Factors associated with psychological distress for couples facing head and neck cancer: A systematic literature review

Cancer patients in supportive relationships display improved health and survival outcomes. Identifying factors that might respond to intervention for Head and Neck Cancer (HNC) dyads is important as HNC patients and their partners experience heightened distress. This article systematically reviewed and evaluated the research findings and methodological quality of studies which identified factors influencing psychological distress for couples facing HNC.

Furthermore, HNC presents difficulties for couple-level functioning relating to intimacy and sexuality; this includes treatmentrelated consequences that impact on physical connection such as the presence of a feeding tube, body image concerns, and oral difficulties such as a persistent dry mouth or excessive salivation. [18][19][20] Additionally, the physical sequelae of HNC can impact upon couple-level communication; the structures commonly affected by treatment are important for both verbal and non-verbal communication, such as speaking and achieving facial expressions. 21 These difficulties are reflected in research indicating a decline in marital functioning one year post-treatment. 22 Another study reported that 83% of HNC spouses and 100% of patients stated that there was increased marital dispute during treatment. 23 It is important to support couples to maintain relationships while they experience the stressors of diagnosis and treatment, as supportive partner relationships are implicated in increasing quality of life, improved adaptation to the cancer, and better survival rates; 6 Research found that being married reduced the chance of death for HNC patients by 33% 24 and substantially improved performance status during treatment in comparison to unmarried patients. 25 Research recommends that developing couple-based interventions for dyads experiencing HNC could focus on communication and collaboration. 26 Understanding a wide range of influencing factors is important, as the development of couple-based interventions has been predominately aimed at patients and partners experiencing breast or prostate cancer. 27 However, a broader understanding of the psychological, clinical, relational, social, and demographic factors 28 that influence dyadic psychological distress, either as causes or exacerbating/buffering factors, is currently missing from the HNC literature. The development of this understanding would make it easier to both identify malleable factors which could be targeted in such couple-based interventions and screen for/assess couples who may be especially vulnerable to experiencing psychological distress.
Psychological distress is variously defined across literature. This review defines it as an aversive emotional state which is experienced by an individual when they are presented with a stressor or demand that causes either temporary or permanent harm. 29

| Rationale for current review
While other reviews have examined factors associated with depression and quality of life in HNC patients, 7,8 no review has systematically investigated the factors influencing distress for both HNC patients and their partners. This review aims to systematically identify and synthesise the key findings of studies that identify these associated factors. Furthermore, the review aims to provide a detailed appraisal of the methodological quality of the identified papers. Due to evidence indicating that couples facing HNC experience heightened levels of distress, 30 a clearer understanding of factors influencing this experience could inform clinical practice and add to existing theories such as the relationship intimacy model of couples' psychosocial adaptation to cancer. 31 This model suggests that couples affected by cancer adopt relationship behaviours which either increase or reduce dyadic closeness: this closeness is hypothesised to influence the couple's adaptation to cancer. 31 A broader understanding of any influencing relational and psychological factors warrants particular attention, given the potential malleability of couple-level relational interactions and individual-level psychological responses/appraisals in reducing psychological distress, as compared to demographic and clinical factors. However, an awareness of the influencing demographic, social, and clinical factors is also important in supporting clinicians/researchers to screen for/ assess couples who are vulnerable to experiencing greater distress.

| OBJECTIVES
(1) Systematically identify and synthesise key findings of studies that identify factors associated with psychological distress for both HNC patients and their partners

| METHODS
The review was pre-registered with the International Prospective Register of Systematic Reviews (PROSPERO: ID = CRD420202 13101)

| Eligibility criteria
Studies were considered for inclusion if they satisfied the following criteria: 1. Consisted of a sample of couples where one member had experienced HNC. Adult patients at any stage of HNC were included.
In line with medical literature 32,33 and previous systematic literature reviews, 34

| Systematic search
Research articles were systematically searched for using the Ovid host (www.ovid.com); the CINAHL, PsycINFO, and MEDLINE databases were searched from the start of the database to the 20 October 2020. A research librarian was consulted to generate individualised search terms for each database (Appendix S1). Search terms consisted of medical subject headings (MeSH) and free-text search terms (Appendix S1). The three key review search concepts were 'partners,' 'head and neck neoplasms,' and 'psychological distress'; however, several search terms were produced for each key term to ensure thorough searches. The search terms were exploded where appropriate to access articles using different terminology.
Following the identification of relevant search terms, the list of terms for each key concept were linked using the Boolean operators of 'or/ and' to produce the search results for the three databases. The search results were gathered and any duplicates between databases were removed using the Mendeley reference manager (www.mendeley.com). The reference lists of included articles were handsearched to identify additional articles.

| Data extraction
Articles were initially screened by title and abstract by a single investigator (LM). All results from one database were additionally screened by a second author (NM) to generate an inter-rater reliability score: overall weighted kappa = 1.00 (perfect agreement).
When it was not possible to assess eligibility through abstract and title screening, full-text copies of the articles were sourced and reviewed by one author (LM) to determine eligibility. Data were extracted using a pre-designed

| Quality assessment
The first author rated the methodological quality of the articles using a quality appraisal tool. There is no 'gold standard' quality appraisal tool 39 so an adapted version of the Mixed Methods Appraisal Tool was used (MMAT). 40 This was the preferred choice as it supplied criteria on both quantitative descriptive studies and quantitative non-randomised studies which was appropriate based on the heterogenous study designs represented by the included papers. The criterion of 'Have relational sample characteristics been fully reported?' was included to enhance the methodological quality assessment regarding which studies had provided full details on the relational details of their sample. Each study was rated as follows against each criterion; 2 = 'Yes,' 1 = 'Cannot tell,' 0 = 'No.' The individual scores were summed, generating a score out of 18; a higher score reflects a study of higher quality. Hong et al. 40 recommend not making MCCABE-WHITE ET AL.
inferences regarding study quality and risk of bias based on the overall score. It is suggested that the ratings of each criterion are supplemented by detailed explanations of how the score was chosen to clarify the quality of the studies. 40 The modified appraisal tool used can be found in Appendix S2; this tool also provides operational definitions for individual quality items. To assess the inter-rater reliability of quality appraisal, a subsample of studies (50%) was randomly selected for double coding by a second reviewer (a coauthor), with any differences of opinion resolved through discussion.
Prior to discussion, the overall weighted kappa = 0.965.

| Coding of influencing factors
A coding system was developed to allow conceptual grouping of factors, producing a common nomenclature for the current review. These factors were coded as follows. The factors were assigned to one of four higher-order (general) categories: Clinical, Sociodemographic and social network, Relational, and Psychological. Within each higher-order category, lower-order (specific) categories were found to identify the relevant factors more specifically. The higher-order and lower-order categories are displayed in Table 1.

| Data analysis
The decision to not conduct a meta-analytic synthesis was informed by the marked heterogeneity across articles, particularly regarding timings of outcome assessments and measurement of psychological distress. A narrative synthesis of the quantitative data is provided and the effect sizes of the association between the influencing factors and psychological distress are reported in Table S2 41 (available as supplementary information) and explored. Specifically, when synthesising across studies/estimates, we report effect-size ranges and medians to quantify associations of interest without making assumptions about underlying distributions. 42 Where a study reports multiple coefficients for a given category of factors or range of (patient and partner distress) outcomes, we take the median estimate

| Study and sample characteristics
Eleven studies were included (see study references list).
Again, observed effects varied (absolute rs from 0.05 to 0.68) but were typically small (median of medians = 0.29). Sociodemographic and social network factors were sub-divided into five separate categories: (1) Social contact; (2)
Observed effects were again wide-ranging (rs from 0.01 to 0.59) though typically small (median of medians = 0.12). The relational fac-

| Psychological variables
Three studies [S8-S10] identified psychological variables associated Findings in study (S10) were mirrored in study (S8), which found a strong negative relationship between active re-engagement coping and patient depression (−0.60) alongside smaller negative associations between active coping and patient and partner anxiety (−0.29 and −0.33 respectively). (across six studies, i.e., a sampling frame of 50%) and, prior to resolving any differences, overall weighted kappa = 0.965 ('almost perfect' agreement). questions. However, one paper (4) employed a self-designed, nonvalidated questionnaire to measure intimacy changes. Only three studies were considered to satisfy the criterion requiring the risk of non-response bias to be low (S2-S3, S9). Study 1 displayed a potential non-response bias, as differences were found between respondents and non-respondents: non-respondents were found to have worse performance status in relation to their cancer treatment and were found to have higher levels of spousal distress at the recruitment stage. In seven of the included studies (S4-S8, S10-S11), it was difficult to judge non-response bias, as these studies did not provide enough information to assess whether those who participated were different from those who did not in terms of the variables of interest.

| Quality appraisal and risk of bias within studies
All studies were judged to have used appropriate statistical analyses. Additionally, all studies provided complete outcome data; although there is no widely agreed cut-off value for what is considered an acceptable level of outcome data completion, this review adopted the conservative figure of 95% to categorise a study as providing complete outcome data. 46 Seven of the studies satisfied the requirement to have fully reported relational sample characteristics: this was an additional quality criterion due to the relational focus of the current review. Four studies did not meet the requirements for this criterion due to studies either only partially reporting characteristics or providing no information (S6-S8, S11).

| Key findings
The current review identified factors associated with psychological distress for couples facing HNC relating to four variables: clinical, sociodemographic and social network, relational, and psychological.
The findings will be discussed in relation to each variable. Overall, the strongest magnitude of association was found for psychological variables (individual differences in coping, illness perceptions, selfefficacy, and goal disturbance).
Regarding clinical factors, several studies identified that the HNC disease burden predicted increased psychological distress for patients and partners. These findings are consistent with research identifying HNC as an especially traumatic cancer due to both the illness and the aggressive treatment options it necessitates. 10,23 There was considerable variation in the reported effect sizes which Note: This appraisal tool is an adaptation of the Mixed Methods Appraisal Tool (MMAT) 40 which can be found in Appendix S1. The scoring refers to whether the criterion is present within the study: 2-'Yes,' 1-'Cannot Tell,' 0-'No.' It was possible to achieve a maximum score of 18.
ranged from small to large and the overall median r was small.
However, two studies did not find an association between these variables for patients (S3, S6). Notably, some studies only reported an effect size in relation to the experienced disease burden for the patient rather than both dyad members. In terms of cancer (disease characteristics), the median r was negligible, indicating that disease characteristics of the cancer did not predict distress for couples as compellingly as disease burden. This aligns with literature from breast cancer populations, which found no effect of either the type of surgery or the type of adjuvant therapy on subsequent distress levels. 28 However, fewer studies in the review investigated the role of disease characteristics in influencing distress, as opposed to disease burden, so it is possible that with larger sample sizes, a stronger relationship may have been found.
Regarding sociodemographic and social network factors, several factors were identified. Two studies (S7, S10) identified large, significant relationships between supportive social contact and lower levels of psychological distress for couple members. These results are consistent with literature suggesting that social support produces less psychological distress for those experiencing HNC. 47,48 The quality of research on this topic would be further enhanced if future research consistently reported effect sizes for both patients and partners. In relation to gender, a negligible median r was found: this aligns with research from another systematic review which found that gender was a significant predictor of distress in only two out of 13 papers. 28 In terms of age, a small, negative r was found, indicating that a younger age was associated with greater distress: this finding reflects research which found that younger age was related to increased anxiety 18 months after an HNC diagnosis. 49 Regarding relational factors, there was a relationship between the perception of relationship quality and psychological distress for both patients and partners as identified by four studies (S1, S2, S4, S6). The reported effect sizes for this relationship displayed considerable variation, ranging from small to large. These findings are important for clinicians and researchers to consider, as research shows that the quality of partnered relationships is linked to psychological adaptation and cancer-related health outcomes. 50 These results might inform future research and clinical practice: when considering potential interventions for HNC dyads experiencing distress, there is potentially greater scope for intervening with more malleable relational factors as opposed to more static clinical and sociodemographic factors. In terms of communication and intra-dyad coping styles, small median rs were found: this was inconsistent with evidence highlighting the importance of intra-dyad coping styles in improving distress outcomes in couples with HNC. 45 Three studies identified psychological factors (S8-S10) which appeared to have the strongest relationships with distress (large median of study medians)-perhaps partially reflecting the interdependence of mood and cognition (individual appraisal and response-style), and common method variance (for factors assessed via subjective self-report measures). Findings from two independent studies (S8, S10) converged in suggesting that greater distress (for both patients and partners) is associated with passive (vs. engaged) ways of coping. In terms of illness perceptions/appraisals, a small median r was found: previous research has found limited support for the role of appraisal of illness and subsequent distress. 28 However, the previous evidence was based on a small number of studies and was based on the patient's perception of the illness: there is a paucity of research investigating the relationship between partner perceptions of the illness and distress outcomes.

| Study limitations
Systematic literature reviews can be associated with limitations such as heterogeneity, issues concerning study selection, and incomplete or inaccurate analyses and outcomes. 51

| Clinical implications
The reviewed studies indicate factors that could influence clinical practice, and potentially reduce psychological distress levels for couples experiencing HNC. These areas include: