diagnosis) and on the follow-up survey 6 months later. We explored selected interaction terms focusing on age, comorbidity, and race and ethnicity with each other and with treatment types. We used type III likelihood ratio-tests to assess each interaction term separately. In addition to the covariates used for the baseline survey model, we added another covariate representing secondary treatments for disease recurrence or progression occurring between the two surveys for the follow-up survey model. We summarized all results using model-adjusted odds ratios (OR) with 95% confidence intervals (CIs). Approximately 5% of all survivors had at least some missing data across variables. We assumed these data were missing at random and we used all data that was available to estimate the models using full information maximum likelihood. All of the regression models were performed using SAS (version 9.4).

Results

There were 1162 PCa survivors included in the analysis of the baseline survey, and 699 survivors who completed both the baseline and follow up surveys. As shown in the first column of Table 1, the baseline survey sample is sociodemographically diverse, including approximately 39% non-Hispanic White (NHW), 28% non-Hispanic Black (NHB), 21% Hispanic, and 12% Asian PCa survivors; 19% had less than a high school education.

Table 1:

Characteristics of Prostate Cancer Survivors on the Baseline Survey by Health-Related Quality of Life (HRQOL) Profile

Variable	Level or Statistic	Overall (N=1162)@	Low HRQOL (N=187)	Moderate HRQOL (N=459)	High HRQOL (N=516)
Age at Diagnosis	<65	610 (52%)	107(57%)	226(49%)	277(54%)
	≥65	552 (48%)	80(43%)	233(51%)	239(46%)
Racial-Ethnic Group	Non-Hispanic White	440 (39%)	47 (25%)	176 (38%)	217 (42%)
Educational Attainment	Non-Hispanic Black	316 (28%)	70 (37%)	111 (24%)	135 (26%)
	Hispanic	233 (21%)	44 (24%)	100 (22%)	89 (17%)
	Non-Hispanic Asian	139 (12%)	16 (9%)	60 (13%)	63 (12%)**
	< High School Grad	219 (19%)	62 (33%)	82 (18%)	75 (15%)
	High School Grad	551 (48%)	94 (50%)	226 (49%)	231 (45%)
	College or Graduate Degree	376 (33%)	29 (16%)	146 (32%)	201 (39%)**
Marital Status	Not Married/Single	274 (24%)	77 (41%)	99 (22%)	98 (19%)
	Married/Living with partner	877 (76%)	107 (57%)	358 (78%)	412 (80%)**
Working Status	Working	513 (44%)	38 (21%)	190 (41%)	285 (55%)
	Not Working	642 (56%)	145 (79%)	268 (59%)	229 (45%)
Insurance	Private/Private + Government	806 (71%)	95 (52%)	308 (68%)	403 (79%)
	Government/No Insurance	324 (29%)	85 (46%)	139 (31%)	100 (20%)**
Smoking Status	Current smoker	137 (12%)	40 (21%)	50 (11%)	47 (9%)
	Former smoker	420 (37%)	70 (37%)	178 (39%)	172 (33%)
	Never smoker	591 (51%)	72 (39%)	229 (50%)	290 (56%)**
Stage at Diagnosis	In situ/Stage I	294 (28%)	30 (16%)	117 (25%)	147 (28%)
	Stage II	654 (61%)	111 (59%)	249 (54%)	294 (57%)
	Stage III/IV	117 (11%)	37 (20%)	81 (18%)	59 (11%)**
Radical Prostatectomy	Yes	487 (42%)	53 (28%)	185 (40%)	249 (48%)
	No	675 (58%)	134 (72%)	274 (60%)	267 (52%)**
Radiation Therapy	Yes	489 (42%)	98 (52%)	207 (45%)	184 (36%)
	No	673 (58%)	89 (48%)	252 (55%)	332 (64%)**
Systemic Therapy	Yes	291 (25%)	76 (41%)	138 (30%)	77 (15%)
	No	871 (75%)	111 (59%)	321 (70%)	439 (85%)**
Comorbidities&	None	436 (38%)	28 (15%)	150 (33%)	258 (50%)
	One	329 (28%)	31 (17%)	138 (30%)	160 (31%)
	Two or more	397 (34%)	128 (68%)	171 (37%)	98 (19%)**
Companionship	Never/Rarely	126 (11%)	44 (24%)	49 (11%)	33 (6%)
	Sometimes	172 (15%)	57 (30%)	85 (19%)	30 (6%)
	Often/Always	835 (74%)	77 (41%)	315 (69%)	443 (86%)**
Financial Well-Being13	Mean (SD)		50.1 (23.6)	60.2 (24.5)	68.7 (24.7)***
Spirituality14	Mean (SD)		30.6 (9.6)	36.9 (8.6)	41.5 (7.2)***

^@Some row variables do not include all 1162 participants due to missing, unknown, refused to answer, or “other” values. The percent of missing values ranged from 0–5% for row variables.

^&Participants were asked if they had ever been told by a health care professional that they had any of 15 different chronic conditions including heart-related conditions (heart attack, heart failure, stroke), asthma, chronic obstructive pulmonary disease (COPD), depression, anxiety.

P-values were calculated by comparing the three profile groups on distribution of non-missing values only. We used Pearson chi-square tests for all categorical row variables, and chi-square rank based group means score statistics for all continuous/ordinal row variables.

^***Indicates P<0.001

^**indicates P<0.01

^*indicates P<0.05

PCa Survivor HRQOL Subgroup Profiles

The 3-profile solution from the LPA model was empirically the appropriate statistical fit with an entropy of 0.897, while also providing a parsimonious and clinically interpretable solution with a sufficient sample size of survivors in each profile (at least 10% in each profile; see Table 1S, online). The three profiles consisted of low, moderate, and high HRQOL subgroups. As shown in Figure 1, those in the low HRQOL profile (red line), including approximately 16% of the PCa survivors, reported impaired functional status with mean PROMIS T-scores ranging between 36 and 41 and high symptom levels (PROMIS T-scores) ranging from 58 to 64. The high HRQOL profile (green line), including approximately 45% of survivors, reported PROMIS function domain T-scores ranging from 56 to 64 and low symptom burden ranging from 39 to 43, nearly an entire standard deviation below the US average scores for the PROMIS measures (e.g., better functioning and less symptom burden). The moderate HRQOL profile (yellow line) fell between the low and high HRQOL profiles and included 39% of the survivors, with mean PROMIS functioning T-scores ranging from 46 to 54 and symptom burden ranging from 47 to 51. Because the 8 PROMIS domains scores were highly correlated across both physical and mental health domains, the three profile groups we observed only differed according to severity or impact with low, moderate, and high HRQOL cluster groups.

Figure 1. Latent Profile Analysis 3-Profile Group Result.

(N=1162 Prostate Cancer Survivors, Baseline Survey)

X-axis: Health-Related Quality of Life (HRQOL) Domains

Left Y-axis: PROMIS T-Score for Function Domains

Right Y-axis: PROMIS T-Score for Symptom Domains

PROMIS^® measures use a T-score metric in which 50 is the mean of a relevant reference population and 10 is the standard deviation (SD) of that population.

PROMIS^®= Patient Reported Outcomes Measurement Information System

Characteristics Associated with Profile Membership: Baseline Survey

Table 1 shows the frequency distribution of the clinical and sociodemographic characteristics across the three HRQOL profiles. Table 2 shows the adjusted ORs and 95% CIs from the adjusted multinomial logistic regression model for each characteristic having a statistically significant association (p<0.05) with profile membership on the baseline survey. Given all other variables are fixed, characteristics of survivors more likely to be in the low HRQOL profile compared with the high HRQOL profile included: being <65 years of age at diagnosis (OR = 2.98, 95% CI 1.71–5.21; vs. ≥65); identifying as NHB race (OR = 2.63, 1.38–5.03; vs. NHW race); not currently working (OR = 3.48, 1.95–6.22); being a former smoker (OR = 2.20, 1.30–3.71; vs. nonsmoker); treated with systemic therapy (OR = 2.71, 1.49–4.94); having two or more comorbidities (OR = 9.24, 5.12–16.7; vs. none); and never/ rarely (OR = 2.62, 1.32–5.20; vs. often/always) or sometimes (OR = 4.95, 2.57–9.55) being able to find companionship when needed. Factors associated with being less likely to be in the low HRQOL vs. high HRQOL profile group included: greater healthcare related financial well-being (OR = 0.81, 0.72–0.92 for a half SD [13 points] increase in financial well-being score), and more spirituality (OR = 0.52, 0.46–0.60 for a half SD [4 points] increase in spirituality score). Notably, surgery, radiation therapy, and stage at disease were not associated with HRQOL profile membership. We found no significant interaction terms.

Table 2.

Multinomial Logistic Regression Analysis of Factors Associated with Prostate Cancer Survivors’ Membership in Health-Related Quality of Life Profiles on the Baseline Survey^*

Variable (# Denotes Reference Category)	Comparator vs Reference HRQOL Profile	Adjusted Odds Ratio (95% Wald CI)
Age at Diagnosis
<65 vs ≥65#	Moderate vs High Low vs High Low vs Moderate	1.08 (0.76, 1.55) 2.98 (1.71, 5.21) 2.75 (1.67, 4.54)
Racial-Ethnic Group
Hispanic vs Non-Hispanic White#	Moderate vs High Low vs High Low vs Moderate	1.45 (0.93, 2.25) 1.95 (0.96, 3.95) 1.34 (0.71, 2.55)
Asian vs Non-Hispanic White#	Moderate vs High Low vs High Low vs Moderate	1.14 (0.69, 1.88) 1.50 (0.64, 3.51) 1.31 (0.60, 2.88)
Black vs Non-Hispanic White#	Moderate vs High Low vs High Low vs Moderate	1.31 (0.88, 1.95) 2.63 (1.38, 5.03) 2.01 (1.11, 3.64)
Working Status
Not Currently Working vs Working#	Moderate vs High Low vs High Low vs Moderate	1.32 (0.94, 1.85) 3.48 (1.95, 6.22) 2.63 (1.54, 4.50)
Smoking Status
Current smoker vs Never smoker#	Moderate vs High Low vs High Low vs Moderate	1.06 (0.64, 1.75) 1.64 (0.80, 3.35) 1.44 (0.77, 2.71)
Former smoker vs Never smoker#	Moderate vs High Low vs High Low vs Moderate	1.32 (0.96, 1.82) 2.20 (1.30, 3.71) 1.66 (1.03, 2.69)
Systemic Therapy
Yes vs No#	Moderate vs High Low vs High Low vs Moderate	2.06 (1.35, 3.14) 2.71 (1.49, 4.94) 1.32 (0.78, 2.22)
Companionship
Never/Rarely vs Often/Always#	Moderate vs High Low vs High Low vs Moderate	1.37 (0.82, 2.29) 2.62 (1.32, 5.20) 1.91 (1.06, 3.45)
Sometimes vs Often/Always#	Moderate vs High Low vs High Low vs Moderate	2.92 (1.76, 4.84) 4.95 (2.57, 9.55) 1.70 (1.01, 2.86)
Comorbidities
One vs None#	Moderate vs High Low vs High Low vs Moderate	1.49 (1.06, 2.11) 1.65 (0.85, 3.22) 1.11 (0.58, 2.10)
Two or more vs None#	Moderate vs High Low vs High Low vs Moderate	3.00 (2.08, 4.34) 9.24 (5.12, 16.7) 3.08 (1.80, 5.28)
Financial Well-Being13 (per ½ SD increase)
	Moderate vs High Low vs High Low vs Moderate	0.89 (0.82, 0.97) 0.81 (0.72, 0.92) 0.91 (0.81, 1.02)
Spirituality (FACIT-SP)14 (per ½ SD increase)
	Moderate vs High Low vs High Low vs Moderate	0.73 (0.66, 0.80) 0.52 (0.46, 0.60) 0.72 (0.64, 0.80)

^*There were 1091 observations used (out of a total of 1162 baseline survey respondents) in the model due to missing data on some of the covariates. The full regression model adjusted for age, race and ethnicity, marital status, educational attainment, SEER region, working status, insurance coverage, smoking status, stage at initial diagnosis, radical prostatectomy, radiation therapy, systemic therapy (hormonal or chemotherapy), ability to find companionship, comorbidities, healthcare related financial well-being, and spirituality. The likelihood ratio test was conducted to calculate the p-value of each variable. Only variables significant at p<.05 level are shown in the table.

Characteristics Associated with Profile Membership: Follow-up Survey

We next examined the transitions in HRQOL profile membership among the 699 PCa survivors who completed both surveys. Membership in the HRQOL profiles remained relatively stable for most PCa survivors over time. About 72% who started in the moderate or high HRQOL profiles groups remained in those groups, and another 15% (n=107) who had been in the low HRQOL profile remained in that profile group. Few survivors (n=86) transitioned into or out of a higher or lower HRQOL group, so we grouped those who transitioned to the lowest group with those who remained in that group over time.

Figure 1S (online) shows the 3-profile LPTA solution for the follow-up survey and Table 2S (online) shows the distribution of the three HRQOL profiles according to the clinical and sociodemographic characteristics. Table 4 shows the adjusted associations with profile membership from the multinomial regression model. We found similar associations to the baseline survey on the follow-up survey, with the exception that we did not observe any associations between systemic therapy, smoking status, or companionship with HRQOL on the follow-up survey. In addition, although we found no association between educational attainment and HRQOL on the baseline survey, on the follow-up survey we observed that survivors who had not completed high school, compared with college graduates or those with post-graduate degrees, were much more likely to be in the low HRQOL profile compared with the higher HRQOL profiles (OR = 8.68, 2.36–31.9 for low vs. high; and 8.40, 2.16–32.7 for low vs. moderate). We also observed that experiencing a recurrence or progression of cancer was associated with being in the low HRQOL profile vs. the high profile (OR = 2.35, 1.06–5.25).

Table 4.

Findings Associated with Membership in Lowest Health-Related Quality of Life Profile and Clinical Implications

Finding	Tailored Clinical Implication for Survivorship Care
Younger Age (< 65 years old)	Routine HRQOL screening;44 referral to lifestyle interventions to promote exercise and nutrition counseling45
Poor physical functioning	Referral to lifestyle interventions to promote walking,46 exercise and strength training;47 multidisciplinary management and care coordination
Poor Psychosocial Functioning	Routine distress screening; referral to mental health support services and/or peer support groups; 48, 49 mindfulness-based stress reduction50, 51
Greater Financial Toxicity	Connection with financial navigators and financial support services; referral to innovative financial toxicity intervention programs52
Poor Social Support	Referral to supportive care services;53 Implementation of evidence-based peer support programs54
Lower Spirituality	Implementation of evidence-based approaches to enhancing spirituality among prostate cancer survivors55

We compared those who did not complete the follow-up survey (n=463) versus those who completed both surveys (n=699). Responders to the follow-up survey were similar to non-responders, except that non-responders were slightly older, were less likely to have completed high school, and were less likely to be NHW compared with those who completed the follow-up survey. We assessed the sensitivity of our results due to non-response on the follow-up survey using an inverse-probability-weighting approach. Overall, we did not observe significant changes in the associations.

Discussion

In this uniquely large and sociodemographically diverse group of PCa survivors treated in diverse community settings, we identified three distinct general HRQOL profiles among PCa survivors experiencing meaningfully different levels of symptom burden and functional deficits. We observed that 16% of PCa survivors were in the low HRQOL profile, with PROMIS T-scores at least 10 points below the general US population average PROMIS scores for most of the eight HRQOL domains. As clinically meaningful differences on most PROMIS T-scores for cancer patients are in the 3–6 point range,¹⁶ the 10-point differences we observed for the low HRQOL group represent striking and clinically significant decrements in HRQOL. The low HRQOL group represents a group with substantial survivorship care needs across multiple correlated domains, and thus distinct from the high HRQOL group with likely lower clinical care needs. Thus, our findings underscore the need for an individualized clinical approach based on a better understanding of the composition of distinct subgroups, as reported in a prior cluster analysis of cancer survivors.²¹

Developing efficient strategies to identify, and provide additional healthcare to the survivors in the high needs, low HRQOL profile group is important. Indeed, creating ways to screen and monitor the well-being of PCa survivors could help them manage, or receive treatment for, their ongoing symptoms and functional limitations. Knowing which groups of survivors are more vulnerable to worse outcomes -- PCa survivors who are younger, identify as NHB race, have lower financial well-being, and greater comorbidities – can direct clinical attention and appropriate resources to mitigate negative outcomes.

We have assessed, using state-of-the-art psychometric analyses and outcome measures, the effects of PCa on general HRQOL during the early phase of survivorship. To our knowledge, although long-term generic HRQOL observational data have been reported,²² we are the first to have modeled multiple symptom and function domains as “profiles” in this survivor population that are common and impactful for survivors across most chronic health conditions. Our approach extends and complements earlier studies on the general HRQOL of early phase PCa survivors.

Most prior studies in the literature on PCa have focused on urinary, bowel, and sexual problems experienced in different treatment groups among men diagnosed with clinically localized PCa.^{2–4, 7} Decrements in these disease-specific domains have been associated with poorer general HRQOL using legacy scales, such as the SF-36.^{6, 7, 23, 24} Importantly, these studies found no associations between treatment type and general HRQOL, consistent with our findings for all stages of survivors, and with other systematic reviews of this topic.^{25, 26} One prior study of PCa survivors found that poorer overall physical health was associated with older age and more comorbidities, while poorer mental health was associated with younger age, more comorbidity, and higher stage at diagnosis.²⁷ Our study identifies sub-groups of PCa survivors who may be at-risk for poor overall health during their survivorship.

We examined the sociodemographic and clinical variables that were most closely associated with being in the low HRQOL profile. We found that younger age was associated with being in the low profile, even when adjusting for all other covariates. It is possible that this finding may be due partly to the unexpected challenges faced by younger PCa survivors, including managing multiple responsibilities at work and home, and greater negative effects of disease-related dysfunction (e.g., sexual, urinary) on psychosocial well-being in this group compared to older survivors.²⁸ Our findings that NHB men with PCa were at greater risk for membership in the low HRQOL compared to NHW men may be due to other factors, including inadequate access to high quality oncology care including supportive care following treatment, and medical mistrust.²⁹

We found a significantly higher risk of being in the low HRQOL profile among PCa survivors with worse healthcare-related financial well-being, which is consistent with the literature regarding the association between financial hardship (or toxicity) and symptom care and HRQOL in other cancer survivors.^{30, 31} For example, higher out-of-pocket costs for ambulatory care and prescription medications are likely to be significant barriers to obtain high quality symptom-related care. We also found that those not currently working were more likely to be in the low profile compared with those currently working, which may be associated with being temporarily unable to work, career disruption, and financial concerns. Prior studies suggest that greater symptom burden is associated with lower income, and with unemployment.^32–35 Our findings for spirituality being less associated with low HRQOL, compared to the high HRQOL profile, may be partly explained by having more effective coping mechanisms for dealing with symptoms, particularly mental health, having a greater sense of meaning and purpose, or being part of a faith community.^36–37 However, a systematic review found no associations between spirituality/meaning and psychological well-being among women with breast cancer.³⁸

We also observed a strong association of having comorbidities with being in the low HRQOL profile, consistent with prior studies of PCa showing that greater comorbidity burden is associated with worse HRQOL.^{27, 39, 40} Men who received systemic therapy (mostly hormonal therapy) were more likely to be in the low HRQOL profile than men who did not receive therapy, adjusting for stage and all other variables, which may relate to hormonal therapy increasing fatigue and affecting general physical function.⁴¹

While most associations were similar in the baseline and follow-up survey, one notable exception was for educational attainment, which was not associated with HRQOL on the baseline survey, but was very strongly associated with HRQOL profiles on the follow-up survey. This may be due to changes in the sample characteristics over time, with only 5 men with a college degree or more in the low HRQOL profile group; or that some individuals with less education face increasing barriers over time due to lower levels of health literacy or lack of confidence in self-management or requesting referrals to specialists for symptom and functional deficits.^{42, 43}

Our findings have several important clinical implications which are summarized in Table 4. First, our use of LPA of multiple symptom and function domains can be used to identify at-risk subgroups of PCa survivors who may require earlier identification and management of symptoms and functional deficits. Table 4 includes several examples of potential interventions, such as multidisciplinary management and care coordination for the subset of high risk PCa survivors to address co-occurring deficits in both physical and psychosocial well-being. The American Cancer Society recommends specialist and primary care provider collaboration with management of side effect burden, and ongoing assessment of long-term HRQOL effects to yield optimal survivorship care.⁴⁴

Our study has several limitations. We did not include domains (e.g., urinary, sexual, and bowel functioning) most commonly studied in prostate cancer. In addition, we had limited clinical detail on risk of recurrence, dose and duration of radiation and systemic therapies, and severity and timing of comorbidity. Other important characteristics potentially associated with HRQOL that we were unable to assess include pre-treatment HRQOL, and access to palliative care services. Our sample of cases diagnosed with regional or metastatic PCa was only 11%, and thus our results may not generalize well to all PCa survivors.

We found that there was a distinct subgroup of PCa survivors belonging to a low HRQOL profile, defined as experiencing co-occurring functional deficits and significant symptom burden reflecting reduced general physical, emotional, social, and cognitive health. Our findings may help to inform screening and earlier intervention for those survivors at the highest risk of poor health outcomes.

Supplementary Material

PV Appendix Figure 1

Figure 1S

Latent Profile Transition Analysis 3-Profile Group Result

(N=699 Prostate Cancer Survivors Who Completed Baseline and Follow-up Surveys)

X-axis: Health-Related Quality of Life (HRQOL) Domains

Left Y-axis: PROMIS T-Score for Function Domains

Right Y-axis: PROMIS T-Score for Symptom Domains

PROMIS^® measures use a T-score metric in which 50 is the mean of a relevant reference population and 10 is the standard deviation (SD) of that population.

PROMIS^®= Patient Reported Outcomes Measurement Information System

Table 3:

Multinomial Logistic Regression Analysis of Factors Associated with Prostate Cancer Survivors’ Membership in Health-Related Quality of Life Profiles on the Follow-up Survey^*

Variable (# Denotes Reference Category)	Comparator vs Reference HRQOL Profile	Adjusted Odds Ratio (95% Wald CI)
Age at diagnosis
<65 vs ≥65#	Moderate vs High Low vs High Low vs Moderate	1.16 (0.69, 1.94) 6.24 (2.75, 14.1) 5.40 (2.30, 12.7)
Racial-Ethnic Group
Hispanic vs Non-Hispanic White#	Moderate vs High Low vs High Low vs Moderate	1.38 (0.76, 2.52) 1.10 (0.40, 3.02) 0.80 (0.27, 2.33)
Asian vs Non-Hispanic White#	Moderate vs High Low vs High Low vs Moderate	1.38 (0.66, 2.88) 2.88 (0.82, 10.1) 2.09 (0.55, 7.99)
Black vs Non-Hispanic White#	Moderate vs High Low vs High Low vs Moderate	1.92 (1.11, 3.32) 2.54 (1.10, 5.88) 1.32 (0.54, 3.22)
Educational Attainment
< High School Grad vs College or Graduate Degree#	Moderate vs High Low vs High Low vs Moderate	1.03 (0.51, 2.11) 8.68 (2.36, 31.9) 8.40 (2.16, 32.7)
High School Grad vs College or Graduate Degree#	Moderate vs High Low vs High Low vs Moderate	1.37 (0.85, 2.21) 7.53 (2.52, 22.5) 5.50 (1.77, 17.1)
Progression or Recurrence
Yes vs No#	Moderate vs High Low vs High Low vs Moderate	1.76 (1.00, 3.09) 2.35 (1.06, 5.25) 1.34 (0.59, 3.04)
Comorbidities
One vs None#	Moderate vs High Low vs High Low vs Moderate	0.83 (0.50, 1.39) 3.07 (1.16, 8.13) 3.68 (1.32, 10.3)
Two or more vs None#	Moderate vs High Low vs High Low vs Moderate	1.30 (0.79, 2.11) 5.91 (2.39, 14.6) 4.56 (1.78, 11.7)
Financial Well-Being13 (per ½ SD increase)
	Moderate vs High Low vs High Low vs Moderate	0.88 (0.79, 0.98) 0.97 (0.82, 1.15) 1.10 (0.92, 1.31)
Spirituality (FACIT-SP)14 (per ½ SD increase)
	Moderate vs High Low vs High Low vs Moderate	0.86 (0.77, 0.95) 0.69 (0.59, 0.81) 0.81 (0.69, 0.96)

^*There were 657 observations used (out of a total of 699 follow-up survey respondents) in the model due to missing data on some of the covariates The full regression model adjusted for age, race/ethnicity, marital status, educational attainment, SEER region, employment status, health coverage, smoking status, stage at diagnosis, receipt of systemic therapy (hormonal or chemotherapy), ability to find companionship, progression or recurrence, healthcare financial well-being, spirituality and comorbidities. The likelihood ratio test was conducted to calculate the p-value of each variable; only variables statistically associated (p<.05) with the outcome (HRQOL profile) are presented in this table.

Context Summary.

Key Objective

To document the prevalence of concurrent symptoms and functional deficits (clusters) reflecting poorer general health-related quality of life (HRQOL) in early prostate cancer survivors, and evaluate numerous characteristics associated with symptom and function clusters.

Knowledge Generated

About 16% of early prostate cancer survivors were categorized as having low HRQOL as defined by experiencing concurrent symptoms and functional deficits corresponding to physical, emotional, and social domains. Factors related to lower HRQOL included younger age, being non-Hispanic black, receipt of systemic therapy, and more comorbidities. Modifiable factors associated with lower HRQOL included lower financial hardship, less companionship, and less spirituality.