Measurement of Patients’ Acceptable Symptom Levels and Priorities for Symptom Improvement in Advanced Lung Cancer

Abstract

Purpose:

Little research has assessed cancer patients’ success criteria and priorities for symptom improvement to inform patient-centered care. Thus, we modified and tested a measure of these constructs for advanced lung cancer patients. We compared acceptable severity levels following symptom treatment across eight symptoms and identified patient subgroups based on symptom importance.

Methods:

Advanced lung cancer patients (N=102) completed a one-time survey, including the modified Patient Centered Outcomes Questionnaire (PCOQ), standard symptom measures, and other clinical characteristics.

Results:

The modified PCOQ showed evidence of construct validity through associations with theoretically related constructs. Symptom severity and importance were moderately correlated. Levels of acceptable symptom severity were low and did not differ across the eight symptoms. Four patient subgroups were identified: (1) those who rated all symptoms as low in importance (n=12); (2) those who rated bronchial symptoms and sleep problems as low in importance and all other symptoms as moderately important (n=29); (3) those who rated nausea and emotional distress as low in importance and all other symptoms as moderately important (n=23); and (4) those who rated all symptoms as highly important (n=33). Subgroups were unrelated to clinical characteristics, except for functional status.

Conclusion:

The modified PCOQ showed evidence of construct validity. Patients considered low symptom severity to be acceptable, irrespective of the symptom. Findings suggest that symptom severity and importance are related yet distinct aspects of the advanced lung cancer symptom experience. Patients have heterogeneous priorities for symptom improvement, which has implications for tailoring treatment.

Introduction

Cancer symptom research has largely focused on symptom severity, frequency, and distress [1]. However, little is known about cancer patients’ success criteria for symptom treatment and their priorities for symptom improvement. Success criteria are levels of symptom severity following symptom treatment that the patient considers to be successful, or acceptable. Additionally, priorities for symptom improvement are symptoms that the patient perceives as important to improve. Understanding patients’ perspectives in these domains is critical for patient-centered care, which has been associated with improved patient satisfaction with health care, health outcomes, treatment adherence, and more cost-effective care [2–5].

One aspect of patient-centered care is shared decision-making in which the clinician describes treatment options, including their risks and benefits, and the patient shares preferences and values related to these options [6]. Thus, patients are empowered to actively participate in their care [7,8]. Patient-defined success criteria for symptom improvement inform shared treatment decisions by providing indices of clinically meaningful improvement from the patients’ perspective. Furthermore, given that lung cancer patients have, on average, 14 symptoms [9], assessing their priorities for symptom improvement also informs shared decision-making.

Success criteria and priorities for symptom improvement have been assessed with the Patient Centered Outcomes Questionnaire (PCOQ) in non-cancer populations, such as patients with chronic pain or Parkinson’s disease [10–18]. This research has found differences across medical populations in the level of symptom reduction considered a treatment success. Patients with chronic pain or fibromyalgia required 52-77% reductions in pain, fatigue, and distress to consider symptom treatment successful [11,13,14,16–18]. Other medical populations, such as patients awaiting liver transplantation and those with Parkinson’s disease, had lower criteria for treatment success [12,15]. Additionally, subgroups of patients with chronic pain have been identified based on their symptom improvement priorities on the PCOQ. Common subgroups include patients who rate all symptoms as highly important to improve and those who rate pain as highly important to improve [11,14,17]. Typically, patients in the multi-focused high importance subgroup have reported greater pain, fatigue, and distress than other subgroups [11,14,18].

To date, only one study has used a modified PCOQ to examine cancer patients’ success criteria and priorities for symptom improvement [19]. For this study, the PCOQ was revised to include 10 common symptoms in metastatic breast cancer. Symptoms requiring the greatest reductions for patients to consider symptom treatment successful were fatigue (49% reduction), cognitive problems (48% reduction), and sleep problems (43% reduction) [19]. Additionally, the following patient subgroups based on symptom improvement priorities were found: 1) those who rated cognitive problems, sleep problems, and fatigue as highly important, 2) those who rated pain as highly important, and 3) those who rated all symptoms as highly important. Few differences in demographic, medical, and symptom severity variables were found across subgroups.

It is important to expand research on success criteria and priorities for symptom improvement to other cancer populations with high symptom burden. In particular, advanced lung cancer patients often experience various symptoms that are a major source of distress, impairment, and disability [20–23]. Lung cancer patients are more likely to report moderate to severe symptoms compared to patients with other cancer types [24,25]. Only one study has examined advanced lung cancer patients’ perceptions of symptom importance, which informed the NCCN-FACT Lung Symptom Index-17 [26]. The top-ranked concerns were fatigue, being able to enjoy life, and worrying that their condition will worsen. However, the final measure did not assess patients’ success criteria or priorities for symptom improvement.

To address these gaps in the literature, we modified the PCOQ to focus on eight common symptoms in advanced lung cancer. The measure assesses usual symptom severity, acceptable symptom severity following symptom treatment, and the importance of seeing improvement in each symptom. Thus, the primary aim of this study was to provide a preliminary evaluation of the construct validity of the new PCOQ measure. Potential correlates of the measure were selected based on the Dodd Symptom Management Model [27]. In general, this model posits that patients’ symptom experiences are correlated with demographics, medical factors, and outcomes such as functional status and quality of life. For primary aim 1, we hypothesized that symptom severity ratings on the PCOQ would be correlated with standard assessments of the same symptoms, medical comorbidities, functional status, quality of life, and the importance of seeing improvement in each symptom. Secondary aims of this study were: (2) to compare acceptable severity levels following symptom treatment across the eight symptoms; (3) to identify patient subgroups based on the importance of seeing improvement in each symptom; and (4) to explore potential demographic and clinical correlates of patient subgroups based on symptom importance.

Methods

Participants and Procedures

This study examined a portion of the data from a cross-sectional survey of advanced lung cancer patients’ health and well-being. The Indiana University (IU) Institutional Review Board approved study procedures. Advanced lung cancer patients were recruited from the IU Simon Comprehensive Cancer Center between March and April 2019. Eligible patients met the following inclusion criteria: (1) at least three weeks post-diagnosis of inoperable stage IIIB, IIIC, or IV non-small cell lung cancer or extensive stage small cell lung cancer; (2) ≥18 years old; (3) fluent in English; and (4) no evidence of severe cognitive impairment based on a clinical cut-point (≥3 errors) on a 6-item validated cognitive screener [28]. Eligibility status was determined via medical chart review and consultation with the primary oncologist.

Potentially eligible patients were mailed a study introductory letter along with a study information sheet. The letter provided contact information for opting out of the study. Research assistants called patients who did not opt out to administer the cognitive screener [28] and obtain verbal informed consent. Consenting patients were sent an online or paper survey, depending on their preference. Reminder calls were made as necessary, and for those sent an online survey, automated emails also served as reminders. Once the survey was received, participants were mailed a $25 gift card.

Of the 176 patients who were sent recruitment mailings, 9 were deceased and 27 were lost to follow-up. Of the 140 reached patients, 115 (82%) completed the screening assessment, 21 refused, and 4 were too ill to provide verbal consent. The most common reasons for refusal were lack of interest, illness, or time constraints. Of the 111 patients who were eligible, all of them consented to participate and 103 (93%) completed the survey. One patient was found to be ineligible due to cancer stage following survey completion.

Measures

Modified PCOQ.

The original PCOQ for chronic pain [14] showed adequate test-retest reliability for usual symptom severity over a 48-hour period and good concurrent validity with measures of pain, disability, and distress [10]. Based on a review of the literature and cognitive interviews with advanced lung cancer patients (see Online Resources 1–3), we modified the PCOQ to focus on eight symptoms in three sections. For each symptom, participants rated their usual symptom severity during the past week on a scale from 0 (none) to 10 (worst imaginable). If their usual symptom severity was greater than zero, participants responded to the following questions: “What level of [symptom] would be acceptable to you if you were to receive treatment for [symptom]?” on a scale from 0 (none) to 10 (worst imaginable) and, “How important is it for you to see improvement in your level of [symptom]?” on a scale from 0 (not at all important) to 10 (most important). In cognitive interviews, patients defined success criteria as symptom severity that would be acceptable to them following symptom treatment. Thus, the term “acceptable” was used in the modified PCOQ.

Symptom Severity.

Four-item Patient-Reported Outcomes Measurement Information System (PROMIS) measures assessed anxiety, depressive symptoms, fatigue, and sleep disturbance, and a 3-item PROMIS measure assessed pain intensity [29]. PROMIS measures were not available for nausea, cough, and lack of appetite, and the PROMIS breathlessness measure only assesses this symptom when performing certain activities. Thus, these four symptoms were assessed with items from the Memorial Symptom Assessment Scale (MSAS) [30].

Demographics and Medical Factors.

Most demographics were self-reported. Age, gender, and cancer information were collected from medical records. Additionally, participants reported whether eight medical conditions had been diagnosed or treated within the past three years [31]. Functional status was assessed with the activities and function item from the Patient-Generated Subjective Global Assessment (PG-SGA), with higher scores indicating worse functional status [32]. Overall quality of life was assessed with one item from the McGill Quality of Life Questionnaire on a scale from 0 (very bad) to 10 (excellent) [33]. Finally, using an author-constructed checklist, participants indicated whether they had received treatment in the past three months for each of the eight symptoms included in the modified PCOQ. Treatment was defined as over-the-counter or prescribed medication, oxygen, psychotherapy/counseling, or other treatments.

Data Analysis

Descriptive statistics were calculated, and the normality of study variables was examined in IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA). For primary aim 1, the preliminary construct validity of the modified PCOQ was assessed via correlations between PCOQ symptom severity ratings, standard measures of the same symptoms, and theoretically related constructs from the Dodd Symptom Management Model (e.g., medical comorbidities, functional status, quality of life) [27].

For secondary aim 2, linear mixed modeling was performed in SPSS with the sample endorsing one or more symptoms (i.e., usual severity rating ≥1 on a 0 to 10 scale) to compare acceptable severity levels following symptom treatment across the eight symptoms. The mixed model allowed for an unbalanced design in which patients who only provided acceptable severity levels for certain symptoms were still included in the model. In the mixed effect model, the predictor was symptom type (e.g., pain, fatigue), a within-subjects factor, and the outcome was the acceptable level of symptom severity.

For secondary aim 3, latent profile analysis (LPA), a model-based approach that classifies individuals into classes, or subgroups, based on study variables [34], was conducted with Mplus version 8 [35] to identify patient classes based on importance ratings for each of the eight symptoms. Six models were estimated with classes added iteratively to determine the best fitting model. Information criteria, the bootstrap likelihood ratio test (BLRT), entropy, and interpretability of the model solution were assessed to determine model fit and the final number of classes. Information criteria included the Bayesian Information Criterion (BIC), Akaike Information Criterion (AIC), and sample size adjusted Bayesian Information Criterion (ssBIC). Lower values of the information criteria, a statistically significant value for the BLRT, and entropy above .80 indicate a good-fitting model [36–38]. Feasibility of class interpretation was also an important consideration when determining model fit [34].

For secondary aim 4, multinomial logistic regressions using Vermunt’s 3-step approach [39] were performed to explore potential correlates of classes, including usual severity of the eight symptoms on the PCOQ, demographics (i.e., age, gender, marital status, employment status, education, and income), and clinical factors from the Dodd Symptom Management Model [27] (i.e., time since the advanced lung cancer diagnosis, current and prior cancer treatments, number of medical comorbidities, functional status, quality of life, and symptom treatment history). Each potential correlate was examined in a separate regression, p-values <.01 were considered statistically significant due to the number of analyses.

Results

Patient characteristics are shown in Table 1. Patients were mostly White (82%) and 54% were female, with an average age of 65 years (SD=11.9). Patients were, on average, 2.6 years (SD=2.5) from their advanced lung cancer diagnosis. The majority had received chemotherapy (61%) and immunotherapy (55%).

Table 1.

Participant demographic and medical characteristics (N = 102)

Characteristic
Age
Mean (S.D.)	64.96 (11.87)
Range	34 – 92
Gender, no. (%)
Male	47 (46.08)
Female	55 (53.92)
Race/ethnicity, no. (%)
Non-Hispanic White	84 (82.35)
African American/Black	13 (12.75)
Othera	5 (4.90)
Married or Living with a Partner, no. (%)	68 (66.67)
Employed, no. (%)	27 (26.47)
Level of Education, no. (%)
No college	34 (33.33)
Some college	32 (31.37)
Graduated college/graduate school	36 (35.29)
Household Income,b no. (%)
$0 - $20,999	10 (9.80)
$21,000 - $30,999	11 (10.78)
$31,000 - $50,999	24 (23.53)
$51,000 - $99,999	24 (23.53)
$100,000 or more	24 (23.53)
Lung Cancer Stage, no. (%)
NSCLC IIIB	16 (15.69)
NSCLC IIIC	1 (0.98)
NSCLC IV	78 (76.47)
SCLC Extensive	7 (6.86)
Years since Advanced or Extensive Stage Diagnosis
Mean (S.D.)	2.60 (2.54)
Range	0.06 – 11.30
Cancer Treatment History,c no. (%)
Surgery	31 (30.39)
Chemotherapy	62 (60.78)
Radiation	36 (35.29)
Chemoradiation	26 (25.49)
Targeted Therapy	39 (38.24)
Immunotherapy	56 (54.90)
Current Cancer Treatmentd no. (%)
Chemotherapy	17 (16.67)
Radiation	1 (0.98)
Chemoradiation	3 (2.94)
Targeted Therapy	27 (26.47)
Immunotherapy	32 (31.37)
Symptom Treatment History,e no. (%)
Breathlessness	28 (27.45)
Cough	21 (20.59)
Fatigue	22 (21.57)
Sleep Problems	24 (23.53)
Pain	36 (35.29)
Nausea	23 (22.55)
Lack of Appetite	9 (8.82)
Emotional Distress	17 (16.67)
Medical Comorbidities
Mean (S.D.)	1.44 (1.04)
Range	0 – 4

NSCLC non-small cell lung cancer, SCLC small cell lung cancer.

^aMulti-racial, Asian American, Native American, Hispanic, and other.

^bMedian household income in Indiana is $57,603 according to U.S. Census 2019 data. Reference: United States Census Bureau (2020) Indiana 2019. https://data.census.gov/cedsci/profile?g=0400000US18. Accessed 12 February 2021

^cTreatment any time before study completion.

^dTreatment ≤4 weeks before study completion.

^eTreatment in the past 3 months for a particular symptom.

Descriptive statistics for the PCOQ are found in Table 2. Patients reported, on average, 4.82 symptoms (SD=2.20). Regarding the construct validity of the PCOQ (aim 1), the usual severity of all symptoms on the PCOQ was highly correlated with standard assessments of the same symptoms, rs(97–102)=0.68–0.91, ps<0.01 (see Table 3). The severity of breathlessness, fatigue, sleep problems, and lack of appetite was positively correlated with medical comorbidities, rs(101–102)=0.29–0.31, ps<0.01, whereas other symptoms showed small, non-significant correlations with medical comorbidities. In addition, the severity of all symptoms was positively correlated with functional status, rs(98–101)=0.36–0.62, ps<0.01, and the severity of all symptoms except for cough and nausea was negatively correlated with quality of life, rs(91–100)=−0.48 – −0.31, ps<0.01. Moderate, positive correlations were found between the severity and importance of cough, fatigue, sleep problems, and pain, rs(57–90)=0.46–.58, ps<0.01. Whereas severity and importance were also moderately correlated for breathlessness, nausea, and emotional distress, results fell short of statistical significance, rs(29–71)=0.30–.39, ps<0.05.

Table 2.

Descriptive statistics for PCOQ constructs

Symptom	Usual Severity				Acceptable Severity				Importance of Improvement
	n	Mean	S.D.	Range	n	Mean	S.D.	Range	n	Mean	S.D.	Range
Breathlessness	102	2.26	2.20	0 – 10	70	2.31	2.18	0 – 8	71	6.42	3.13	0 – 10
Cough	100	1.87	2.25	0 – 9	58	2.09	1.92	0 – 8	57	5.47	3.46	0 – 10
Fatigue	101	4.39	2.72	0 – 10	88	2.60	1.88	0 – 9	90	6.54	2.94	0 – 10
Sleep problems	102	3.09	2.78	0 – 9	73	2.70	2.31	0 – 10	72	5.76	3.09	0 – 10
Pain	99	3.04	2.80	0 – 9	66	2.71	2.13	0 – 10	66	6.85	2.93	0 – 10
Nausea	100	1.00	1.96	0 – 8	29	2.14	2.28	0 – 10	29	5.55	3.38	0 – 10
Lack of appetite	102	1.79	2.74	0 – 10	40	3.10	2.18	0 – 9	39	5.18	3.03	0 – 10
Emotional distress	102	2.22	2.64	0 – 10	58	2.41	1.70	0 – 8	58	6.09	3.24	0 – 10

PCOQ Patient Centered Outcomes Questionnaire.

Table 3.

Correlations between PCOQ symptom severity and hypothesized variables

Standardized Severitya				Medical Comorbiditiesb		Functional Statusc		Quality of Lifed		Symptom Importancee
		r	p-value	r	p-value	r	p-value	r	p-value	r	p-value
1.	Breathlessness	0.72	<.001*	0.31	.002*	0.54	<.001*	−0.34	<.001*	0.30	.012
2.	Cough	0.87	<.001*	0.16	.108	0.36	<.001*	−0.25	.015	0.58	<.001*
3.	Fatigue	0.81	<.001*	0.29	.004*	0.62	<.001*	−0.44	<.001*	0.46	<.001*
4.	Sleep problems	0.76	<.001*	0.29	.003*	0.40	<.001*	−0.31	.002*	0.57	<.001*
5.	Pain	0.84	<.001*	0.15	.134	0.45	<.001*	−0.39	<.001*	0.49	<.001*
6.	Nausea	0.91	<.001*	0.07	.467	0.50	<.001*	−0.26	.010	0.39	.038
7.	Lack of appetite	0.79	<.001*	0.29	.003*	0.56	<.001*	−0.38	<.001*	0.26	.116
8.	Emotional distress	0.68, 0.69f	<.001*	0.20	.044	0.41	<.001*	−0.48	<.001*	0.30	.021

Pairwise correlations. PCOQ Patient Centered Outcomes Questionnaire.

^*p<.01

^aMemorial Symptom Assessment Scale measures were used for breathlessness, cough, nausea, and lack of appetite, and Patient-Reported Outcomes Measurement Information System (PROMIS) measures were used for fatigue, sleep problems, pain, and emotional distress (i.e., anxiety and depression), ns = 97-102.

^bns = 99-102.

^cns = 98-101.

^dns = 97-100.

^ens = 29-90.

^fSeparate correlations were computed between PCOQ emotional distress and PROMIS anxiety and depression measures, rs=0.68, 0.69, respectively.

On average, of those experiencing the relevant symptom, patients required a small reduction from usual severity (−.93 to −2.30 within a 0-10 scale) to consider these symptoms acceptable. Estimated marginal means indicated that the lowest acceptable severity was 1.73 for nausea, whereas the highest acceptable severity was 2.71 for lack of appetite. For secondary aim 2, linear mixed modeling results yielded no significant differences in acceptable severity across the eight symptoms (see Table 4).

Table 4.

Acceptable levels of symptom severity using linear mixed modeling (n = 97)

	Estimated Marginal Mean	S.E.
Symptoms
Breathlessness	2.28	0.24
Cough	2.01	0.25
Fatigue	2.62	0.22
Sleep problems	2.48	0.23
Pain	2.52	0.24
Nausea	1.73	0.33
Lack of appetite	2.71	0.29
Emotional distress	2.37	0.25

Each symptom was rated on a 0 to 10 scale.

For secondary aim 3, six latent profile models were estimated to identify patient classes based on symptom improvement priorities (see Online Resource 4). Considering all of the criteria for model selection, we chose the 4-class model because it yielded the lowest BIC value and provided a better fit than the 3-class model based on the BLRT. More importantly, the 4-class model was the most conceptually meaningful. The characteristics of the four classes are shown in Figure 1 and Table 5. Class 1 rated all symptoms as low in importance (n=12, 12%). Class 2 rated bronchial symptoms (i.e., breathlessness, cough) and sleep problems as low in importance and all other symptoms as moderately important (n=29, 30%). Class 3 rated nausea and emotional distress as low in importance and all other symptoms as moderately important (n=23, 24%). Class 4 rated all symptoms as highly important (n=33, 34%).

Fig. 1.

Patient subgroups’ mean importance ratings on the Patient Centered Outcomes Questionnaire (PCOQ). N = 97

Table 5.

Descriptive statistics for classes based on symptom importance

	Class 1		Class 2		Class 3		Class 4
	Estimated Marginal Means (S.E.)	Range	Estimated Marginal Means (S.E.)	Range	Estimated Marginal Means (S.E.)	Range	Estimated Marginal Means (S.E.)	Range
Symptom Importance
Breathlessness	2.70 (0.72)	0-5	3.94 (0.50)	0-7	6.74 (0.51)	2-10	9.27 (0.39)	5-10
Cough	1.39 (0.57)	0-4	2.30 (0.35)	0-5	6.92 (0.48)	3-10	9.23 (0.37)	8-10
Fatigue	1.73 (0.69)	0-5	6.73 (0.45)	1-10	6.60 (0.50)	4-9	8.31 (0.41)	1-10
Sleep problems	2.49 (0.77)	0-5	3.78 (0.50)	0-9	6.14 (0.56)	3-10	8.39 (0.49)	3-10
Pain	1.92 (0.74)	0-6	6.34 (0.46)	3-10	5.54 (0.49)	1-9	9.52 (0.37)	8-10
Nausea	1.51 (0.89)	0-4	4.48 (0.58)	1-8	3.76 (0.67)	0-5	9.25 (0.59)	5-10
Lack of appetite	0.97 (1.35)	0-1	4.44 (0.66)	0-7	4.54 (0.67)	1-9	7.91 (0.69)	0-10
Emotional distress	1.29 (0.69)	0-2	5.90 (0.43)	3-9	3.16 (0.41)	0-6	9.47 (0.31)	8-10
	Mean (S.D.)	Range	Mean (S.D.)	Range	Mean (S.D.)	Range	Mean (S.D.)	Range
Differing Variable
Functional status	0.58 (0.67)	0-2	1.17 (0.76)	0-3	1.48 (0.90)	0-3	1.25 (0.92)	0-3

Class 1 Low Importance, n = 12; Class 2 Moderate Importance except for low bronchial symptoms and sleep importance, n = 29; Class 3 Moderate Importance except for low nausea and emotional distress importance, n = 23; Class 4 High Importance, n = 33.

For secondary aim 4, multinomial logistic regressions using Vermunt’s 3-step approach were conducted to explore differences between classes on demographic and clinical factors. In sequential analyses, classes 1, 2, and 4 were used as reference groups in order to compare all classes. Only one significant difference (p<.01) emerged between classes; worse functional status was associated with a greater likelihood of being in class 3 than class 1 (OR=5.25, 99% CI [1.03, 26.65]).

Discussion

The modified PCOQ showed evidence of construct validity in advanced lung cancer patients. Correlations between this measure and theoretically related constructs were largely consistent with the Dodd Symptom Management Model [27]. Additionally, on average, patients considered low symptom levels to be acceptable following symptom treatment, a preference that did not differ across the eight symptoms. Lastly, patients had heterogeneous priorities for symptom treatment, which were largely unrelated to demographic and clinical factors.

Regarding the preliminary construct validity of the PCOQ, results were largely consistent with our hypotheses. The severity of all symptoms on the PCOQ was related to standard measures of the same symptoms and functional status. In addition, for many symptoms on the PCOQ, severity was correlated with medical comorbidities and quality of life. Moderate associations were found between the severity of most symptoms and their importance. Findings suggest that symptom severity and importance are related but distinct aspects of the symptom experience in advanced lung cancer. Other studies in chronic pain and cancer also have found significant associations between usual symptom severity and patient subgroups based on symptom importance, although the size of these associations was not reported [11,12,14,18,19]. Only two studies, both of which had small samples, did not replicate these associations [15,17].

During cognitive interviews regarding our modified PCOQ, advanced lung cancer patients reported several reasons for their symptom importance ratings other than symptom severity. Some patients attributed their lower symptom importance ratings to their ability to tolerate higher symptom severity, rather than actual symptom severity. One patient based her importance ratings on symptom interference with daily activities. Another patient stated that her low importance ratings reflected a focus on survival rather than symptoms. These findings provide possible explanations for the moderate correlations between symptom severity and importance that warrant examination in future research.

Differences in acceptable severity levels were not found across the eight symptoms; low severity was considered acceptable for all symptoms. For those experiencing the relevant symptom, patients required small reductions to consider symptom severity acceptable. Fatigue severity was the furthest from its acceptable level, consistent with research suggesting that fatigue is the most bothersome symptom for advanced lung cancer patients [40]. The current sample’s acceptable severity levels were comparable to success criteria for symptom improvement among patients with chronic pain and metastatic breast cancer [13,14,17,19]. Patients with chronic pain often adjust their success criteria such that higher symptom levels are more acceptable after experiencing partial pain relief with treatment [10]. Thus, our current sample’s low acceptable symptom levels may be related to their generally mild to moderate symptom levels and low rates of recent symptom treatment.

Four distinct patient classes, or subgroups, were identified based on priorities for symptom improvement. Previous work in chronic pain has found similar heterogeneity [11,12,14,17]. Only one other cancer study identified patient subgroups based on symptom importance [19]. Similar to the current study, one subgroup rated all symptoms as highly important. Other subgroups differed from those of the current study, which may reflect differences in sample characteristics (e.g., metastatic breast vs. lung cancer) and methodology.

When exploring demographics and clinical factors as possible correlates of patient subgroups based on symptom importance, only functional status was associated with subgroups. Associations between functional status and subgroups based on symptom importance have rarely been examined, with one cancer study finding null associations [19]. Our results are not consistent with the Dodd Symptom Management Model, positing that demographic and health-related characteristics affect the symptom experience [27]. However, results replicate findings in the general PCOQ literature that has largely focused on chronic pain; few demographic and clinical correlates of subgroups based on symptom importance have been found [11,12,14,15,17–19]. Conversely, among patients with cancer and chronic pain, age, education, and usual symptom severity have been correlated with these subgroups [11,12,14,18,19]. In our study, symptom treatment history was unrelated to patient subgroups based on symptom importance. The low rates of recent symptom treatment may have contributed to null findings. In addition, because experiences with symptom treatment are not uniformly successful, the relationship between symptom treatment history and symptom importance is likely to be complex.

Limitations of this study should be noted. We enrolled patients with advanced lung cancer at one academic medical center in the midwestern United States, most of whom were Caucasian. Although the sample’s demographics were typical for this center, patients who agreed to participate in our survey may have differed in important ways from those who did not participate. Additionally, the relatively small sample size and range restriction for certain variables contributed to null findings. Lastly, the cross-sectional design did not allow for examination of test-retest reliability in our study and change in acceptable symptom severity and symptom treatment priorities over time. These factors are likely to vary throughout the cancer experience.

Our findings have significant implications for the tailoring of symptom treatment for advanced lung cancer patients. In general, results point to the importance of considering patient preferences, such as their acceptable symptom severity and symptom improvement priorities, when discussing treatment options to foster shared decision-making. Specifically, results suggest that patients typically require low symptom severity levels to consider treatment outcomes acceptable, highlighting the need for patient-provider discussions about possible outcomes. Findings also suggest that advanced lung cancer patients have heterogeneous priorities for symptom improvement. Thus, assessing patients’ priorities and goals for symptom treatment can inform shared decision-making and patient-centered care. For example, after assessing the severity of symptoms, a provider could ask, “What is the most important symptom to treat?”

Furthermore, our findings indicate that fatigue may be an important symptom for intervention in advanced lung cancer patients, as it required the greatest reduction in severity to be acceptable and was also rated as highly important. Previous research with advanced lung cancer patients has found that fatigue often co-occurs with other symptoms [41,42]; therefore, treating other symptoms may help reduce fatigue. For example, evidence suggests that cognitive-behavioral interventions may improve the severity of the pain-fatigue-sleep symptom cluster in patients with various cancers [43]. However, treatment for one symptom, such as steroids or stimulants for fatigue, may negatively impact other symptoms, such as anxiety and sleep problems [44,45], Thus, providers should discuss symptom treatment side effects with advanced lung cancer patients to ensure that treatment is consistent with their priorities.

Further research is needed to examine the psychometric properties of our measure in larger, more diverse samples that fully represent the advanced lung cancer population. Additionally, longitudinal research may examine changes in acceptable symptom levels and symptom improvement priorities over the entire lung cancer trajectory. Researchers could also modify and validate the PCOQ in other advanced cancer populations. Additionally, stepped-care intervention trials targeting symptom clusters could consider patients’ priorities for symptom improvement when determining the sequence of care. This research could ultimately inform patient-centered approaches to optimizing quality of life and functioning in advanced cancer.