2.1 Study eligibility criteria

Studies were deemed eligible if their population was exclusively or mainly men with PCa, and were reporting on body image, self-esteem and masculinity outcomes, written or fully translated into English. Due to the paucity of literature in specific validation studies of tools to measure masculinity, body image and self-esteem, search criteria were widened to include experimental, observational and randomised controlled studies that used questionnaires as part of a predefined tool to measure the above outcomes. Studies involving mixed cancer populations where results specific to the PCa cohort were not reported were excluded, as were systematic reviews. Studies that did not calculate any validity measures for their tool (such as internal consistency/reliability) in a PCa cohort were also excluded.

2.2 Information sources and search

Pubmed, Embase, Medline and Psychinfo were searched for eligible studies from inception up to 6th May 2020. The search terms used were: (prostate cancer OR prostate neoplasm) AND (body image); (prostate cancer OR prostate neoplasm) AND (masculine OR masculinity); (prostate cancer OR prostate neoplasm) AND (self esteem). A reference review of identified systematic reviews was conducted for additional studies, and grey literature (unpublished studies) was searched for via abstracts available on EMBASE.

2.3 Study selection

Two independent reviewers (J.B. + O.B.) screened titles, abstracts and subsequently full texts against eligibility criteria post de-duplication of results. Discrepancies after full text review were discussed until 100% agreement was reached. After study selection, the tools used in each study were summarised.

2.4 Data collection and items

Data extraction was conducted by two independent authors (J.B. + O.B.). All data was extracted to a pre-defined extraction sheet detailing the number of patients the tool was used in, the country and language in which it was used, age and demographic data of the participants, primary outcomes of each study and method of collecting responses. Data extracted specific to the tools used were name of the tool used, author of the tool, whether the tool had been modified for the study, domains, response format, scoring method and whether the tool had been developed specifically for PCa patients, and if available, information on structural validity, internal consistency, discriminative validity and overall response rate from participants. Measures were extracted to objectively assess validity of each tool used based on criteria set out by the COnsensus-based Standards for the selection of health status Measurement INstruments (COSMIN) checklist.¹¹ This checklist assesses the methodological quality of studies assessing the measurement properties of patient reported outcome measures (PROMs). It determines the extent to which aspects of a PROM have been assessed and reported on by studies, specifically if they have evaluated the reliability, validity, responsiveness and interpretability. Thereby, it provides an overall assessment of the quality of a PROM being assessed.

One author (Hoyt) was contacted personally for more information about the domains assessed and scoring method for the tool used in their studies.^{12, 13} We received the scale and scoring method which was used to complete Table 3.

2.5 Risk of bias assessment

Risk of bias of individual studies was assessed using the COSMIN risk of bias tool¹¹ – an adapted version of the COSMIN checklist specifically for use in systematic reviews to assess the methodological quality of single studies on measurement properties of PROMs. This evaluated the quality of individual constructs assessed in a study, including PROM design, validity, consistency and responsiveness. Pre-determined criteria for evaluating the quality of each construct exist, giving them a rating of “very good,” “adequate,” “doubtful” or “inadequate” Relevant sections of the tool were used depending on which constructs were reported in studies, with an overall risk of bias table generated for each construct and study.

3.1 Study selection

A total of 2126 studies were identified, with a further eight added through reference review and conference abstract databases (Figure 1). Initial screening resulted in 136 full-text articles assessed for eligibility. Of these, a final 46 studies were included in the review.

The final number of tools identified was 17. Of these, seven were modified versions of tools already identified, either using subscales,^{4, 8, 14-16} specific items,^17-20 removing specific items²¹ or translating the tool to an alternative language.^{22, 23} There was one tool used to measure self-esteem, nine to measure masculinity, and seven to measure body image. Tools identified as well as validation articles in PCa patients are summarised in Table 1.

3.2 Body image

Body image was assessed in a total of 2753 patients, using seven different tools across 10 studies (Table 2).^{4, 21-29} The most frequent of these was the Body Image Scale (BIS), used in 697 participants.³⁰ Participants ranged in age from 54 to 90 and were undergoing a variety of treatments, including radical prostatectomy, ADT and radiation. None of the body image tools were specifically developed for PCa cohorts, instead being tested in other cancer populations,³⁰ adults and adolescents³¹ and American undergraduates.³² No studies aimed to specifically validate the tools used; seven evaluated internal consistency using Cronbach's alpha.

The BIS by Hopwood et al³⁰ was used twice in English,^{26, 28} twice in other languages^{22, 23} and once with one of the items removed.²¹ The BIS, although intended for use in all cancer populations, was originally developed and validated in breast, large bowel, testicular, gynaecological cancers and lymphoma. In the studies identified by this review, internal consistency was stated for four of the five studies, being sufficient in all (Cronbach's alpha = 0.72-0.90).

The “Concern about Body Image” scale was used in one study of 1089 patients undergoing radical prostatectomy.⁴ It was adapted from research done by Clark et al on men with metastatic PCa, where body image was identified as a dimension of QoL.³³ The scale was found to have acceptable internal consistency (Cronbach's alpha = 0.82) and was tested on the largest group of patients of all body image tools identified.

The Physical Self Perception Profile was used in one study of 58 men, having been originally validated in a college-aged, largely active population, based on applying self-esteem theory to physical perception.³² Three of the subscales of this tool relate to and are predictive of physical activity – strength, sports participation and conditioning. This suggests this scale may be less adequate in measuring body image in men with cancer, where fatigue is one of the most commonly experienced side effects of treatment.³⁴

The body image tool used by Moinpour et al in 737 men consisted of 15 treatment-specific items generated from input from nurses and physicians, (Table 2), although no specific psychometric evaluation of the tool in PCa patients is mentioned.²⁷

The MBRSQ-Appearance Evaluation subscale was used in one study of 104 men,²⁴ having been developed by Cash et al³⁵ for use with college students. In previous studies, the entire Appearance Evaluation subscale has been found to be unstable, in particular the item “Most people would consider me good-looking,” which has shown associations with both gender and weight, where overweight was associated with instability.³⁶ Internal consistency was sufficient (Cronbach's alpha = 0.981).

A modified BIS adapted from the Derogatis Sexual Functioning Inventory³⁷ was used by Taylor-Ford et al,²⁹ in a study of 74 men yet to undergo treatment. This was developed in university students, and in initial evaluation the body image subscale was found to have one of the lowest coefficients of internal consistency (0.58, N = 325).³⁷ Derogatis et al also found the body image subscale was profoundly affected by sexual functioning, where female and male patients experiencing sexual dysfunction reported body image scale values 1.5 standard deviations below the normative mean. Considering the prevalence of sexual dysfunction in men diagnosed with PCa, use of this body image scale on its own may be in fact measuring body image related to men's sexual dysfunction instead of body image as a discrete concept.

3.3 Masculinity

Masculinity was measured in 9021 patients using a total of nine different tools (Tables 1 and 3), and was the outcome measured most frequently, being used in 32 studies.^{5, 8, 12-20, 38-58} The age range of all patients was 42 to 87 years, and studies were carried out in the United States (US), Canada, South Africa and Australia. Five tools had validation articles where they were used specifically in men with PCa in their complete form.

The Bem Sex Role Inventory – Short Form was used in three studies of 280 patients total, with one study also adapting the tool to create an Importance of Sex Role Inventory by changing the response options to ranking the importance of each item instead of agreement.⁴⁶ The short form of the tool was shown to have higher levels of reliability in university students (alpha = 0.82-0.89) compared to the long form (alpha = 0.81-0.85),⁵⁹ while a shortened, Spanish translation of the scale validated in an older Brazilian population (mean age = 74.4) was shown to differentiate two factors corresponding to masculine and feminine scales.⁶⁰

Cancer Related Masculine Threat is a tool developed by Hoyt, and used in both of his studies measuring masculinity in 127 men with PCa.^{12, 13} Development was informed by current literature as well as interviews with men with cancer to measure the discord between men's experiences and beliefs regarding cancer, and their masculinity.¹³ It was found to have moderate positive correlations with other measures endorsing traditional aspects of masculinity: the Male Role Norms Inventory – Revised⁶¹ (r = .52, P < .001), and the unmitigated agency subscale of the Extended Personality Attributes Questionnaire⁶² (r = .48, P < .01). This confirmed Cancer Related Masculine Threat as a related although distinct concept. The tool also showed stability between the scores recorded at baseline and 8 weeks (r = .85, P < .001).¹³

The Conformity to Masculine Norms Inventory (CMNI) was used in four studies totalling 769 patients, either in an abbreviated form,¹⁴ or with certain subscales used.^{8, 39} The CMNI is described as encompassing dimensions of masculinity in the United States.⁶³ Factor analysis and development were carried out with students, with internal consistency ranging from coefficient alpha = 0.72 to 0.91 for each of the 11 subscales, with test retest validity for the 11 subscales over 2 weeks ranging from 0.76 to 0.95.⁶³ Internal consistency in PCa patients ranged from Cronbach's alpha = 0.56 for the Dominance subscale,¹⁵ to 0.89 in the Emotional Control subscale.³⁹ Only one study reported on missing responses, stating 88/353 (24.9%) participants left entire measures, or >10% of any given measure blank.

The EORTC-QLQ-PR25, PC-QoL and FACT-P are all scales developed specifically to measure health outcomes in men with PCa, and were used in a total of 19 studies, comprising 7369 men. Four studies used item 19 from the EORTC-QLQ-PR25 “Have you felt less masculine as a result of your illness or treatment?,” although none reported on internal consistency of the measure. In an international field study validating the EORTC-QLQ-PR25 in 13 different countries and 642 PCa patients, the hormonal treatment-related subscale, which contains item 19, showed the lowest internal consistency both at baseline and after treatment (Cronbach's alpha = 0.41 at baseline, 0.39 after treatment), leading the authors to recommend that results from this scale be reported at the individual item-level,⁶⁴ supporting the use of item 19 alone.

The PC-QoL masculine self-esteem subscale was used by 14 studies spanning three countries. Three of these were follow-up studies of the same cohort in which the subscale had previously been used.^{40, 42, 43} The scale was developed using qualitative interviews with 130 PCa patients, and validated in a sample of 349,⁵ leading to the definition of 11 domains, with some specific to those with PCa. The masculine self-esteem subscale showed good internal consistency (0.91), although three items concerning feeling less of a man converged with the sexual intimacy subscale, with the correlation between masculine self-esteem and sexual dysfunction higher in PCa patients (−0.21 vs −0.42 in comparison patients). This suggests PCa patients were more likely to conflate masculinity with sexual performance than their counterparts. All studies reporting satisfactory internal consistency values (Table 1) demonstrated that masculine self-esteem exists as a discrete and reliable construct.

The Masculinity in Chronic Disease Inventory (MCD-I) was used in two studies.^{16, 58} The original tool was developed specifically for use with PCa patients (Table 3), undergoing the most extensive psychmetric evaluation out of all tools in this review (Appendix 1). Chambers et al developed the tool from focus groups with 26 expert patients and healthcare professionals and 15 survivors, and conducted initial face validity testing with a convenience sample of 19 men with PCa. It was subsequently tested in 403 Australian men with PCa, and was found to be acceptable, valid and reliable, with convergent and divergent validity analyses confirming that the six domains assessed were relevant to men with PCa. Goodwin et al used the scale in a group of 622 men with chronic diseases such as hypertension, hypercholesterolaemia, osteoarthritis and skin cancer and recommended the use of a 5 factor structure in future, which also showed sufficient structural validity according to the COSMIN criteria (Appendix 1).⁶⁵

One study used the “I am able to feel like a man” item from the FACT-P tool, developed by Esper et al⁶⁶ as an adjunct to the Functional Assessment of Cancer Therapy (FACT),⁶⁷ which has been used to date in over 26 000 patients.⁶⁸ The PCa subscale, within which the “I am able to feel like a man” item is contained, showed acceptable internal consistency in initial evaluation (alpha = 0.65-0.69) within two cohorts of US men with localised and advanced PCa,⁶⁶ and was able to differentiate between patients at different disease stages.

The MRNI-R restrictive emotionality subscale (Table 2) was originally developed and tested in a group of US undergraduates,⁶¹ and the restrictive emotionality subscale showed good internal consistency within the initial sample (Cronbach's alpha = 0.87) and PCa cohort (Cronbach's alpha = 0.85). The results from Darabos et al, where the 66 men sampled somewhat disagreed with restrictive emotionality norms, has been supported in Levant's use of the tool in older men, but also suggests a tool more specific to the often older PCa cohort may have been warranted.⁶⁹

3.4 Self esteem

Self-esteem was measured in four studies totalling 624 US patients,^{3, 70-72} all using the Rosenberg Self-Esteem scale (RSE; Table 4).⁷³ The scale was developed in 1965 and validated in American students.⁷³ Internal consistency values were reported by two studies and were both acceptable (0.84³ and 0.88⁷¹). However, a review of the use of the scale in cancer patients by Curbow et al suggested that items used by the RSE may not be the only important concepts relevant to cancer patients (and furthermore, PCa patients), as cancer patients do not seem to differ from the general population in their levels of self-esteem, and because null effects are often seen in interventions trying to target self-esteem in cancer patients.⁷⁴

5.1 Review limitations

As with any systematic review, this study has limitations. The large amount of heterogeneity between tools, and the use of subscales or abbreviated forms of tools for which no validation had occurred limited the comparisons that could be drawn between different methods of measurement. Furthermore, many tools underwent limited psychometric evaluation, meaning a full COSMIN assessment was not possible. Finally, despite the comprehensive search strategy utilising two authors, pertinent articles within the literature could still have been missed within this review considering the broad area of interest within in this review.

5.2 Clinical implications

The reframing of PCa as a chronic disease has led to increased emphasis on treatment side effects, benefits and harms that men may experience, as these will have a lasting impact on their lives and mental wellbeing. Tools to measure men's health related QoL can be used to help guide treatment decisions, understand what is important to men in the course of their diagnosis and treatment, and facilitate discussions with clinicians.

Body image, masculinity and self-esteem can all impact men's mental health extensively, which further affects their QoL.^{8, 14, 42, 70, 71} Furthermore, there is a relationship between physical functioning and quality life measures. Multiple studies have found sexual functioning was linked to men's perceptions of their masculinity,¹⁹ with a similar relationship between sexual functioning and self-esteem,⁸⁰ and depression.⁵⁰ The connection between physical functioning and QoL indicators, although difficult to ascertain casuality, highlights the fact that better tools to measure these factors could help clinicians recognise difficulties in physical and mental functioning their patients may be experiencing. This review also emphasises that tools to measure body image, masculinity and self-esteem in PCa patients should be used with caution, as many of these tools have undergone limited validation, and may not be measuring constructs which are relevant to men with PCa.

We highlight the current lack of tools specifically developed for PCa patients, showing how current studies and clinicians using these tools to measure the outcomes of body image, masculinity and self-esteem in PCa patients may not be measuring these constructs in a way that is sensitive to the unique position of these patients. Therefore, more qualitative research to help establish the domains of these outcomes that are important to men with PCa, followed by validation studies of current tools is needed. This will enable clinicians to use these tools in clinical practice to help understand how their patients' mental wellbeing can be best supported during their diagnosis and treatment.