Exploring Symptom Clusters and Their Measurements in Patients With Lung Cancer: A Scoping Review for Practice and Research

Despite the many advances in cancer treatment options and the increasing number of cancer survivors, patients with lung cancer continue to experience significant symptom burden and distress (Siegel et al., 2020; Sung et al., 2017). Patients with lung cancer often experience multiple side effects from treatment and symptoms from the disease itself (Bouazza et al., 2017), affecting their functional status, general health perception, and overall health-related quality of life (Ferrans et al., 2005). Systematic use of patient-reported outcome measures (PROMs) during treatment and follow-up is fundamental in the era of person-centered care, and symptom assessment is the first step of the symptom management process. Using PROMs is one method to improve symptom management, enhance quality of care, and promote patient satisfaction for patients with cancer (Graupner et al., 2021). Within oncology nursing, continuous symptom management, including assessment, is critical in providing holistic person-centered care of high quality.

Symptom Clusters in the Oncologic Setting

In clinical practice and research, symptoms occurring in clusters have a synergistic and cumulative effect on patient outcomes compared with single symptoms, hence having important implications for clinical practice (Miaskowski et al., 2017). Research has focused on single symptoms, and most patients with cancer experience multiple co-occurring, related symptoms (Dodd, Miaskowski, & Paul, 2001; Kim et al., 2005). The present study relies on the symptom cluster definition by Kim et al. (2005) as established groups of symptoms (two or more) related to each other and relatively independent of other clusters revealing specific underlying concepts of symptoms and that may share the same etiology. Symptom clusters have been identified in qualitative and quantitative research by exploring nonpredefined clusters de novo (Chen & Tseng, 2006; Walsh & Rybicki, 2006) or predefined clusters a priori (Gift et al., 2003; Hammer et al., 2022; Miaskowski et al., 2006). The de novo approach means that relationships between many symptoms are explored and clusters consisting of a varied number of symptoms are identified. The a priori approach denotes that symptom clusters, generally consisting of two to three symptoms, are investigated based on a beforehand assumed relationship (Xiao, 2010). The number and specific symptoms within each symptom cluster and symptom dimensions measured have been found to be highly variable (Ward Sullivan et al., 2018), and the constellations depend on patient characteristics, disease stage, treatment, time frame of the measuring instrument, and statistical method used (Fan et al., 2007; Kim et al., 2009; Kirkova et al., 2010).

The middle-range theory of unpleasant symptoms (TOUS) (Lenz et al., 1997) has been applied in research within the oncology setting regarding multiple symptoms (Chan et al., 2005; Fox & Lyon, 2007; Gift et al., 2004; Hoffman et al., 2007; Kim et al., 2015; Wu et al., 2015; Xiao et al., 2021). TOUS has the following three main reciprocal components: the multidimensional symptom experience, the influencing factors, and performance (outcome). The conceptualized symptom experience includes four symptom dimensions as follows: intensity (strength or severity), timing (duration and/or frequency of occurrence), distress (level of discomfort), and quality (specific perceptions experienced with the symptom). Factors influencing the symptom experience are important in symptom assessment and management, including physiologic, psychological, and situational antecedents, which can interact. Performance is the result of the symptom experience, including functional and cognitive activities (Lenz et al., 1997), and quality of life may be recognized as an outcome of the symptom experience (Lenz & Pugh, 2014).

The instruments used in symptom research within the oncologic context evaluate different symptom dimensions, sometimes combined with outcomes regarding function. They are cancer-specific, disease-specific, or symptom-specific instruments designed for self- or interviewer administration, and minor cultural adaptations may apply. Symptom instruments may include single or multiple symptoms, varying between either single or multiple items per symptom. A wide variety of 18 symptom and functional assessment tools were used in symptom cluster research (Xiao, 2010). When searching for comparable health-related quality-of-life parameters among patients with lung cancer, 17 different instruments were identified (Damm et al., 2013). A more detailed description of symptom assessment instruments and the symptom dimensions creating the symptom cluster was reported by Ward Sullivan et al. (2018), where the MD Anderson Symptom Inventory (MDASI) (Cleeland et al., 2000) and Memorial Symptom Assessment Scale (Portenoy et al., 1994) were the most commonly used. Although instruments may measure several dimensions of the symptom experience, composite indexes are presented in the results, and symptom dimensions have been sparingly used as outcomes (Henoch et al., 2018).

Many issues remain in symptom cluster research (Miaskowski et al., 2017; Ward Sullivan et al., 2018), such as whether the number and varieties of symptom clusters differ based on the dimensions used to create the cluster and how cluster symptoms are related (Miaskowski, 2016; Miaskowski et al., 2017).

It is essential to have a relevant scope of the literature within the field of interest to ensure that data can be presented for comparison to develop symptom management interventions. Hence, the present scoping review maps findings from symptom cluster research within the specific context of lung cancer and the oncologic setting to help clinical healthcare personnel and researchers. The aim is to explore symptom clusters in patients with lung cancer, describing how included symptoms and symptom dimensions are measured.

Methods

Study Design

A scoping review methodology was chosen because the intention was to map the knowledge regarding the contents and measurements of symptom clusters. The PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) was used (Tricco et al., 2018). The scoping review followed the five stages proposed by Arksey and O’Malley (2005) and the updates regarding alignment with the PRISMA-ScR (Peters et al., 2020) and the clarification and enhancement of each stage (Levac et al., 2010): (a) identifying the research question, (b) finding relevant studies, (c) selecting studies, (d) charting the data, and (e) collating, summarizing, and reporting the results. The optional sixth stage—consulting with the reference group—was not used.

Research Questions

The research questions were as follows:

• Which symptom clusters exist in lung cancer research?
• How do the de novo and a priori clusters differ in symptom constellation?
• Which instruments are used in symptom cluster research (i.e., how are the symptoms measured)?
• How are the symptom dimensions (intensity, timing, distress, and quality) made evident in quantitative and qualitative research?

Search Strategy

The key search terms were arranged (Peters et al., 2020) by defining the population (patients with lung cancer), concept (symptom clusters and symptom dimensions), and context (oncologic setting) and comprised the following: lung cancer OR lung neoplasm AND patient experiences OR patient descriptions OR patient reported outcomes AND symptoms OR symptom cluster OR multiple symptoms OR symptom distress OR symptom burden OR symptom dimensions OR symptom intensity OR symptom quality OR symptom severity OR symptom frequency OR meaning. Searches were performed in PubMed^®, CINAHL^®, Scopus^®, and the Cochrane Library. Specific headings, MeSH (Medical Subject Headings) terms, and keywords were used in combination with free-text search terms by using Boolean operators adapted to the specific databases to obtain breadth and depth. Searches were conducted with no limitation as to the earliest year of publication, and the searches ended at December 31, 2021.

Study Selection and Eligibility Criteria

Eligibility criteria guiding study selection were discussed until consensus was reached. Studies published in English in peer-reviewed journals concerning self-reported experiences of multiple co-occurring symptoms in a population of adult patients with lung cancer at or after the time of diagnosis were considered for selection. The selection process is presented in a PRISMA-ScR flow diagram (Page et al., 2021) (see Figure 1). The first author performed the database searches, with guidance from an experienced librarian. Selected articles (n = 2,371) were exported to EndNote X8/X9 for sorting and removing duplicates (n = 848), and the Rayyan QCRI software was used for further review processing. Through Rayyan, 1,523 articles were screened for inclusion. The eligibility criteria were used to screen the title/abstract, resulting in the exclusion of 973 articles. Following full-text screening (n = 550), 53 articles from 48 studies met the eligibility criteria. The screening was blinded and conducted by a minimum of two authors for each article. Any discrepancies were discussed within the author team until consensus was reached.

All included articles were screened for ethical approval, and a quality appraisal was performed (Daudt et al., 2013) using the Mixed Methods Appraisal Tool (Hong et al., 2018). The Mixed Methods Appraisal Tool includes five domains that vary depending on study design. Each assessment was scored from 0 to 100, with a higher score indicating higher quality. The quality appraisal was performed by the first author and then discussed and verified by another author.

Charting the Data

An extraction template consistent with the aim and research questions was developed to organize the data extraction. The symptom dimension–related data were identified using the TOUS (Lenz et al., 1997) for guidance regarding the specific inference of each dimension.

Collating, Summarizing, and Reporting the Results

The initial analysis divided the articles by their methodology, some investigating a priori symptom clusters and others de novo symptom clusters. The de novo section was arranged into two parts, when applicable, according to whether the included article had a distinct study aim to explore symptom clusters. All identified symptom clusters were charted per article. The most prevalent symptoms among the symptom clusters were identified. A comparison was also made regarding which other symptoms these prevalent symptoms were inclined to cluster with depending on methodology.

Because of the vast variety of obtainable symptom items from the variety of assessment scales used in the included articles, related individual symptom items were combined and further sorted into seven symptom categories to make comparisons possible. For example, the breathing-related symptoms dyspnea, difficulty breathing, shortness of breath, and breathlessness were combined into “dyspnea symptoms” as a part of the “respiratory symptoms.” “Pain symptoms” include the symptom items pain, neuropathy, numbness, and tingling.

Results

Sample Characteristics

The final sample included 53 articles (see Tables 1 and 2). Overall, the sample represented 11,948 participants with lung cancer who, in terms of their cancer care trajectory, were recently diagnosed, were currently undergoing treatment, or had finished treatment, and were as many as five years after diagnosis, with a mean age ranging from 54 to 72.8 years. Six articles had a qualitative design and 47 a quantitative design; one was a mixed-methods study from 15 different countries. The years of publication ranged from 1993 to 2021. Because one study may yield articles with both a priori and de novo methodology, the included records are referred to as articles instead of studies.

The quality assessment showed an overall moderate to high quality score in 46 studies, with a score of 80–100, and 7 scoring 40–60 (range = 0–100). The main area producing a lower score was the assessment regarding the risk of nonbias because the information was lacking or indicated high nonresponse bias.

Research Questions 1 and 2

The total number of symptom clusters in the final sample of 53 articles was 289. Reoccurring symptom clusters were fatigue and depression, fatigue and dyspnea, fatigue and cough, and dyspnea and cough.

A Priori and De Novo Symptom Clusters

A total of 19 a priori defined symptom clusters were investigated in 13 articles (Chan et al., 2011, 2013; Cheville et al., 2011a, 2011b; Fox & Lyon, 2006; Gift et al., 2003; Henoch & LÖvgren, 2014; Hoffman et al., 2007; Lee, 2020; Molassiotis et al., 2021; Reyes-Gibby et al., 2013; Steffen et al., 2020; Wang et al., 2014). Three clusters—fatigue, anxiety, and breathlessness; fatigue and pain; and fatigue, dyspnea, and cough—were described in more than one article, making the total number of unique clusters 15.

The prevalence of some symptoms was more evident than others. Fatigue was the prevailing symptom, being present in all articles and in 11 different clusters. Dyspnea symptoms were the second most common, being present in eight articles and four different clusters, followed by pain in seven articles and five unique clusters, and depression in four studies and five clusters. Cough was present in three studies but only two clusters.

These most prevalent symptoms in a priori clusters were found to cluster with specific symptom categories. Fatigue is clustered commonly with psychological and nutritional impact symptoms. Dyspnea, pain, and depression had a similar pattern of clustering most often with fatigue and psychological symptoms. Cough was found only in clusters with other respiratory symptoms and/or fatigue. Few clusters included the nutritional impact symptoms, elimination/abdominal symptoms, or the body/hair/skin–related symptoms.

A total number of 270 de novo symptom clusters were identified in 40 articles, where 15 (Brown et al., 2011; Choi & Ryu, 2018; DeClue et al., 2020; Gift et al., 2004; Hamada et al., 2016; Henoch et al., 2009; Khamboon et al., 2015; Li, Li, et al., 2021; Li, Wu, et al., 2021; Liu, Liu, et al., 2021; Maguire et al., 2014; Sarna & Brecht, 1997; Wang & Fu, 2014; Wang et al., 2006, 2008) described an intention in their methodology to investigate symptom clusters de novo (hereby defined as “with cluster aim”). The other 25 articles (Andersen et al., 2020; Belqaid et al., 2016, 2018; Chien et al., 2021; Henoch et al., 2008; Kiteley & Fitch, 2006; Kuo & Ma, 2002; Lavdaniti et al., 2021; Lee, 2021; Liu, Hou, et al., 2021; Lowe & Molassiotis, 2011; Maguire et al., 2019; Maric et al., 2010; McFarland et al., 2020; Mody et al., 2021; Molassiotis, Lowe, Blackhall, & Lorigan, 2011; Molassiotis, Lowe, Ellis, et al., 2011; Morrison et al., 2017; Nishiura et al., 2015; Okuyama et al., 2001; Reyes-Gibby et al., 2007; Sarna, 1993; Tanaka et al., 2002; Tchekmedyian et al., 2003; Turcott et al., 2020) did not do this as a part of their aim (hereby defined as “without cluster aim”), but the results have presented patient-reported co-occurring symptoms—that is, symptom clusters. The number of symptom clusters varied from 1 to 49 variations per article.

Among the 15 articles with a cluster aim, the most prevalent symptoms were fatigue in all 15 articles (Brown et al., 2011; Choi & Ryu, 2018; DeClue et al., 2020; Gift et al, 2004; Hamada et al., 2016; Henoch et al., 2009; Khamboon et al., 2015; Li, Li, et al., 2021; Li, Wu, et al., 2021; Liu, Liu, et al., 2021; Maguire et al., 2014; Sarna & Brecht, 1997; Wang & Fu, 2014; Wang et al., 2006, 2008), followed by pain in 13 (Brown et al., 2011; Choi & Ryu, 2018; DeClue et al., 2020; Hamada et al., 2016; Henoch et al., 2009; Khamboon et al., 2015; Li, Li, et al., 2021; Li, Wu, et al., 2021; Maguire et al., 2014; Sarna & Brecht, 1997; Wang & Fu, 2014; Wang et al., 2006, 2008), depression (n = 12) (Choi & Ryu, 2018; DeClue et al., 2020; Hamada et al., 2016; Henoch et al., 2009; Khamboon et al., 2015; Li, Li, et al., 2021; Li, Wu, et al., 2021; Liu, Liu, et al., 2021; Sarna & Brecht, 1997; Wang & Fu, 2014; Wang et al., 2006, 2008), appetite loss (n = 12) (Choi & Ryu, 2018; Gift et al, 2004; Hamada et al., 2016; Henoch et al., 2009; Khamboon et al., 2015; Li, Li, et al., 2021; Li, Wu, et al., 2021; Liu, Liu, et al., 2021; Maguire et al., 2014; Sarna & Brecht, 1997; Wang et al., 2006, 2008), cough (n = 10) (Brown et al., 2011; Choi & Ryu, 2018; Henoch et al., 2009; Khamboon et al., 2015; Li, Li, et al., 2021; Li, Wu, et al., 2021; Liu, Liu, et al., 2021; Maguire et al., 2014; Sarna & Brecht, 1997; Wang et al., 2006), and dyspnea (n = 10) (Brown et al., 2011; Choi & Ryu, 2018; Henoch et al., 2009; Khamboon et al., 2015; Li, Li, et al., 2021; Li, Wu, et al., 2021; Maguire et al., 2014; Sarna & Brecht, 1997; Wang et al., 2006, 2008). These symptoms clustered with the following symptom categories: fatigue and pain with nutritional and psychological symptoms. Depression was usually present in constellations with other psychological symptoms and fatigue, and appetite loss with other nutritional impact symptoms and fatigue. Cough was present mostly along with dyspnea and psychological symptoms, and dyspnea clustered mostly with other respiratory symptoms, pain, and psychological symptoms.

Among the 25 articles without a cluster aim, fatigue was the most prevalent symptom, being present in 19 articles (Andersen et al., 2020; Belqaid et al., 2016; Chien et al., 2021; Henoch et al., 2008; Kiteley & Fitch, 2006; Kuo & Ma, 2002; Lee, 2021; Liu, Hou, et al., 2021; Lowe & Molassiotis, 2011; Maric et al., 2010; McFarland et al., 2020; Molassiotis, Lowe, Blackhall, & Lorigan, 2011; Molassiotis, Lowe, Ellis, et al., 2011; Morrison et al., 2017; Nishiura et al., 2015; Okuyama et al., 2001; Reyes-Gibby et al., 2007; Sarna, 1993; Tchekmedyian et al., 2003), followed by dyspnea symptoms (n = 18) (Andersen et al., 2020; Belqaid et al., 2016; Henoch et al., 2008; Kiteley & Fitch, 2006; Kuo & Ma, 2002; Lavdaniti et al., 2021; Lee, 2021; Li, Li, et al., 2021; Liu, Hou, et al., 2021; Lowe & Molassiotis, 2011; Maguire et al., 2019; McFarland et al., 2020; Molassiotis, Lowe, Blackhall, & Lorigan, 2011; Molassiotis, Lowe, Ellis, et al., 2011; Morrison et al., 2017; Okuyama et al., 2001; Sarna, 1993; Tanaka et al., 2002), pain (n = 15) (Andersen et al., 2020; Belqaid et al., 2016; Chien et al., 2021; Kuo & Ma, 2002; Lavdaniti et al., 2021; Lee, 2021; Lowe & Molassiotis, 2011; Maguire et al., 2014; McFarland et al., 2020; Molassiotis, Lowe, Blackhall, & Lorigan, 2011; Morrison et al., 2017; Nishiura et al., 2015; Reyes-Gibby et al., 2007; Sarna, 1993; Turcott et al., 2020), depression (n = 14) (Andersen et al., 2020; Belqaid et al., 2016; Chien et al., 2021; Henoch et al., 2008; Kuo & Ma, 2002; Lavdaniti et al., 2021; Liu, Hou, et al., 2021; Maguire et al., 2019; McFarland et al., 2020; Nishiura et al., 2015; Okuyama et al., 2001; Reyes-Gibby et al., 2007; Tanaka et al., 2002; Tchekmedyian et al., 2003), appetite loss (n = 11) (Belqaid et al., 2018; Kiteley & Fitch, 2006; Kuo & Ma, 2002; Lavdaniti et al., 2021; Lee, 2021; Lowe & Molassiotis, 2011; Maguire et al., 2019; Maric et al., 2010; Mody et al., 2021; Okuyama et al., 2001; Sarna, 1993), and cough (n = 10) (Belqaid et al., 2016; Kiteley & Fitch, 2006; Kuo & Ma, 2002; Maguire et al., 2019; Molassiotis, Lowe, Blackhall, & Lorigan, 2011; Molassiotis, Lowe, Ellis, et al., 2011; Morrison et al., 2017; Okuyama et al., 2001; Sarna, 1993; Tanaka et al., 2002).

These most prevalent symptoms clustered with various symptom categories. Fatigue was commonly clustered with psychological and respiratory symptoms. Dyspnea and pain clustered mostly with psychological symptoms and fatigue. Depression clustered commonly with fatigue and other psychological symptoms. Appetite loss mostly co-occurred with other nutritional impact symptoms and fatigue. Cough was likely to cluster with dyspnea and fatigue.

As with the a priori cluster findings, the analysis of the de novo constellations revealed that few constellations included the elimination/abdominal or the body/hair/skin–related symptoms.

Comparisons are made between the prevalent symptoms and symptom cluster constellations (see Figure 2).

Research Questions 3 and 4

Symptom Measurement Instruments

Overall, 36 different validated instruments containing symptom measurements were identified within the 47 included articles with quantitative and mixed-methods designs. In addition, there were some variations in the numeric rating scale items because this scale was used for several symptoms, such as dyspnea, pain, nausea, and fatigue. In addition, there were some nonconventional author-developed symptom scales in seven articles.

The instruments were cancer, disease, or symptom specific. The instruments mainly used were cancer specific, such as the European Organisation for Research and Treatment of Cancer Quality-of-Life Questionnaire–Core 30 (EORTC QLQ-C30) or the MDASI questionnaire, and disease specific, like the lung cancer modules of these instruments along with the Lung Cancer Symptom Scale. Some instruments were symptom specific, such as the Cancer Dyspnea Scale, Hospital Anxiety and Depression Scale, or Center for Epidemiological Studies–Depression.

The most frequently used instrument was the EORTC QLQ-C30 (n = 12), a general questionnaire developed for populations of patients with cancer (Aaronson et al., 1993). This 30-item instrument contains four domains, where the symptom items constitute half the questionnaire. The supplementary lung cancer–specific module LC13 with 13 additional symptom items (Bergman et al., 1994) was also used frequently (n = 6), as well as the MDASI (n = 8), which contains 13 core cancer symptom items (Lin et al., 2007).

Among the symptom-specific questionnaires, the most frequently used was the Hospital Anxiety and Depression Scale (n = 7), either in covering all 14 items or the depression subscale only (Zigmond & Snaith, 1983), the 12-item Cancer Dyspnoea Scale (n = 3) (Tanaka et al., 2000), and the 10- to 20-item (depending on version) Center for Epidemiological Studies–Depression (n = 3) (Radloff, 1977).

Symptom Dimensions Measured in Quantitative Studies and Described in Qualitative Studies

The author-developed symptom scales measured either intensity or timing. The validated instruments measured mainly intensity and/or timing, with only a few containing the distress and quality dimensions. The qualitative studies provided several descriptions within the distress, timing, and quality dimensions but less of intensity.

Intensity: Intensity was the most frequently measured dimension, existing in 15 instruments as the solitary measured dimension. The MDASI, EORTC scales, and SF-36^® are examples. Intensity and timing are present in nine instruments, such as the Memorial Symptom Assessment Scale and Symptom Distress Scale; intensity and distress in four instruments, the Brief Fatigue Inventory, Brief Pain Inventory, Functional Assessment of Cancer Therapy–Fatigue, and Cancer Fatigue Scale. One instrument, the Cancer Dyspnea Scale, measures intensity and quality.

The intensity dimension was not as prominent in the qualitative studies. Because qualitative data do not provide a severity grading in numbers, the intensity dimension is explained in other terms, such as being bothersome, offensive, or unbearable. This dimension is also explained through relationships between symptoms, such as the presence of one symptom was the reason another symptom worsened (Belqaid et al., 2018; Kiteley & Fitch, 2006) or that symptom severity was graded compared with other physical problems (Kiteley & Fitch, 2006). The severity dimension was also partially related to illness status because disease progression intensified some symptoms. Noticeably, co-occurring symptoms that individually have relatively low severity appear to have a cumulative impact on patients and may cause higher distress (Kiteley & Fitch, 2006; Molassiotis, Lowe, Blackhall, & Lorigan, 2011).

Distress: Four instruments, the Brief Fatigue Inventory, Brief Pain Inventory, Cancer Fatigue Scale, and Functional Assessment of Cancer Therapy–Fatigue, measured the intensity and distress dimension. In the qualitative studies, the distress dimension was commonly described as highly dependent on the patients’ experience and knowledge of the symptoms. Insecurity and novelty of symptoms increase distress, and experience, acceptance, and knowledge infer lower distress. Patients also described how the level of distress was related to how the symptoms affected their daily life. A higher interference causes more distress (Belqaid et al., 2018; Kiteley & Fitch, 2006; Maguire et al., 2014; Molassiotis, Lowe, Blackhall, & Lorigan, 2011; Molassiotis, Lowe, Ellis, et al., 2011). Also, symptoms possibly related to disease progression or treatment not working would cause higher distress for some (Moassiotis, Lowe, Ellis, et al., 2011) but be of less importance to others (Molassiotis, Lowe, Blackhall, & Lorigan, 2011). Similar to intensity, the distress dimension involved descriptions of the additive effects of multiple symptoms (Maguire et al., 2014; Molassiotis, Lowe, Blackhall, & Lorigan, 2011; Molassiotis, Lowe, Ellis, et al., 2011). The level of distress may vary depending on setting but can differ in descriptions between studies. For example, cough was described as being particularly prominent and causing a higher level of distress because of the association with visibility and embarrassment in public or disturbing family members (Maguire et al., 2014; Molassiotis, Lowe, Ellis, et al., 2011), but also as not being particularly bothersome to others because many were former smokers and had previous experience of cough in the past (Molassiotis, Lowe, Blackhall, & Lorigan, 2011). The symptoms perceived as most distressing would change during the course of their illness trajectory (Molassiotis, Lowe, Blackhall, & Lorigan, 2011). Not being met with respect and understanding for bothersome symptoms and lack of emotional support were described as causing increased distress (Belqaid et al., 2018; Molassiotis, Lowe, Ellis, et al., 2011).

Timing: Six instruments, including the Hospital Anxiety and Depression Scale and Center for Epidemiological Studies–Depression, measured the timing dimension. In the qualitative articles, symptoms were described as having unpredictable timing (Belqaid et al., 2018) and being long-lasting and ever present (Belqaid et al., 2018; Molassiotis, Lowe, Blackhall, & Lorigan, 2011) and related to certain activities and time points (Belqaid et al., 2018; Maguire et al., 2014; Molassiotis, Lowe, Ellis, et al., 2011). Some symptoms could increase at a specific time related to oncologic treatment, such as anxiety building up in the days before a chemotherapy cycle or aggravated cough because of radiation therapy treatment (Molassiotis, Lowe, Blackhall, & Lorigan, 2011). Patients described symptoms such as fatigue and pain in relation to how much they needed to rest and how they felt before and after their rest (Kiteley & Fitch, 2006). Several descriptions described the occurrence of one symptom attributing to the onset of another (Belqaid et al., 2018; Maguire et al., 2014; Molassiotis, Lowe, Ellis, et al., 2011), and symptoms induced or intensified by being in a certain environment (Belqaid et al., 2018; Molassiotis, Lowe, Ellis, et al., 2011).

Quality: Only one instrument, the Cancer Dyspnea Scale, was used to assess the quality dimension. Narratives regarding the quality dimension were sensory and location descriptions related to the symptoms (Belqaid et al., 2018; Kiteley & Fitch, 2006; Maguire et al., 2014; Molassiotis, Lowe, Ellis, et al., 2011) and bodily perceptions (Kiteley & Fitch, 2006; Maguire et al., 2014), particularly when describing how the symptoms made them feel. Some described how symptoms responded to interventions (Belqaid et al., 2018; Molassiotis, Lowe, Ellis, et al., 2011). There were patients who struggled to find the right words to explain the sensation but knew that something had changed (Belqaid et al., 2018; Molassiotis, Lowe, Ellis, et al., 2011).

Discussion

Results

A large diversity of cluster constellations appeared in the 53 articles. Most studies (40 of 53) explored symptom clusters de novo and the rest a priori, differing in content regarding symptoms and number of symptoms in their constellations. Additional comparisons between a priori and de novo approaches are necessary for understanding symptom clusters (Xiao, 2010), and further considerations may be needed when selecting a priori clusters. Fatigue was the predominant symptom across all studies, and other commonly occurring symptoms in clusters were dyspnea, pain, depression, and cough. In de novo clusters, there was a greater occurrence of nutritional impact symptoms.

Forty-seven articles used quantitative/mixed methods with a large number of different symptom assessment instruments, and a few nonconventional authors developed instruments for measuring intensity alone or in combination with timing. Six were qualitative articles describing distress, timing, and quality dimensions but had fewer accounts of the intensity dimension. Because the findings indicate cluster differences depending on the methodology, there is a need to refine how symptom clusters are measured and managed, where fatigue and nutritional aspects stand out. It is likely that nausea and vomiting are temporary and treatment related; therefore, appetite loss was selected as the nutritional symptom for comparison because it may affect the patients during a greater part of the cancer care continuum. Symptoms affecting nutritional intake and development of malnutrition in patients with lung cancer affecting health-related outcomes have been identified (Kiss, 2016; Polański et al., 2021). Across a variety of treatment modalities and disease stages, malnutrition ranges from 45% to 69% (Kiss et al., 2014). Nutritional screening and supportive care are likely significant in symptom management because patients experience better health-related quality of life, lower overall symptom burden, and better prognosis (Gul et al., 2021; Polański et al., 2017).

Fatigue has a range of causes and coexists with many other symptoms; therefore, treatment requires an interprofessional effort with medication, exercise, nutrition, and other therapeutic approaches (Stone et al., 2023). A multimodal symptom management intervention targeting exercise and nutrition can improve PROMs and symptom burden in patients with advanced lung cancer (Ester et al., 2021). Relative to depression and anxiety, fatigue yields a more negative effect on lung cancer survivors’ health-related quality of life (Jung et al., 2018), but exercise has been shown to improve capacity and health-related quality of life in cancer survivors (Peddle-McIntyre et al., 2019). There is also a challenge in considering fatigue as one symptom or as two separate phenomena: physical fatigue and mental fatigue (de Raaf et al., 2013). In this review, multidimensional aspects of fatigue have been considered, but it is defined as one symptom. Regarding measurement of cancer-related fatigue, European Society for Medical Oncology guidelines suggest a 10-point numeric rating scale for fatigue as the best screening tool. However, using a more specific questionnaire, such as the Brief Fatigue Inventory, to assess moderate to severe fatigue could be necessary. Because fatigue often occurs with related symptoms, a screening tool that captures multiple symptoms may also be of clinical value. There is no clear recommendation regarding the most appropriate subjective measure, so there is a need for comparable data to reliably detect changes over time (Fabi et al., 2020).

Patients with lung cancer experience a large diversity of symptom cluster constellations that fluctuate and vary in intensity, timing, distress, and quality over time. “Unpleasant symptoms, in all their synergy, interaction, and complexity, are what the whole patient presents” (Lenz et al., 1997, p. 25). This complex multidimensional symptom experience is important to consider when measuring symptoms, for example, because a patient may score high on intensity but may not be so bothered by that symptom because the timing dimension may be limited or their knowledge and coping skills make it manageable. Conversely, a lower-intensity symptom may be more incapacitating because of the quality aspects or meaning for the patient. The intensity dimension was the most commonly measured dimension among the quantitative studies. However, intensity was not the most prominent dimension from the patients’ perspective in the qualitative studies, but rather distress, timing, and quality. The TOUS can provide a clinically relevant conceptual mapping of the symptom dimensions and suggest how and why they are important to measure and a selection of variables for clinically useful research. The restriction of measuring an individual dimension is considered inadequate, and it is recommended that each symptom be measured separately with multidimensional measures (Lenz et al., 1997). Symptom dimensions should be important when developing and evaluating symptom management strategies, not just for the assessment itself (Dodd, Janson, et al., 2001; Humphreys et al., 2014). When and by which instrument a symptom assessment is conducted will affect the accuracy and relevance of that measurement. Using multidimensional scales to measure the complex nature of a symptom cluster was suggested by Barsevick et al. in 2006, but this scoping review indicates the methodologic issues in measurement and comparison of symptom items and dimensions because of the vast number of instruments used and absence of multidimensional assessments within the lung cancer context alone. There is a need to further evaluate the validity, reliability, and responsiveness of PROMs instruments in symptom cluster research (Miaskowski et al., 2017) because this stringency is still missing.

Method

The scoping review gives width and possibilities in the searches but is a challenge when summarizing the large amount of data collected. The authors’ original plans for the review process developed into dividing their research questions into two parts. Part 1 has been presented in this scoping review, and part 2 will be presented as a separate study (in manuscript) as an integrative review with the same sample but other specific research questions linked to the theoretical framework as stated in the authors’ protocol (Karlsson et al., 2020). Part 2 aims to explore how influencing factors affect the patients’ experience of symptom clusters and the consequences in their daily life. An option could have been to deselect the de novo articles without a cluster aim to render a smaller final sample with a full focus on explicit symptom cluster research. No limitation was set regarding earliest year of publication because co-occurring related symptoms have been investigated before the concept of symptom clusters was introduced. A limitation to more current years could have also decreased the final sample but then omitted important findings.

Although lung cancer is a divided group of molecularly and histologically heterogeneous subtypes (Travis et al., 2015), and the incidence, mortality, and therapy options vary between subtypes (Howlader et al., 2020), this review included a wide population with varying subtypes and in various phases of the cancer care continuum: before, during, or after treatment. This may affect the occurrence of specific symptom clusters, depending on when the symptoms have been measured and dimensions measured. For example, nausea and vomiting are symptoms more likely to occur during chemotherapy than at diagnosis, and the presence of cough may decrease as treatment decreases the tumor burden or increase as a side effect of treatment. As reported by Kalantari et al. (2022), changes occur in the relationships and interconnections between and among symptom clusters, depending on the time point in the treatment period and type of cancer. Therefore, these results should be interpreted with caution regarding specific treatments, such as immunotherapy that has a different action mechanism and side effects compared with standard chemotherapy or radiation therapy or subgroups of patients who may have a different symptomology. However, the key symptoms of fatigue, distress, and appetite loss appeared among the five symptom clusters (general, immunotherapy related, pulmonary, gastrointestinal, and neural) discovered in a study of symptom clusters in patients with lung cancer who were treated with programmed cell death protein 1 immunotherapy (Zhang et al., 2022) were comparable to the current results. In the study by Zhang et al. (2022), five symptom clusters were identified as follows: (a) the general cluster, (b) the immunotherapy-related cluster, (c) the pulmonary cluster, (d) the gastrointestinal cluster, and (e) the neural cluster.

The authors did not analyze the methodologic approaches regarding statistical methods used, which may have provided additional knowledge regarding the most appropriate analytical method to create symptom clusters and awareness concerning common and unique underlying mechanisms of symptom clusters (Ward Sullivan et al., 2018). The authors did not limit the inclusion related to a specific level of correlation of symptoms in a cluster regarding the p value, which contributed to the large number of individual clusters.

Methodologic rigor was shaped by involving two to five team members in the five steps (Arksey & O’Malley, 2005), supporting a systematic and transparent review process.

The theoretical framework provided direction regarding the categorization of symptom dimensions in the PROMs instruments, but the authors’ interpretation of the dimensions may not equate to others’ interpretations.

Implications for Nursing

Symptom clusters among patients with lung cancer are numerous and with varied symptom patterns, some more common than others. Fatigue is the most prevalent symptom among the many symptom clusters, and it is essential for oncology nurses to perform timely assessments, ensure that person-centered supportive care is provided to reduce fatigue, and continuously evaluate patients’ needs. Compassionate nursing and patient education regarding symptom management is recommended to reduce distress, as well as psychosocial interventions and individually adapted physical activity. If patients’ fatigue-related distress is addressed, it may also positively affect the presence of related psychological symptoms. Because nutritional impact symptoms are highlighted in this population, interventions and symptom management to improve nutritional status and reduce symptom burden are important and may also reduce fatigue and psychological symptoms.

In clinical practice, the patient–nurse relationship enables assessment of multiple symptom dimensions by the use of appropriate instruments in combination with person-centered communication. Influencing factors need to be considered when assessments are made and interventions proposed. Having access to the advanced expertise and continuity of high-quality care provided by an oncology nurse specialist may be vital for this population.

Because patients with lung cancer are not always capable of engaging in the recommended moderate to high levels of physical activity because of fatigue-, dyspnea- and nutrition-related problems, the authors recommend a team approach involving oncology nurses, dietitians, physicians, and other healthcare professionals to develop, support, and follow up on individual person-centered rehabilitation plans.

The symptom assessments and interventions need to be performed with patients’ life expectancy in mind because not all patients would benefit from the same types of interventions.

Conclusion

This scoping review shows an abundant variety of symptom clusters among patients with lung cancer. The symptoms occurring in clusters vary if researchers have defined the clusters in advance (a priori) or not (de novo). Overall, fatigue is the symptom found to most frequently co-occur with other symptoms in clusters.

Noticeably, the a priori clusters contain dyspnea more often, and the de novo clusters contain pain, cough, depression, and nutritional symptoms more often. Because of the differences between symptom clusters related to the a priori or de novo approach, future symptom cluster research should consider this.

The qualitative studies, although being quite few, complement the findings from the quantitative studies in this sample, endorsing the presence of clinically significant symptom clusters among patients with lung cancer, and contribute evidence regarding the quality dimension that is missing in most symptom assessment instruments.

The importance of considering the multidimensional aspects of the symptom cluster experience is vital in future research and clinical practice to sustain the holistic approach to patient-centered care.

Because certain symptoms are more likely to cluster than others, future research regarding a potential underlying biologic etiology by measuring specific biomarkers could be valuable.

The authors gratefully acknowledge Annelie Ekberg-Andersson, librarian at Karlstad University, for valuable assistance with finding the relevant literature and Catriona Kennedy, PhD, professor in the School of Nursing and Midwifery at Robert Gordon University in Aberdeen, Scotland, and Siew H. Lee, PhD, lecturer in the School of Nursing and Midwifery at Robert Gordon University, for advice regarding the analysis.

About the Authors

Katarina Karlsson, RN, is a PhD student in the Department of Health Sciences in the Faculty of Health, Science, and Technology at Karlstad University and an oncology nurse in the Department of Clinical Oncology at the Central Hospital of Karlstad and the County Council of Värmland, all in Sweden; Cecilia Olsson, RN, PhD, is an associate professor in the Department of Health Sciences in the Faculty of Health, Science, and Technology at Karlstad University and an associate professor in the Department of Bachelor in Nursing at Lovisenberg Diaconal University College in Oslo, Norway; Ann Erlandsson, PhD, is an associate professor and senior lecturer in the Department of Environmental and Life Sciences/Biology in the Faculty of Health, Science, and Technology at Karlstad University; Karin M. Ahlberg, RN, PhD, is a professor in the Sahlgrenska Academy Institute of Health and Care Sciences at the University of Gothenburg; and Maria Larsson, RN, PhD, is a professor in the Department of Health Sciences in the Faculty of Health, Science, and Technology at Karlstad University, all in Sweden. This research was supported, in part, by the County Council of Värmland, Sweden. Karlsson completed the data collection. All authors contributed to the conceptualization and design, provided the analysis, and contributed to the manuscript preparation. Karlsson can be reached at katarina.karlsson@kau.se, with copy to ONFEditor@ons.org. (Submitted June 2023. Accepted September 12, 2023.)