Abstract
For most people, the diagnosis of cancer represents a major and catastrophic life event. Patients must deal with stress associated with having a life-threatening illness and the aversive side-effects of treatment. Many patients also experience changes to their work status, social relationships, physical abilities and family roles [1–3]. A substantial proportion of patients diagnosed with cancer experience psychological morbidity following diagnosis and surgery, with the predominant disorders being anxiety and depression [4–6]. Apart from their effect on patients' quality of life, anxiety and depression may also increase the severity of physical symptoms experienced during chemotherapy [7], and in some cases may affect adherence to treatment [8].
Estimates of the proportion of patients who are experiencing psychological morbidity vary from 23% [9] to 47% [4]. Patients identified as particularly vulnerable to high levels of distress include inpatients, those with more advanced disease, severe pain or functional impairment and patients with a history of affective disorder [8,10–12]. While the extent of psychological morbidity among cancer patients has been widely reported in overseas research, to date few studies have examined its prevalence among Australian patients.
One recent Australian study reported on psychological morbidity among women with newly diagnosed, primary-stage breast cancer [13]. The prevalence of psychiatric disorder, as diagnosed through a structured clinical interview was 42%, with the majority of identified cases being anxiety and depressive disorders. This high prevalence of psychological morbidity was surprising, given that patients had reasonably good prognosis and, presumably, low levels of pain and disability. It is possible that the findings were affected by selection bias, as all patients in the sample had agreed to participate in a randomised trial which would involve their attending some form of group psychological intervention. The fact that a substantial proportion (38%) of patients who were originally approached to participate in the study declined also increased the likelihood of bias.
In the last two decades, as increased attention has been focused on the psychological aspects of cancer diagnosis, there has been a growth in the number of support services being made available to cancer patients. These include professionally led and self-help patient support groups, information and counselling services offered by the New South Wales Cancer Council and hospital-based counselling provided by social workers, psychologists and psychiatrists. Patients also have the option of accessing psychological services through private practitioners. To date, there is little information regarding the proportion of patients who access these services, or the degree to which patients perceive them to be helpful. Given the limited resources available for the psychosocial care of cancer patients, it would be useful to evaluate whether patients who are most in need of psychological support are using and benefiting from available support services. This information could be instructive for the future planning of psychosocial care for cancer patients.
We report on the results of a patient survey, which was conducted at four hospitals within the Sydney metropolitan region. A primary aim of the study was to measure the prevalence of clinically significant levels of anxiety and depression within a cross-section of cancer patients who were attending outpatient clinics. A further aim was to assess the degree to which patients use available support services (support groups, counsellors, psychiatrists etc.) and their level of satisfaction with these.
Method
Sample
Adult cancer patients (513) who attended oncology outpatient clinics at four hospitals in the Sydney metropolitan area (Royal North Shore Hospital, Sydney Adventist Hospital, Manly Hospital and St Vincents Hospital) participated in the survey. The eligibility criteria were: a diagnosis of cancer, at least 18 years of age, ability to speak and read English and ability to give informed consent.
Procedure
Ethics committee approval for the study was provided by each of the four hospitals. Cancer patients attending the hospital outpatient centres were approached by a research assistant and were informed of the purpose of the survey, both verbally and in writing. Patients who agreed to participate were asked to sign a consent form, and then complete a background information questionnaire and the Hospital Anxiety and Depression Scale (HADS) [14].
The HADS is a brief self-report screening instrument which has been designed to identify cases of anxiety and depression among medically ill patients. The majority of studies which evaluated the HADS in oncology settings have found it to have adequate psychometric properties [10,14–19], although one study found its sensitivity to be low [20]. The fact that it is brief, and easy to complete and score makes it particularly suitable for busy oncology settings. In accordance with the criteria suggested by its authors [14], patients who scored below 8 on either subscale were defined as ‘non-cases’, those who scored between 8 and 10 were defined as borderline cases, and those who scored above 10 were classified as experiencing clinically significant anxiety or depression.
The background questionnaire sought information on patients' demographic and clinical characteristics, as well as on their level of social support, use of support services and degree of satisfaction with the services used. Patients were asked to indicate their level of activity using a three-point scale: fully active, activity is somewhat/moderately restricted or more than 50% of daytime spent in bed. Patients were also asked to indicate their degree of satisfaction with their available social support on a five-point Likert scale (1, completely satisfactory; 5, totally inadequate).
As much of the data was of a nominal or ordinal quality, its analysis consisted of frequencies and Pearson χ-squared cross tabulations using SPSS (version 7.5.1, SPSS, Chicago, IL). Multivariate statistical techniques (i.e. logistic regression) were used to examine the differences within and between groups of patients. Statistical tests on demographic and clinical outcomes (on measures other than adjuvant treatment) were evaluated with a critical p of 0.05. Due to the number of contrasts used in measuring the effects of adjuvant treatment (chemotherapy and radiotherapy), a critical p of 0.0125 was used for this item.
Results
Of the 513 patients who responded, 504 valid surveys were obtained. Patients ranged in age from 20 to 93 years, with median age being 62 years. Females comprised 55% of the sample. The vast majority of patients (94%) nominated English as the primary language spoken at home. Secondary schooling had been completed by 89% of patients, of whom 47% had completed post-secondary qualifications (a degree, certificate or diploma). A total of 64% of patients were married or living in a de facto relationship. The most common cancer sites were breast and gastrointestinal (33% and 29% respectively), with head and neck, lung and haematological cancers comprising the majority of other cancers. Forty-two per cent of patients in the sample had either stage I or stage II cancers, and 43% had stage IV cancer. Forty-one per cent of patients were receiving radiotherapy and 31% were receiving chemotherapy at the time of responding to the survey. Forty-three per cent of the sample stated that they remained fully active, while 52% reported their activity had been restricted somewhat, and 4% reported spending over half their waking time in bed.
Table 1 displays the prevalence of anxiety and depression among the sampled patients, as measured by the HADS. Using a cut-off score of 11 on the HADS, the prevalence of clinically significant anxiety was 11.5%, and a further 11.7% of patients showed borderline levels of anxiety. The prevalence of clinically significant depression was 7.1%, while a further 11% experienced borderline levels of depression. Overall, 14.5% of patients were experiencing clinically significant anxiety and/or depression. When borderline cases were included, this figure increased to 30.6%.
Prevalence of anxiety or depression in patient sample (n = 504)
The relationship between demographic and/or medical factors and levels of anxiety and/or depression, based on Pearson χ-squared tests is displayed in Table 2. A significantly higher proportion of women than men experienced borderline and clinically significant anxiety (χ2 = 4.03, df = 1, p = 0.045), however, there were no significant gender differences on depression (χ2 = 0.006, df = 1, p = ns). Clinically significant anxiety was more prevalent among patients under 65 years of age than those who were 65 years or over (χ2 = 4.49, df = 1, p = 0.034). Age was not related to clinically significant depression (χ2 = 0.03, df = 1, p = ns). Patients who spoke a language other than English at home were more likely to experience clinically significant depression than other patients (χ2 = 3.87, df = 1, p = 0.049). They also showed a trend towards increased prevalence of clinically significant anxiety (χ2 = 3.80, df = 1, p = 0.051).
Relationship between demographic/medical factors and clinically significant anxiety and depression, using Pearson χ-squared tests
Patients with moderately or severely restricted levels of activity experienced more frequent clinically significant anxiety (χ2 = 7.89, df = 1, p = 0.005) and depression (χ2 = 15.34, df = 1, p < 0.001) than patients who were fully active. Patients with advanced (stage IV) cancer experienced higher levels of depression (χ2 = 6.05, df = 1, p = 0.014) than patients with more early stage cancer. Stage of disease was not related to anxiety levels (χ2 = 0.61, df = 1, p = NS). Patients who were diagnosed in the last 3 months did not show greater anxiety (χ2 = 0.68, df = 1, p = NS) or depression (χ2 = 0.01, df = 1, p = NS) than patients with an earlier diagnosis.
These relationships were also tested via logistic regression. The key risk factors associated with clinically significant anxiety using this method were: restricted activity level (p = 0.002) and being female (p = 0.021). Risk factors associated with clinically significant depression were: restricted level of activity (p = 0.001) and a non-English-speaking background (p = 0.030).
Patients responses on measures relating to social support suggested that majority of patients had adequate social support. On the five-point Likert scale, 79% of patients indicated that their available social support was ‘completely satisfactory’, while a further 14% indicated that it was ‘reasonably satisfactory’. Only 6% of patients reported having less than satisfactory social support. Sixteen per cent of surveyed patients had previously attended a cancer support group. Of these patients, 83% stated that they had found it ‘extremely’ or ‘reasonably’ helpful, while 4% stated attendance made them feel worse. A small proportion of patients (6.5%) were currently attending a support group. Of the sample, 17% had received individual counselling in relation to their cancer, of whom 64% were female. The majority of counselling was provided by patients' general practitioners (29%), social workers (23%) or ‘counsellors’ (23%); 15% of patients saw a psychologist, and 10% saw a psychiatrist. Of patients who received individual counselling, 80% did not personally pay for the service. For the 20% of patients who paid for their counselling, fees ranged from $4 to $210 per session, with the median fee being $75. Of the patients who had received individual counselling, 86% stated that they had found the experience ‘extremely’ or ‘reasonably’ helpful, while 4% stated it had made them feel worse. Twenty-five per cent of patients who were experiencing clinically significant levels of anxiety or depression at the time of the survey had received some counselling in relation to their cancer.
The relationship between patients' social support and/or use of support services and clinically significant anxiety and depression was measured using Pearson χ-squared tests. There was no relationship between patients' reported level of social support and their experience of anxiety (χ2 = 0.69, df = 1, p = NS) or depression (χ2 = 2.47, df = 1, p = NS). Similarly, patients who were attending a support group did not show higher levels of anxiety (χ2 = 1.45, df = 1, p = NS) or depression (χ2 = 0.23, df = 1, p = NS) than those who were not. In fact, the vast majority (82%) of patients who were attending support groups did not show clinically significant psychological symptoms. On the other hand, patients who had received individual counselling in relation to their cancer showed higher levels of anxiety (χ2 = 3.59, df = 1, p = 0.05) compared with those who did not. They did not differ on depression levels (χ2 = < 0.01, df = 1, p = NS). While patients with clinically significant anxiety and depression more frequently stated that they wanted to attend a support group, only 8.3% were currently doing so.
Discussion
The prevalence of clinically significant anxiety and depression among patients in the sample was 11.5% and 7.1% respectively. While the prevalence was reasonably low, it is possible that the finding was biased by the sociodemographic characteristics of the sample. Participants in the study comprised a cross section of patients attending four hospitals in the Sydney metropolitan region. The fact that the hospitals were in the more affluent parts of Sydney (North Shore and Eastern suburbs), the patients were predominantly English speaking and a high proportion of patients had completed tertiary education suggests that the sample was of high socioeconomic status. This factor may limit the generalisability of the findings of this study. It is possible that sampling in a wider geographical region would have resulted in higher prevalence of psychological distress. While social desirability effects (i.e. a desire to appear as well adjusted) might also have influenced some patients responses, this is a potential risk in any study of this nature, and is not likely to explain the relatively low incidence of patient distress identified in this study.
Although the prevalence of anxiety and depression were substantially lower than those reported in studies which assessed psychological morbidity using structured diagnostic interviews [4,13,15], they were not dissimilar to the rates reported by other studies which used the HADS. For instance, a North American study of hospital outpatients [10] found the prevalence of clinically significant anxiety and depression on the HADS to be 18% and 7.8%, respectively. A South African study which used the HADS to assess depression among oncology outpatients found an prevalence rate of 8% [21]. A British study which administered the HADS to women with advanced breast cancer [22] found a prevalence rate for clinically significant anxiety, depression or the coexistence of both at 9%, 9% and 9%, respectively.
It is possible that the relatively low level of psychological morbidity identified in this and previous studies which used the HADS reflects the limitations of the instrument. Hall et al. [20] found the HADS to have a low level of sensitivity (i.e. ability to correctly identify cases) when the recommended threshold of > 10 for clinical ‘caseness’ was used. While lowering the threshold to 8 increased the sensitivity of the instrument, its specificity decreased markedly (i.e. a higher number of non-cases were incorrectly identified as cases). However, other studies have found the HADS to be a sensitive and specific tool in screening for psychiatric disorders in oncology populations [15–17]. The HADS has also been found to correlate well with the Brief Symptom Inventory [23] and the General Health Questionnaire [24], which are more precise measures of psychiatric symptomatology [16,25].
A further criticism that has sometimes been directed at the HADS is the fact that the depression subscale focuses largely on anhedonia, rather than the full range of clinical symptoms associated with depression. It should, however, be stressed that because the instrument is designed for use with physically ill patients, the somatic symptoms of depression have been deliberately omitted. In addition, it should be noted that the HADS is a screening instrument, and is therefore not designed for the accurate diagnosis of depression or anxiety. The purpose of a screening instrument is to enable the assessment of a large number of patients, and to identify those who appear to be experiencing high levels of psychological difficulty. These patients may subsequently be targeted for further, more sensitive assessment.
Patients who were identified in this study as being at high risk of psychological morbidity were: females, patients who were less than 65 years of age and patients from a non-English-speaking background. More advanced disease and more restricted levels of movement were also associated with poorer psychological adaptation. Patients' level of social support was not associated with psychological adjustment. This may have been due to the low level of variability in scores (92% of patients indicated ‘completely satisfactory’ or ‘reasonably satisfactory’ levels of social support). This limited variability may have been a genuine phenomenon, or it may have reflected the limitations of the measure that was used to assess social support. As the five-point Likert scale was not a psychometrically validated instrument, it is not known whether it was sufficiently sensitive to provide accurate information on adequacy of social support. In addition, patients' responses may have been affected by social desirability effects, as some patients may prefer to be perceived as having good social support.
Of the patients who were attending support groups at the time of the study, the vast majority (82%) were not experiencing clinically significant psychological symptoms. It may be that patients who attend support groups do so for reasons other than psychological morbidity (e.g. companionship, information, relaxation). Alternatively, it is possible that the patients who were attending support groups had experienced psychological benefits from their attendance, and were coping well as a direct result of this factor.
Of the patients who were experiencing clinically significant anxiety or depression, the majority (75%) had not received any counselling or psychological treatment, and only 8.3% were attending support groups. This suggests that the majority of distressed patients were not accessing any kind of psychological support outside of their family and social network. The reasons for this are unclear, although it is possible that these patients were not aware of the psychological support services that are available (counselling and support groups), or did not see them as potentially beneficial. It may be that unless clinicians are prepared to refer patients or encourage them to attend, the majority of patients will not regard these services as relevant to their needs. Given the high degree of satisfaction expressed by patients who accessed individual counselling (86% found it ‘extremely’ or ‘reasonably’ helpful) or who attended support groups (83% found them to be ‘extremely’ or ‘reasonably’ helpful), it appears that potentially beneficial resources are being poorly used. Additional investigation of the reasons why patients do not to attend support groups or counselling could provide useful information regarding how the promotion of these services could be better targeted, or how they could more effectively serve patients' needs.
Although the prevalence of clinically significant anxiety and depression identified by the HADS in this study is relatively low, the number of affected patients is substantial. In a typical metropolitan hospital (such as the Royal North Shore Hospital) approximately 110 cancer patients attend for consultation or treatment each day. The findings of this study suggest that using the HADS, approximately 17 patients suffering from clinically significant anxiety or depression can be identified in the average day. If borderline cases were also included, this number would increase to 33. As there are currently no screening procedures in place, there is no way that psychologically distressed patients can be systematically identified and referred for treatment if necessary. Further, as very few patients are independently seeking counselling or taking advantage of existing support services, a substantial proportion of patients may be experiencing psychological distress that could be effectively treated. Indeed, there is considerable evidence that psychological interventions with cancer patients (in both group and individual format) can achieve significant reductions in anxiety, depression and other negative symptoms (see reviews [26–28]).
The experience of anxiety and depression has a negative impact on wellbeing and quality of life. High levels of distress also affect the course of treatment. Patients with psychological morbidity experience more severe physical symptoms associated with treatment [7], poorer treatment compliance [8] and greater use of health-care resources (see review [29]). Two prospective studies also show that patients suffering from depression or hopelessness have shorter survival duration relative to better-adjusted patients [30,31]. The early identification and treatment has the potential to impact on both psychological and health outcomes.
Screening as standard for patients newly referred to oncology departments has been suggested [16,32]. The program would enable early diagnosis and referral of patients experiencing psychological difficulties. Although not as sensitive as more specific diagnostic instruments, the HADS is brief. It is therefore useful in settings where large numbers of patients are evaluated. If identified cases of anxiety and depression were followed up with additional testing and/or intervention, such a program would have the potential to improve patients' quality of life, and assist in treatment.
Ultimately, the viability of screening will depend on a willingness to make resources available. This will require a recognition that psychological morbidity associated with cancer is a serious problem for a significant minority. With limited funds implementation of such a program remains a challenge.
Footnotes
Acknowledgements
We thank the Royal North Shore, Sydney Adventist, Manly and St Vincent's Hospitals.