Abstract
Identifying risk factors for psychological distress in patients with cancer may help providers more efficiently screen, identify, and manage distress. This article presents predictors of psychological distress in a large heterogeneous sample of cancer patients. In total, 836 patients were enrolled in a large randomized control trial and completed computerized psychosocial assessments Mental Health Assessment and Dynamic Referral for Oncology. Multivariate regressions examined predictors of distress and interest in mental health services. Final models suggest that psychological distress was related to six variables, and interest in mental health services was related to previous history of mental health counseling, total number of cancer-related symptoms, and race/ethnicity. Results may be used to identify high-risk patients who may benefit from proactive psychosocial interventions.
Introduction
At least one-third of oncology patients experience psychological distress (Carlson and Bultz, 2003). Distress has been defined by the National Comprehensive Cancer Network (NCCN) (2012) as “a multifactorial unpleasant emotional experience of a psychological (cognitive, behavioral, emotional), social, and/or spiritual nature that may interfere with the ability to cope effectively with cancer, its physical symptoms, and its treatment.” A better understanding of psychological distress in patients with cancer is imperative as untreated psychological distress may result in significant negative outcomes (Sarna et al., 2002), which could be short-term (e.g. non-adherence) or longer term (e.g. clinical depression). Distress affects quality of life (QOL) because of the obvious personal suffering it can cause, but also because it can adversely impact social functioning and relationships (Deimling et al., 2006). Some researchers have even hypothesized that high rates of psychological distress may have the potential to affect survival rates in oncology patients (Brown et al., 2003; Jacobsen, 2007). Individuals may experience psychological distress during any phase of cancer treatment, from the moment of diagnosis to active treatment and even into survivorship. Henselmans et al. (2010) examined distress in patients both at the time of diagnosis and during active treatment in 171 women newly diagnosed with breast cancer and found that 33 percent of patients were highly distressed immediately after diagnosis and throughout active treatment. Furthermore, though distress in cancer survivors often stems from different concerns than those who are in active treatment, it is estimated that at least 22 percent of cancer survivors continue to experience psychological distress after treatment is complete (Sarna et al., 2002).
Although the impact of psychological distress has been well documented, there are inconsistencies in the literature regarding rates of psychological distress in oncology populations. For example, while Lynch et al. (2008) found that 47 percent of their sample was considered psychologically distressed, Skarstein et al. (2000) only found that 17 percent of their sample experienced such problems. Such discrepancies may be due to a number of factors, including the demographic and clinical characteristics of samples, such as age, race, gender, and type of cancer. Zabora et al. (2001) examined cancer sites and psychological distress by analyzing a database of 4469 patients with different types of cancer who had completed the Brief Symptom Inventory (BSI). The researchers found that patients who were younger, not married, and had lung cancer had the highest levels of distress. A number of other studies have found similar results. Parker et al. (2003) assessed the relationship between depression and anxiety and age, marital status, level of education, social support, and QOL in 351 outpatients with cancer. Older, married patients reported fewer depressive symptoms, and greater perceived social support was correlated with fewer depressive symptoms, decreased anxiety, and better QOL. Furthermore, Deimling et al. (2006) conducted a study looking at race and psychological functioning in a random sample of 321 cancer survivors with either breast, prostate, or colorectal cancer from the cancer tumor registry of the Ireland Cancer Center (ICC). Using a relatively diverse sample (38% Black/African American, 62% White/Non-Hispanic), the authors found that Black/African American patients reported significantly less depression and anxiety symptoms than their White counterparts. Finally, Thomas et al. (2011) examined the relationship of gender and psychological distress in 760 patients (357 males and 366 females) who had a range of cancer diagnoses and found that female patients were at higher risk of overall distress compared to men. Regardless of the population providers are working with, screening and addressing psychological distress is now considered to be part of standard practice in oncology settings. However, very few providers are able to keep up with consistent psychosocial distress screening due to overall time shortages as well as an absence of mental health professionals in oncology settings.
The purpose of this study was to evaluate demographic, physical, and psychosocial predictors of psychological distress in a large sample of cancer patients with a range of disease sites. A secondary purpose of this article was to investigate both physical and psychosocial predictors of interest in mental health counseling services offered to patients during their oncology visits.
Method
Participants
The sample used in these analyses were collected as part of a large randomized control trial (RCT) that examined whether or not a computerized psychosocial assessment and intervention program for patients with cancer could improve psychological, social, and physical outcomes. The larger RCT results are forthcoming in a future publication as we are still collecting follow-up data. These analyses were performed on the baseline data collected from 836 cancer patients recruited from three comprehensive cancer centers: the University of Massachusetts Medical School Cancer Center (n = 581; 70%), the Cancer Institute at Cooper Hospital in New Jersey (n = 126; 15%), and the University of Texas MD Anderson Cancer Center (n = 129; 15%). Eligibility required (1) a past or current cancer diagnosis, (2) age greater than or equal to 18 years, (3) a permanent residence in the same state as treatment center, and (4) a working telephone number. The following excluded enrollment: altered mental status (e.g. psychosis, delirium, disorientation, dementia, unresponsiveness), hostile or agitated behavior, severe illness that would preclude conversation or computer use (e.g. persistent vomiting, severe pain), or factors precluding follow-up (e.g. transient residence or lack of a telephone). Patients were recruited regardless of type, duration of illness, stage of cancer, or phase of treatment. The study was approved by the Institutional Review Boards of all three institutions.
Measures
The baseline Mental Health Assessment and Dynamic Referral for Oncology (MHADRO) assessment consisted of a maximum of 80 items. Branching logic ensured that participants only answered applicable items. Demographic and disease information including gender, race, ethnicity, age, marital status, type of cancer, and time since cancer diagnosis was collected on all participants. The psychosocial constructs that were measured in the MHADRO’s baseline assessment included depression, anxiety, functional disability, subjective well-being, Behavioral Health Status (BHS) index (a mixture of anxiety, depression, functional disability, and subjective well-being scores), NCCN distress thermometer, social support, alcohol abuse, tobacco abuse, patient–provider relationship, and cancer symptoms and side effects (pain, tiredness or fatigue, nausea or vomiting, insomnia or sleep difficulties, difficulty with bowel movements, and sexual difficulties or lack of interest in sex). For all of these scales, higher scores indicated more positive outcomes with the exception of the alcohol use, tobacco use, total number of symptom items, and the Self-Reported Distress Scale (NCCN-DT). A full description of the assessment’s major constructs of this study has been published (O’Hea et al., 2013). Finally, the MHADRO program also contained questions about mental health diagnostic and treatment history, which queried any existing mental health problem at present or in history (e.g. diagnoses, hospitalization, psychotropic prescriptions, counseling). Participants also indicated whether they believed seeing a counselor would be beneficial (yes, no, not sure).
Procedures
Recruitment/enrollment
All participants signed a written consent prior to participating in the study. All subjects were treated in accordance with the National Institutes of Health (NIH)’s Guidelines for Protecting Human Subjects. Participants arriving for routine oncology (treatment or follow-up) appointments or chemotherapy infusions were screened for eligibility and approached in an exam room or an infusion chair. Each patient was given information about the study’s purpose and participant requirements (one computerized assessment and three follow-ups). Participants were randomly assigned to one of the two conditions described below: control or intervention. For the control group, participants completed the MHADRO baseline assessment and follow-up assessment; but did not receive the three feedback reports or the option for a dynamic referral. At the end of the 12-month follow-up assessment, all control group participants were given the opportunity to take advantage of the dynamic referral option.
Participants in the intervention group received tailored feedback reports including a Health Care Provider Report and Patient Feedback Report, and if they reached a certain criteria, they were also offered a Mental Health Referral Report upon completion of the baseline assessment. Patients in the intervention group who were not already in therapy received the option for having a faxed referral sent if any of the following applied: (1) a score of 3.2 or less out of a possible 4 on the MHADRO’s depression scale (with higher scores indicating better outcomes), (2) a BHS score falling in the most severe range (scores in the top 70%), (3) an answer of “yes” or “not sure” when asked whether seeing a counselor or therapist would be beneficial, (4) a rating of 9 or 10 on the self-reported measure of emotional distress, or (5) sexual difficulties endorsed. Participants who accepted a faxed referral signed an authorization form to release Personal Health Information. Upon completion of the assessment, a referral request was automatically sent to an appropriate mental health provider, based on the participant’s zip code and insurance. A copy of the mental health provider’s information was also provided in the patient report, which was given to all patients in the intervention condition as they left their appointment. When requested, the mental health provider was given 72 hours to contact the patient who requested the referrals. A research assistant followed up with the mental health provider to confirm receipt of the request and also called the patient to ensure whether he or she was contacted by the mental health provider. Control group participants were given the opportunity to have a dynamic referral sent after the completion of their final, 12-month follow-up assessment.
Statistical analysis
Basic descriptive statistics were used to characterize the sample. To assess predictors associated with psychological distress we used a manual, backward model selection process using multiple linear regression. As a first step, all variables under consideration were included (race/ethnicity, sex, marital status, education, alcohol use, cancer type, time since diagnosis, medication for emotional problems, counselor visits, mental health diagnosis, age, Heavy Smoking Index, social support, patient–provider partnership, number of cancer symptoms, rating of pain in last 2 weeks, rating of fatigue in the last 2 weeks, and rating of concern about sexual difficulties). At each subsequent step, the model fit and statistical significance was assessed and a decision made as to the next variable to be removed from the model (based on the largest p value). This procedure was repeated until a model with predictors significant at the p < 0.05 level was obtained. For categorical variables that remained significant, we calculated least squares means with an adjustment for multiple comparison testing (see Tables 1 and 2).
Characteristics of the sample in the MHADRO study (N = 836).
MHADRO: Mental Health Assessment and Dynamic Referral for Oncology; BHS: Behavioral Health Status; SD: standard deviation.
Final multivariable model of predictors of psychological distress.
All group differences are significant at least at the p < 0.05 level.
As a final step, we examined factors associated with the acceptance of dynamic referral (see Tables 3 and 4). Our initial analyses included descriptive statistics and proceeded to multivariable modeling. Similar to our approach of determining the predictors of psychological distress, we employed a manual, backward selection process using multiple logistic regression. Using the same regressors from the model predicting psychological distress, we removed variables from the model until only predictors significant at the p < 0.05 level was obtained. Resulting odds ratios (ORs) and 95 percent confidence intervals (CIs) are reported. All statistical analyses were performed using SAS 9.3 (SAS Institute Inc., Cary, NC, USA).
Factors associated with acceptance of dynamic referral.
Final multivariable model of acceptance of dynamic referral: the MHADRO study.
MHADRO: Mental Health Assessment and Dynamic Referral for Oncology; CI: confidence interval.
Results
The patient sample of 836 participants was primarily non-Hispanic White (90.7%), female (86%), and married (65%) with a mean age at intake of 59 years (standard deviation (SD) = 11.6 years). A large percentage of participants (56%) were alcohol users but did not screen positive for risky use at the time of assessment. The most prevalent cancer type in the sample was breast cancer (49.2%), with the majority of participants having been diagnosed more than 6 months ago (66%). Sample participants, in general, had few mental health challenges, with 69 percent indicating they have never taken medication for emotional problems, 63 percent never having seen a counselor or therapist to manage emotions or stress, and 65 percent reporting no mental health diagnosis (Table 1).
Multiple linear regression modeling revealed that participants who had never taken medication for emotional problems and those who had but currently were not taking medications had lower psychological distress scores on the BHS than those who were currently taking medications, respectively (Table 2). Additionally, ever having seen a therapist for emotional problems and reported mental health diagnoses were associated with higher psychological distress. Heavier smokers also had significantly higher psychological distress than non-smokers, as did participants with more pain and fatigue in the past 2 weeks (Table 2).
Of the 836 participants enrolled in the study, 415 were in the intervention group and therefore eligible to receive dynamic referral. The offering of dynamic referral among those eligible did not differ by sociodemographic characteristics (race/ethnicity, sex, marital status, and education), behaviors (alcohol use, Heavy Smoking Index), time since diagnosis, social support, or patient–provider partnership. Participants offered dynamic referral tended to be younger, had higher distress as measured by the BHS, had a higher number of cancer symptoms or side effects, higher ratings of pain and fatigue, and more concern about sexual difficulties or lack of interest. Those offered dynamic referral also were more likely to have reported a mental health diagnosis, had seen a counselor previously for help managing emotions or stress, and either previously or currently were on medication for emotional problems.
Of the 185 participants offered dynamic referral, because they met the aforementioned criteria, two changed their minds about being contacted and were removed from subsequent analysis. Of the remaining 183 participants offered the dynamic referral, 27.3 percent (N = 50) accepted. Several factors were associated with the acceptance of a dynamic referral including higher psychological distress, race/ethnicity, the number of cancer symptom side effects, and pain and fatigue ratings (Table 3). However, only three variables were retained in the final model. Specifically, those who had never seen a counselor or therapist for emotional help were less likely to accept a dynamic referral than those had seen one before but not currently (OR = 0.48, CI = 0.23–0.99). (Note that none of the participants in this subsample indicated that they were currently seeing a counselor (Table 4)). Likewise, individuals from a non-Hispanic Black background or a Hispanic/Latino background were 5.53 and 3.98 times more likely to accept a dynamic referral than non-Hispanic Whites. Finally, each additional cancer symptom or side effect increased the odds of accepting a dynamic referral by 35 percent.
Discussion
The variables related to psychological functioning have established histories of being linked to distress, either in individuals with cancer or the general population. First, many researchers have demonstrated that younger oncology patients tend to experience more psychological distress than older patients (Hack et al., 2010). One recent study included 1057 newly diagnosed patients with either breast or prostate cancer and demonstrated that age was a significant predictor of distress as younger women were more distressed than older women with cancer (Hack et al., 2010). We also found younger age to be significantly related to psychosocial distress in our large, heterogeneous sample. These findings may be attributed to the responsibilities and dependents a young person has compared to older adults. For example, a young person may have young children or even still be paying off college loans. Furthermore, being young when facing cancer has the additional burden of both feeling like an anomaly and facing mortality much sooner than some others. In addition, individuals diagnosed with cancer at a younger age may also have to live longer with the disease and its physical as well as psychosocial repercussions.
History of psychological problems, as indicated by previous diagnoses, past use of psychotropic medications, or history of counseling/therapy, was also related to psychological distress in this. This link has been established in the past (Jim et al., 2012; Senf et al., 2010) and is important because it has the potential to impact clinical practice related to cancer care for high-risk patients. That is, at the present time, although psychiatric history may be documented in cancer care, many practices do not use this information to systematically screen, monitor, and address psychological distress in this potentially high-risk group. It may be beneficial to use previous psychiatric history to systematically, and therefore more consistently, place such patients in a “higher risk” group that can be monitored for psychological distress more proactively and closely than the average patient. This proactive approach would be different than the more reactive approach where patients are referred to mental health specialists after the patient has begun to show significantly psychosocial distress symptoms. Having a mental health specialist on board from diagnosis may prevent higher symptom severity.
Two other common predictors of distress were significant in our model: smoking and reported low levels of social support. These variables are meaningfully related to distress and have established histories of being linked with negative outcomes, such as psychological dysfunction, in the general population (Lawrence et al., 2009) as well as in oncology samples (Ditre et al., 2011; Humphris and Rogers, 2004). For example, Ditre et al. (2011) found higher pain ratings in patients with cancer who were smokers as compared to never-smokers. Also, Humphris and Rogers (2004) followed up patients with oral cancer over 15 months and noted significantly greater psychological distress in smokers versus never-smokers. In regard to social support, in a relatively recent study of 2595 women with breast cancer, Bardwell et al. (2006) found that in comparison to a number of other variables such as physical health, physical activity, age, symptoms, and pain, only low social support remained a significant predictor of distress. Similar to previous history of psychological problems, these personal variables, such as low support, could potentially be used in clinical oncology practices as a way to identify patients who have been recently diagnosed with cancer and who may be at high risk of psychological problems during their care. Furthermore, more integrative and intensive psychosocial care may be warranted for patients who are identified as having low levels of a history of psychological problems or low social support.
The last group of variables that were linked to high distress was physical problems. Specifically, high ratings of pain and fatigue were related to greater distress in our sample. Recent studies have established that interventions such as yoga (Wren et al., 2011), exercise (Cheville et al., 2013), and cognitive behavioral therapy/interventions (Tatrow and Montgomery, 2006) can help cancer patients with pain, fatigue, sleep problems, and general distress. Ideally, providers could screen for such problems, using a program like the MHADRO, to identify patients early and, because there are empirically supported interventions, provide treatment before symptoms become too severe or debilitating.
It is notable that type of cancer, time since diagnosis, marital status, and race did not impact distress levels, which is interesting as these variables have been linked to distress in cancer in the past (Bardwell et al., 2006; Deimling et al., 2006; Parker et al., 2003; Zabora et al., 2001). However, as was established in the beginning of this article, many times other studies have focused on only one type of cancer, which could impact findings.
In addition to predicting distress, we also examined variables that were linked with acceptance of mental health services, when offered to patients. Not surprisingly, those who had never seen a counselor or therapist for emotional help were less likely to accept a dynamic referral to see mental health counselor than those had seen one before but not currently. This may be related to familiarity with the process or past success and symptom relief from therapy. Since this was not queried, however, we can only make speculations at this time. Also, having more symptoms or side effects of cancer was also related to increased interest in seeking mental health services, which makes sense in that the more difficulties a person experiences, perhaps the more open he or she is to receiving help in managing such symptoms. We did not expect to find racial differences regarding interest in mental health services, but nonetheless, non-Hispanic Black and Hispanic patients were significantly more likely to be interested in mental health services than their non-Hispanic White counterparts. Because the present sample was not racially or ethnically diverse, this finding should be interpreted with caution and future research is needed to determine whether this is indeed a valid and reliable finding. Along a similar line of thought, although this study consisted of a very large sample of patients with various disease types who were attending three different cancer centers at various geographical locations, the sample was predominately women of a White/Caucasian background. Thus, all of these findings should be interpreted cautiously in terms of extrapolation to men and patients of diverse backgrounds. Future research, of a similar scale, with men and patients of diverse ethnic/racial backgrounds is needed to determine whether the same or different factors are predictive of psychological functioning.
There are a few notable limitations to this study. First, in future studies, it may help to assess patient attitudes toward computerized screening and assessment as this may have impacted responses and willingness to share information. Second, it may have been more helpful to have the oncology provider review the patient report before giving the patient the opportunity to send a dynamic referral for psychological services. The oncology provider’s review of the symptoms may have yielded more patients who were interested in reaching out for help from a mental health provider. Finally, although the purpose of our study was to use a computerized program to screen and refer for psychological distress, in-person evaluations and interactions are still preferred, when possible, as they provide a professional evaluation and management of symptoms that a computer cannot. However, for clinics or healthcare providers who do not have the time or resources for personalized screenings, programs like the MHADRO may be helpful.
In sum, these findings established that, in a large sample of patients with various types of cancers, certain psychosocial variables may increase the risk of psychological distress. The important variables identified in this study were the following: being younger at diagnosis, having a history of mental illness, high levels of tobacco addiction, experiencing low levels of social support, or reported high levels of pain and fatigue. In addition, patients who had a history of therapy, who were experiencing high levels of physical symptoms and side effects, or who were from a Latino/Hispanic or Black/African American background, were more interested in counseling/therapy than patients with other characteristics. These results may be used to inform screening and intervention practices as they provide some insight into attributes that may place oncology patients at higher risk of developing psychological problems during their oncology care. Finally, intervention studies are needed that use these high-risk identifiers as a way to provide early assessment and intervention for patients with cancer. Such studies would be able to help us determine whether early identification and management can indeed improve longer term psychosocial outcomes for people with cancer.
Footnotes
Acknowledgements
Pre-trial registration number: NCT01442285.
Declaration of conflicting interests
This grant was funded through a Small Business Technology Transfer (STTR) mechanism from the NIH/NIMH. Intellectual property, licencing and revenue income from the Mental Health Assessment and Dynamic Referral for Oncology (MHADRO) program are shared between Polaris Health Directions and the University of Massachusetts Medical School. Dr Boudreaux and Dr O’Hea also consult for Polaris Health Directions. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research received grant funding from the National Institutes of Health/National Institute of Mental Health (NIH/NIMH) (grant no. R42 MH078732).