SEARCH

SEARCH BY CITATION

Keywords:

  • cancer;
  • screening;
  • oncology;
  • distress;
  • depression;
  • acceptance of treatment;
  • desire for help

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Objectives: Despite documented high rates of psychological distress, it is not clear how to identify those who are willing to accept help. The aim of this study was to investigate whether asking patients receiving chemotherapy if they want help with emotional problems is valuable and to investigate the type of help they want.

Methods: Patients attending a chemotherapy suite were asked to complete the Hospital Anxiety and Depression Scale, the Brief Patient Health Questionnaire (PHQ) and the Emotion Thermometers tools. Results were compared with a single question on desire for help.

Results: In this study, 128 patients completed questionnaires for distress, depression, anxiety and desire for help at initial interview. Only one in five unselected patients had a perceived need for help, and in distressed patients only 36% expressed a desire for help. The addition of the help question to the two questions (PHQ-2) about mood and interest improved the ability to rule-in depression by increasing the specificity. However, by addition of this question, sensitivity was significantly reduced. Desire for help was modestly associated with severity of distress, anxiety and depression.

Conclusions: The addition of a help question appears to have limited value in screening for psychological symptoms, but it may highlight those who are willing to accept addition support. Clinicians should attempt to offer a range of psychosocial interventions that will be acceptable to patients with distress. Copyright © 2010 John Wiley & Sons, Ltd.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

It is well recognised that many cancer patients suffer appreciable rates of psychiatric morbidity. The prevalence of distress is approximately 48% in studies that used a criterion standard 1. Importantly, about two-thirds of cancer patients report at least one unmet psychosocial need 2, 3. However, even in cancer settings, not all patients want help with their problems and this may be especially so in relation to psychosocial difficulties. For example, Scholten et al. (2001) found that only 40% of 43 women with advanced breast cancer were willing to accept psychological counselling 4. In a large Canadian sample of 2776 patients, almost 50% of those who met criteria for distress had neither sought professional psychosocial support and nor intended to do so in the future 5. Klank-Rießen et al. (2007) found that 40.9% of 115 women with breast cancer expressed an interest in psychosocial support 6. Graves et al.7 found that among 333 lung cancer patients, only 22.5% wanted help with their problems and, even in those with a high rate of distress, only 30% wanted help. In a cross-sectional study of cancer patients in the Netherlands, Tuinman et al.8 screened 277 patients using the Distress Thermometer (DT) and the Hospital Anxiety and Depression Scale (HADS). Of those who scored above five on the DT, only 14% definitely wanted to be referred for further help, while 29% would consider a referral. Interestingly, of those with low distress, 5% still wanted to be referred and 13% would have considered a referral 8. Clearly, expression of help is linked with the acceptability of the treatment on offer. For example, Shimizu et al. (2005) found that only 28.2% of patients who screened positive on the DT and Impact Thermometer would accept a referral to a psychiatrist 9. Similarly, Curry et al. (2002) found that of those offered psychosocial referral only 22% accepted. Female patients and individuals with a moderate-to-high level of depression were more likely to accept such services 10.

Self-reported need for help is increasingly recognised as an important construct in mental health care. In the 2002 Canadian Community Health Survey, 22% of patients reported a 12-month unmet need for psychological care. Of these, 35% of the patients reported preferring to self-manage symptoms, 19% said that they did not get around to seeking care, 16% did not know how to obtain help, and 15% were afraid to ask for help 11, 12. Walters et al.13 found that 42% of primary care patients preferred to deal with distress on their own, while 47% indicated that they would like help. Those with mild-to-moderate distress preferred support from friends and family, relaxation, massage and support from their GP, whereas those with more severe distress were more likely to identify a preference for psychotherapy, medication and support groups 6. In the 2003 European Study of the Epidemiology of Mental Disorders (ESEMeD Project) involving 8796 adults, 9% of the total sample perceived some need for mental health care in the past 12 months. Among those who had a mental disorder in the past 12 months, 33% perceived a need for help. Van Beljouw et al. (2010) found that 22% of individuals with depression or anxiety in the NESDA study did not perceive a need for care 14. In those with no perceived need, 51% said they would rather manage alone and 36% said they didn't think help would be effective.

This interest in the help construct has led to consider whether asking patients whether or not they would like help for psychosocial problems could be a useful screening question in itself. One study in primary care suggested that the help question combined with the Patient Health Questionnaire (PHQ)2 is potentially useful 15, but as yet there is no consensus on how to measure this domain in cancer settings. Therefore, we aimed to clarify how many people with early cancer have a perceived need for help, whether a simple help question is a useful screening tool to identify depression in cancer settings and further what factors can predict those with a high perceived need for help.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Between March and August 2007, patients who had a diagnosis of cancer and were attending for their first chemotherapy treatment at Leicester Royal Infirmary were asked to participate in this study. The patients were given questionnaires to complete at their first chemotherapy treatment and were then asked to complete the same questionnaires again 2–4 weeks later when they attended for their follow-up treatment. This study looks at the results from the questionnaires completed at the initial appointment. Patients were excluded from the study if they were unable to consent, did not speak English or were too unwell to participate. The research was carried out with approval from the North Nottinghamshire Local Research Ethics Committee.

Several short or ultra-short mood questionnaires were used in this study. The Emotion Thermometers (ET) tool 16 includes and expands on the well-known DT 17 and includes the Help Thermometer (HT), as well as domains of anxiety, depression, anger all in a visual-analogue format. Preliminary cut-offs of 3v4 in each domain are suggested. In the HT, patients were asked to rate their desire for help with emotional problems on an 11-point Likert scale, where 0 indicated that they felt they could manage entirely on their own and 10 indicated that they felt they desperately needed help. The cut-off used for the DT as a measure of possible distress was a score of 4 or above 18. As a method of comparing the desire for help, we used a simple questionnaire, which asked patients the question: ‘Do you want help for emotional or psychological concerns at this stage?’ The patient was able to answer yes or no to this question by ticking the appropriate box. Patients who indicated that they wanted help were asked what type of help they wanted (the options were: information about their illness, medication, face-to-face contact, support group, self-help guides, complementary therapy, support for family and carers, and ‘other’) and who they would want to see for help. Those patients who indicated that they did not want help were asked to indicate the reason why they did not want help. In all cases, patients were able to select as many options as they felt were appropriate.

All the patients were also asked to complete the HADS, measuring depression and anxiety using the relevant subscales and the total score was used to measure levels of distress. The cut-off values for the HADS were set as follows: for the HADS total score (HADS-T), a cut-off of 15 or more was used to suggest distress; for HADS depression (HADS-D), a cut-off of 8 or more was used to identify depression; for HADS anxiety (HADS-A), a cut-off of 8 or more was used to identify anxiety 19. The patients also completed the PHQ-9, which was used as a measure of the severity of depressive symptoms. A cut-off of 10 or more was used on the PHQ-9 to indicate a possible depressive illness. The first two questions on the PHQ-9 were also examined as a separate two-item screening tool (PHQ-2). If the patient answered either of these questions positively, the test was considered positive. The results from this PHQ-2 were then compared with the results from the combination of the PHQ-2 and the help question, where a positive answer to the help question was that the patient requested help.

The data obtained from the questionnaires were analysed using the statistical package R, version 2.7.0, AMOS 5, which calculates path analysis weights adjusting for intercorrelations and Microsoft Excel 2007.

The difference in the prevalence of caseness on each of the ET for those wanting help and those not wanting help was calculated and the significance of these differences was tested using a Pearson's χ2 analysis with Yates' continuity correction. To establish diagnostic accuracy, the following statistics were used: sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and fraction correct. Receiver-operating characteristic (ROC) curves were used to calculate the optimum cut-off point on the HT.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Description of the sample

Of 173 individuals approached, 130 (75.1%) were willing and able to participate in this study. One hundred and twenty-eight of the 130 patients completed all of the questionnaires at initial evaluation. One patient did not complete the HADS and one patient did not complete the PHQ-9. Forty-six (35%) of the patients were men and 84 (65%) were women. The mean age was 58.5 years and the mean time from diagnosis was 10.8 months.

In this sample, 62 (48.1%) patients were anxious according to HADS-A, 24 (18.6%) were depressed on HADS-D, 37 (28.7%) were distressed using HADS-T, 29 (22.5%) rated as depressed on PHQ-9 and 27 (20.9%) scored positively on PHQ-2.

Description of perceived desire for help

Twenty-six of the 130 patients (20%) indicated that they would like help on the single-item help questionnaire. Of those thought to be distressed on the DT, 36.7% wanted help. Of those who indicated that they wanted help, nine (34.6%) scored 4 or above on all four-symptom-related ETs, fourteen (53.8%) scored 15 or above on HADS-T, and nine (34.6%) scored 10 or above on the PHQ-9. The mean rating on the ‘desire for help’ thermometer was 4.4 for this group of patients.

Of those saying that they did not want help, eleven (10.6%) scored 4 or above on four-symptom-related ETs. Twenty-three patients (22.1%) scored 15 or above on HADS-T and 20 (19.2%) scored 10 or above on the PHQ-9. The mean rating on the ‘desire for help’ thermometer for this group was 1.3. There were two patients who wanted help but scored 3 or below on all 5 ETs, 9 or below on PHQ-9 and 14 or below on HADS-T. A comparison of scores on the emotion thermometers for those wanting help and those not wanting help is shown in Figure 1. All of the differences between the thermometer ratings between those wanting help and those not wanting help were observed to be significant at p<0.05 using Pearson's χ2 test.

thumbnail image

Figure 1. Prevalence of caseness on ETs in those wanting and not wanting help. Figure shows proportion of people answering positively to the help question who scored above 3 on each thermometer and proportion of people answering negatively to the help question who scored above 3 on each thermometer

Download figure to PowerPoint

The most common request for help was for face-to-face contact, with 16 people (61.5%) indicating that they would like this type of assistance. Looking at other types of help that people wanted revealed that 57.7% of patients requested complementary therapy, 42.3% wanted support for family and carers, 34.6% wanted information about their illness, 30.8% wanted self-help guides and 30.8% wanted support groups. The least popular option was medication (19.2%). The preferred sources of help were: nurse specialists (53.8%), family and friends (46.2%), hospital doctors (26.9%), GP (23.1%), psychologist (19.2%), spiritual advisor (7.7%) and a psychiatrist (3.8%).

In this study, 73 of 81 (90.1%) of those not distressed did perceive a desire for help. The most common reasons were: receiving support from elsewhere (56.7%), feeling well (41.3%) and coping on their own (23.1%). Less common explanations were being afraid of side effects (7.7%), low confidence in treatment (4.8%), unwillingness to talk with a professional (7.7%) and fear of stigma (2%).

Is the help question a useful screening tool?
(a) The help question used alone

The help question data were compared with the HADS and PHQ9 rating scales at conventional cut-offs. These results are illustrated in Table 1. The data show that the help question used alone was moderately successful at confirming the presence of anxiety (PPV=0.808) but could not be used as a case-finding method for distress or depression. However, it was able to rule-out depression, identifying the non-depressed (NPV=0.806 or higher). Rule-out accuracy for anxiety was the least successful. The values for sensitivity were low when the help question was compared with the PHQ-9, HADS-T, HADS-A and HADS-D.

Table 1. Help question alone in identification of distress, anxiety or depression
 SensSpecPPVNPVLR+LRORFraction correct
  1. Sens, sensitivity; Spec, specificity; PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio, LR, negative likelihood ratio; OR, odds ratio; Fraction correct, overall accuracy

Distress by HADS-T (>14)0.3680.8680.5380.7672.7880.7283.8300.721
Depression by HADS-D (>7)0.4400.8560.4230.8643.0560.6544.6730.775
Anxiety by HADS-A (>7)0.3390.9250.8080.6024.5200.7156.3220.643
Depression by PHQ-9 score (>9)0.3100.8300.3460.8061.8240.8312.1950.713
(b) The help question in combination with PHQ-2

The addition of the help question to the PHQ-2 was investigated, compared with either method used independently (Table 2) for identification of distress (HADS-T) and depression (PHQ-9). The results show that the addition of the help question to the PHQ-2 increased specificity but reduced sensitivity. The addition of the help question to the PHQ-2 increased specificity but reduced sensitivity and based on overall accuracy (fraction correction), the addition of the help question did not help in the identification of PHQ defined depression or HADS-T defined distress.

Table 2. Help question combination in identification of distress or depression
 SensSpecPPVNPVLR+LRORFraction correct
  1. Sens, sensitivity; Spec, specificity; PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio; LR, negative likelihood ratio; OR, odds ratio; Fraction correct, overall accuracy.

Comparison against distress by HADS-T
PHQ-2 alone0.5530.9340.7780.8338.3790.47917.4930.822
PHQ-2 + Help0.2370.9780.8180.75410.7730.78013.8120.760
Help alone0.3680.8680.5380.7672.7880.7283.8300.721
Comparison against depression by PHQ-9
PHQ-2 alone0.6900.9300.7410.9129.8570.33329.6010.876
PHQ-2 + Help0.2760.9700.7270.8229.2000.74612.3320.814
Help alone0.3100.6300.3460.8061.8240.8312.1950.713
What predicts perceived desire for help?
(a) Use of the HT as a predictor of desire for help

We examined the accuracy of the HT as a predictor of desire for help based on the single desire for help question. Accuracy at each cut-off point is shown in Table 3. Using ROC curve analysis, the cut-off on the HT, which maximised sensitivity and specificity, was a score of 3 or above when compared with the desire for help question. At this threshold, the HT had a sensitivity of 1.00 and a specificity of 0.66 and a PPV of 0.43 and a NPV of 1.00. At a cut-off of 3 or above, the HT correctly identified 80% of responses to the desire for help question.

Table 3. Accuracy of the HT against perceived desire for help
Cut off (greater than)SensitivitySpecificityPPVNPVFraction correct
  1. PPV, positive predictive value; NPV, negative predictive value; Fraction correct, overall accuracy.

01.000.000.20N/A0.64
11.000.550.361.000.72
21.000.650.421.000.80
30.810.800.500.940.84
40.620.890.590.900.82
50.420.920.580.860.82
60.190.970.630.830.81
70.150.970.570.820.81
80.120.980.600.820.81
90.080.990.670.810.80
(b) Emotional correlates of desire for help

Using path analysis, the variables most associated with desire for help were: (1) distress (DT standardised mean weight (SMW)=0.271) and (2) anxiety (HADS-A SMW=0.225) and depression (HADS-D SMW=0.122). However, even collectively variables explained only 42% of variance in desire for help (Figure 2).

thumbnail image

Figure 2. Path analysis of variables associated with desire for help. Numbers indicate standardised mean weights after adjustment for intercorrelations

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We found that only 20% of unselected patients and only 36% of distressed patients are willing to accept help for emotional or psychological difficulties. These results are similar to those previously reported 7, 20. Notably, although we found that patients who experienced distress were more likely to report wanting help, 64% of such patients did not feel that they required help with their symptoms or were not able to express their wish. The most common reason for declining in those with distress was receiving informal help elsewhere or preferring to manage on own; a finding that has been documented in primary care 21. In this preliminary sample, we found that reliance on the help question alone or the help question in combination with the PHQ-2 was not effective as a screening tool for emotional complications of cancer. Previously, one study conducted in a primary care setting 15 advocated the help question as a screening tool for depression, but in this study, we found a combination of the help question with the two items from the PHQ-2 significantly lowered the sensitivity with a minor gain in specificity and an overall reduction in accuracy. Nevertheless, clinicians may wish to consider asking all patients whether or not they feel that they need additional support as a prelude to asking about what sort of help the patients felt that they needed at that particular time point. This could allow treatment to be targeted at those who have unmet needs and who are willing to accept to the intervention. Related to this, we propose that desire for help can be quantified using a simple thermometer format. Indeed, a cut-off of 3 or above on the HT correctly identifies about 80% of responses to the desire for help question.

We also attempted to examine predictors of desire for help, but were not able to construct an entirely satisfactory model. Although there were unique contributions from distress, anxiety and depression, only 42% of variance in desire for help was explained. We looked at reasons for declining help and most common was receipt of pre-existing support from elsewhere although a few had concerns about whether or not any help would work for them and/or feared stigma of mental health treatment. Because those with severe distress may be reluctant to see a psychiatrist 22, it is important to offer help in a form that is acceptable and available to patients. In our study, the type of help that patients most requested was face-to-face contact and complementary therapy. It is likely that other variables not measured in this study contribute to desire for help. In fact few studies have examined predictors of desire for help or predictors of willingness to offer help. In one study from Amsterdam, 48 of 98 consecutive patients declined psychosocial help. Reasons given were that they had no need (n=17), they found it too burdensome (n=16) or they had received psychosocial treatment previously (n=1) 23. Merckaert et al. (2009) examined predictors of desire for help. Younger people, working men and women receiving hormone therapy had a greater desire for psychological support 24. Referral was associated with younger age, unmarried status, living alone, presence of more depressive symptoms, hopelessness, and attachment anxiety, and with less social support, self-esteem, and spiritual well-being. Ellis et al. (2009) examined predictors of psychosocial referral in 326 patients with metastatic gastrointestinal or lung cancer. Referral was associated with younger age, unmarried status, and presence of more depressive symptoms 25. Other factors of importance in acceptance of support or referral include practical barriers and accessibility 26 and concerns that staff may be too busy 27. In trial settings uptake of help is generally higher than in clinical practice but in a recent trial of psychological treatment of depression delivered by cancer nurses, 20% of patients declined participation, largely because of scepticism about whether the intervention would be effective 28.

Strengths of this study include the fact that two measures were used as comparison tools and patients were all given the questionnaires at the same time points within their treatment. However, there are a number of limitations. The sample size was small and the diversity of stages of disease and diagnosis meant that we were unable to carry out analysis by gender, diagnosis, age or ethnicity. In particular, the sample size for those who indicated that they wanted help in this study was small. Potential diagnosis was also made based on PHQ-9 and HADS scores; there is ongoing debate about the optimal cut-off point on the HADS and whether or not the total score should be used or whether the use of subscales is more optimal. The HADS and PHQ-9 were being used as the gold standard in this study, potentially limiting the interpretation of any results as a better gold standard would have been a structured interview. A further limitation of the study is that patients were asked to rate symptoms over the last week or 2 weeks depending on the tool used. Patients may prefer different sources of help at different points in their treatment 5, 29. For example, individual counselling was viewed as being useful for coping with distress during the course of treatment, but couple counselling may be more helpful for dealing with the effects of illness on other relationships once the acute treatment has ended 29. However, our study only focussed on those patients who were commencing chemotherapy.

We conclude that desire for help has an approximate relationship with mood complications of cancer but asking patients about their desire for help gives valuable information not achieved by mood screening alone. In general, those with higher levels of distress expressed most desire for help but some patients who were distressed did not want help or were reluctant to ask for it, whereas some patients who did not score highly on measures of psychological symptoms nevertheless wished to receive further assistance. Asking patients whether or not they want help, and quantifying this need is a useful means of detecting unmet need and focussing interventions on those willing to accept it.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We would like to thank Dr Abhro Chaudhuri and Dr Sridhar Thiagarajan for help with the data collection. There are no known conflicts of interest for any of the authors.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References