CRF is one of the most common, distressing conditions experienced by people with cancer. It has detrimental impacts on daily functioning and overall quality of life [19, 20]. The absence of an accepted definition and standardized approach for assessing CRF has contributed to the heterogeneity of methods and results across published studies. Synthesizing the current literature in adults with cancer that includes CRF as a primary outcome is essential to identify gaps in the literature and provide recommendations for future research to improve health outcomes in adults with cancer. This scoping review identified, collected, and summarized information to: 1) describe how CRF has been defined, 2) determine how CRF has been assessed, and 3) characterize the articles and samples reviewed. Out of 150 included articles, CRF definitions and methodological procedures (e.g., study designs, clinical measures and dimensions) varied widely, confirming the need for a single agreed-upon CRF definition and assessment battery.

Consensus has not been reached regarding how to define CRF, as reflected in this review in which only two-thirds of the articles defined/described CRF. Less than half of the articles that defined/described CRF used the NCCN’s definition, and the articles with non-NCCN definitions/descriptions were highly heterogeneous, providing further evidence that the field is lacking consensus on how to define or conceptualize CRF. As the debate continues whether CRF is a definable condition, the need to use a consistent definition is critical to translate research findings to the clinical setting and accurately describe, diagnose, and manage CRF.

Of the articles that contained definitions/descriptions, the majority included the word ‘multidimensional’ and/or listed more than one dimension. Some authors may have avoided using the term multidimensional to describe CRF because it is still debated whether CRF should be considered a multidimensional construct (in which all dimensions share the same etiology) or a unidimensional construct (in which each dimension has a stand-alone pathogenesis) [21, 22]. In addition, all dimensions that end up being included in the agreed-upon definition should also be operationalized to solidify a collective understanding of the construct of CRF.

A recent review of CRF among childhood cancer survivors formulated an explicit definition based on the included reports, defining it as “a subjective, persistent, and multidimensional experience that differs from normal fatigue in the physical, emotional and/or cognitive spheres [23].” Conducting a similar thematic review in adults with CRF may be helpful to establish the groundwork for an explicit definition. In addition, creating research-based case definitions of CRF and delineating specific clinical subtypes may assist with understanding CRF’s etiology and formulating optimal management strategies; this was a recommendation made during the National Cancer Institute Clinical Trials Planning Meeting in 2013 [24] that is still unmet.

In this review, we found that nearly 30 measures have been used to assess CRF, with each measure differing widely in scope, including varying numbers of items, scale formats (e.g., visual analogue scales, Likert ratings), scoring interpretations, and reference periods for which CRF was assessed (e.g., present day, past week). The MFI-20 was the CRF measure used most often; the same questionnaire was also the most commonly used in our previous review of fatigue in non-oncologic conditions [25], perhaps because it broadly measures general fatigue as well as specific fatigue dimensions like physical, motivational, cognitive, and mental fatigue. In the current review, only about half of the articles included validity descriptions with their measures; it’s important for this information to be reported in future publications so that readers know whether the tool is validated in the specific cancer population being examined, and they can use that information when interpreting the results. Across all CRF measures, 30 unique dimensions were found, with some dimensions used more heavily (“Physical”, “Mental”, “Cognitive”) than others, perhaps based on the most common symptoms reported by a specific cancer type (i.e., breast cancer).

The current number of available tools and their heterogeneity pose a challenge for comparing and interpreting CRF findings across articles. An additional challenge is that each tool assesses unique dimensions of CRF, and these dimensions are not operationalized. Therefore, it is imperative for the field to either identify which existing CRF measure(s) sufficiently encompass the breadth of CRF as a behavioral construct or create a new CRF measure (or battery) to move toward a standardized way of assessing CRF. In doing so, the number of dimensions used to measure CRF will need to be reduced, and each dimension will need to be operationalized.

Many articles included in this review failed to provide accurate and/or complete information regarding measures used to assess CRF, including: (1) the correct name and/or citation of the assessment(s), (2) the version of the tool (e.g., short form vs. long form, English vs. French), and (3) interpretation of scores in relation to CRF. These reporting inconsistencies/omissions led us to collapse all versions of a measure into the same group with the original measure and describe the psychometric properties of only the original versions. In the future, authors must report CRF measure information in full (i.e., proper name, correct citation, version, scoring interpretation) so that consumers can confidently determine whether measures used across articles are identical and interpret findings accordingly.

CRF was primarily examined in high-income countries, in that half of the studies were conducted in either the United States, Germany, Netherlands, or China. A significant portion (43%) of articles contained small sample sizes (fewer than 100 people). Sample demographics were grossly underreported, with missing data commonly occurring for race, age, and gender; based on what was reported, samples primarily consisted of White middle-aged females. These small homogenous samples limit the ability to translate findings to the broader population, impeding the development of effective therapies. In future studies, demographic information is crucial to report because these characteristics can impact cancer-related symptoms. For example, those who are female [26] or of older age [27, 28] have been found to experience higher levels of CRF. Demographic factors that are considered social determinants of health are particularly important to include in publications because certain groups of people experience disproportionately poorer access or outcomes related to cancer care due to structural disparities and inequities [29].

Cancer type was unspecified in one-sixth of the articles. Breast cancer was the most common cancer type in which CRF was evaluated, which is not surprising since breast cancer was the number one cancer diagnosis world-wide as of 2020 [30]. The most common cancer treatment (observational articles only) was chemotherapy, likely because it is a common primary treatment for invasive breast cancer and its behavioral toxicities are well-documented [31, 32]. It was encouraging to find that longitudinal designs were more common than cross-sectional, suggesting that the long-term implications of CRF have been considered and evaluated throughout the course of disease; similarly, quantitative designs were more common than qualitative, suggesting that authors have stived to obtain objective, reliable, and generalizable results. However, observational designs were more common than experimental, likely due to the need to observe the natural trajectory of CRF as a late side effect of most cancer treatments. Although observational study designs may be the only way researchers can explore certain questions, they only allow us to determine associations with CRF (not causality) and typically contain more uncontrolled confounds, limiting the interpretations we can yield from the findings.

Although an extensive review of the literature was conducted, it is possible this review may have missed relevant articles due to the selected search terms or applied filters. Regarding search terms, CRF was challenging to operationalize due to the number of words that are synonymous with fatigue, so we opted to be as specific as possible with the words used to describe CRF (see Online Resource 2) and left out more general fatigue-related terms (e.g., lethargy, weakness). As for applied filters, we only included articles written in or translated to English, so the findings here may be more applicable to English-speaking countries and not truly representative of CRF studies worldwide. Lastly, despite our best intentions to do so, our review did not include the dimensions included in CRF definitions; this is because many of the articles provided general descriptions rather than true definitions, and it was unclear whether these descriptions were being used to operationalize the term for the study or merely re-count how it has been described in previous literature.