Can a tailored telephone intervention delivered by volunteers reduce the supportive care needs, anxiety and depression of people with colorectal cancer? A randomised controlled trial

Authors


Centre for Behavioural Research in Cancer Council, The Cancer Council Victoria, 1 Rathdowne Street, Carlton, Victoria, Australia 3053.

E-mail: Vicki.White@cancervic.org.au

Abstract

Objective

The objective was to assess the effectiveness of a volunteer-delivered tailored telephone-based intervention in reducing prevalence of unmet supportive care needs, elevated levels of anxiety and depression among people with colorectal cancer over a 9-month period.

Methods

There were 653 participants who completed the baseline questionnaire and were randomised to usual care (n = 341) or intervention (n = 306). Three follow-up questionnaires were completed at 3-monthly intervals (response rates: 93%, 87%, 82%, respectively). All four questionnaires contained the Supportive Care Needs Survey (SCNS), Hospital Anxiety and Depression Scale (HADS) and checklists for colorectal cancer symptoms and use of support services. The intervention consisted of trained volunteers providing emotional support, service referral and information and was delivered after completion of each of the first three questionnaires. Primary outcomes were prevalence of moderate to high SCNS needs and elevated levels (scores of 8+) of HADS anxiety and depression.

Results

Over the study period, SCNS needs decreased similarly for both groups, and prevalence of elevated depression did not change for either group. There was a greater reduction in the prevalence of elevated anxiety in the intervention than usual care group (p < 0.01), with the intervention group decrease mainly occurring between baseline and the first follow-up survey (p < 0.01). However, the prevalence of elevated anxiety was similar between the two groups at each follow-up point.

Conclusions

The intervention had no effect on supportive care needs or depression, although it may be associated with a greater reduction in anxiety. Future research should test the intervention with patients closer to diagnosis. Copyright © 2011 John Wiley & Sons, Ltd.

Introduction

Many people with cancer report unmet supportive care needs, particularly regarding information and psychosocial support [1-7]. As higher levels of supportive care needs are associated with elevated levels of psychological morbidity and distress [8-11], identifying effective strategies to reduce needs is important.

Although several randomised control trials have tested interventions to specifically reduce unmet supportive care needs [12-15], none have found an intervention effect. All have involved assessment of needs and a tailored intervention based on identified needs. Interventions have been delivered by medical staff (oncologists, GPs, nurses) and have generally been associated with the cancer treatment centre. Two studies using oncologists/GPs for intervention delivery reported low levels of discussion and referral for supportive care needs [13, 15], with only 48% of clinicians in one study documenting actions taken in response to identified needs [15]. Two studies have used nurses attached to the cancer treatment centre to deliver the intervention [12, 14]. In one study, the doctor, the person with cancer and the nurse discussed identified needs, and the nurse developed a management plan and made referrals as appropriate [14]. This study reported a low uptake of referrals.

Although many support services and information resources for people with cancer are provided by organisations external to the treatment setting, these organisations have rarely been utilised for the delivery of interventions to reduce supportive care needs. In one study that has explored this [15], the second intervention arm involved nurses attached to a cancer Helpline operated by a large cancer charity. Participants in this study arm were more likely to have their issues discussed and be referred to services than participants in the oncologist/GP intervention arm. Although at follow-up, fewer participants in this intervention arm reported unmet supportive care needs compared with the usual care and oncologist/GP groups, this difference was not statistically significant.

If community-based organisations are to be involved in the delivery of supportive care interventions, finding a mechanism that is cost-effective and acceptable to people with cancer and to the organisation is important. Volunteers are frequently used by cancer charities, hospitals and other community groups to provide support for people with cancer. Although volunteers are commonly involved with peer support programmes, they also provide non-peer based practical, informational, emotional and supportive care for people with cancer [16, 17]. We designed a volunteer-based intervention delivered under the auspices of The Cancer Council Victoria (CCV), an independent cancer charity mandated to deliver supportive care services to people with cancer. The CCV runs a telephone helpline and is experienced in running volunteer-based support programmes. The intervention aimed to provide people with cancer with emotional support and information and to refer them to other resources when needed. The intervention was tailored to suit an individual's type and level of need and was delivered over several time points to allow for the change in needs experienced over time [2]. The telephone was chosen as the delivery mechanism, as it has been shown to be an effective tool to convey information to people with cancer [18, 19] and has been used in volunteer-based programmes delivering support to people with cancer [16]. Telephone delivery also enables the intervention to be delivered at a population level.

We tested whether our volunteer-delivered intervention improves psychosocial outcomes of people with cancer over a 9-month period in a randomised controlled trial (RCT). This period covers treatment, finishing treatment and potentially first anniversary of diagnosis, time points when different needs may arise for people with cancer. Specifically, we examined whether the change over time in the proportion with supportive care needs and elevated levels of anxiety and depression was greater in the intervention group than the usual care group. The study utilised people with colorectal cancer; the second most common cancer among men and women in Australia [20]. As secondary outcomes, we examined whether the intervention reduced the number of colorectal symptoms and increased the use of support services in the intervention group relative to the usual care group.

Method

Research design

Randomised controlled trial with consenting participants allocated to intervention or usual care after completing the baseline (Time 1) questionnaire was conducted. Participants completed follow-up questionnaires at 3- (Time 2), 6- (Time 3) and 9-month (Time 4) post-study entry. All questionnaires were self-administered.

Sample

Eligible participants were those diagnosed with stages I–IIIb colorectal cancer, aged between 18 and 80 years, registered with the population-based Victorian Cancer Registry (VCR) within 3 months of their diagnosis, considered able to participate by their clinician and could understand English.

Measures

Demographics, treatment and cancer stage

In all questionnaires, participants answered questions regarding treatments received, age, employment and marital status. The VCR provided information on extent of disease at diagnosis.

Supportive care needs

The 59-item Supportive Care Need Survey (SCNS) [1, 21, 22] assessed level of need in all questionnaires. For each item, degree of need for help in the past month was indicated using a five-point scale (no need not applicable; no need satisfied; low need; moderate need; or high need). Items cover the domains of psychological, physical and daily living, patient care and support, health system and information and sexuality. The internal consistency of the full scale and the subscales is high, with Cronbachs' alpha greater than 0.84 for each domain. Concurrent validity is acceptable [21].

To assist intervention delivery, in each questionnaire, participants identified up to five SCNS needs that they most wanted help with (priority needs), with an option to indicate none.

Anxiety and depression

The Hospital Anxiety and Depression Scale (HADS) [23, 24] was completed at each time point. Scores for the anxiety and depression subscale range from 0 to 21 and can be classified into low/normal (scores 0–7), borderline (8–10) or clinically significant (11 or higher). Internal reliability for each scale is high [25]. On return of questionnaires, anxiety and depression scores were monitored, with those scoring 11 and above contacted by a cancer nurse independent of the study to assess distress and determine suitability of study continuation.

Colorectal symptoms

In consultation with nurses, oncologists and cancer survivors, we developed a nine-item colorectal cancer symptoms checklist. In each questionnaire, participants indicated (yes or no) if they had experienced the following symptoms in the past month: wind, diarrhoea or constipation, urination problems, stoma problems, weight changes, appetite changes, hair loss, body image and bowel pains. Positive responses were summed to give an overall colorectal symptoms score.

Use of health services

In each questionnaire, participants indicated their use of 11 different supportive care services (e.g. psychologist, home help, peer support services) in the previous 3 months. Positive responses were summed to give an overall use of services score.

Social support

In each questionnaire, participants completed the 19-item Medical Outcome Study (MOS) Social Support Scale [26]. Higher scores on this scale indicate greater social support.

Procedures

The study had institutional ethics approval. The VCR identified eligible people and contacted their clinician by mail to inform them of the study. If the clinician did not recommend otherwise, the VCR contacted identified people by mail to obtain permission to pass their contact and cancer details to the researchers. The research team approached those agreeing by mail and obtained written consent for study participation.

Randomisation

Participants were randomised to usual care or intervention within four age groups (18–49; 50–59; 60–69; 70–80).

Study arms

Usual care

Participants received a letter informing them of their allocation to this group and received usual care.

Intervention: the Pathfinder volunteer programme

Figure 1 summarises the intervention procedures that were implemented following the Times 1, 2 and 3 questionnaires. The intervention was premised on a supportive care model that suggests that all patients should be screened for unmet needs [27]. The different components of the intervention and the theoretical constructs drawn on to develop these components are as follows:

  1. Feedback on unmet needs and levels of anxiety and depression. Many behaviour change models assert that awareness of a problem or risk is needed for behaviour change to occur [28]. Based on this, the first step of our intervention involved participants receiving a tailored letter reflecting their priority supportive care needs (if any), colorectal symptoms (if any) and anxiety and depression levels if scores were 8 or above, as determined from their questionnaire responses. When there were no priority needs and anxiety and depression levels were normal, letters indicated that participants were doing well. The letter included the name of their allocated Pathfinder.
  2. Pathfinder contact. Pathfinders received a letter (called a case summary) summarising their assigned participant's needs and anxiety and depression levels. Pathfinders contacted their participant by phone within 1 week of receiving this letter. Pathfinder's contact with participants was based on social support theory that suggests that the availability of people who can assist in meeting resource and psychological needs contributes to positive adjustment and helps to reduce distress [29]. Pathfinders provided three of the four types of social support posited: emotional (e.g. offer reassurance), informational (offer advice) and instrumental (offer to provide resources) support. Pathfinders were trained in supportive communication techniques, including active listening skills, and used normalisation to assist in the provision of emotional support. Pathfinders were equipped with a manual (see succeeding paragraphs) that identified possible actions for resolving unmet needs or to reduce anxiety and depression, enabling them to provide informational and instrumental support. Drawing on concepts utilised in self-management education programmes [30], Pathfinders worked with participants to help them identify the actions that they wanted to utilise to address their unmet need. Participants were offered a written copy of the action plan (called a needs plan). Pathfinders called participants 2 weeks later, and if needs were not resolved, the needs plan was reviewed. After each call, Pathfinders completed a contact report detailing topics discussed, resources to be sent and length of call. Additional calls could be made to or from the participant, as required.
  3. Assessment of use of strategies. In subsequent questionnaires, participants receiving a written needs plan reported on their use of each listed strategy.
  4. Primary care doctor involvement. During Pathfinder calls, intervention participants indicated if they wanted their doctor to receive a copy of their tailored letter.
Figure 1.

Pathfinder training and contact with patients for the intervention

Recruitment and training of Pathfinders

Fifty-seven volunteers acted as Pathfinders throughout the study. Volunteers undertook a 3-day training programme and were only accepted into the programme upon satisfactory completion of several practise calls with programme staff. Volunteers were assessed on their skills in following the call protocol, responding to patients' emotional state, developing a needs plan with the patient and the volunteer's emotional response throughout and after the call. Pathfinders received ongoing supervision and training. Nearly all volunteers were female (93%), and 52% had a nursing/psychology background. We intended the Pathfinder programme to refer patients to peer support programmes rather than be a peer support service. To ensure that the programme did not become a peer support service, volunteers with a diagnosis of colorectal cancer were excluded.

Programme manual

A cancer nurse, with experience working on CCV's Cancer Helpline and in training volunteers for a support role, developed the Pathfinder Manual utilising the Helpline's database of services and support programmes for people with cancer. The manual provided a detailed guide to supporting people with cancer over the telephone and contained information regarding each SCNS need and colorectal symptom assessed, anxiety and depression and seven topics assessed in other areas of the questionnaire (e.g. work issues). The manual included a list of resources (services or liter ature) and strategies that might address each need/area. Strategies were based on published evidence and clinical experience. For example, for the need ‘lack of energy and tiredness’, the manual provided information around why this occurs, strategies for reducing fatigue such as taking several shorter rest periods rather than one long one, and listed written resources (e.g. a booklet entitled ‘Coping with fatigue caused by cancer treatment’).

Statistical analyses

Analyses were conducted using stata© (STATA Corp., TX, USA) [31]. Those not agreeing to randomization at Time 1 (n = 6) were removed. Data from 10 intervention group participants declining Pathfinder calls were included in the intervention group data. Primary outcomes were proportion of participants reporting the following: a moderate to high need on any of the 59 SCNS items and anxiety and depression scores ≥8. Secondary outcomes were number of colorectal cancer symptoms and support services used. Generalised linear latent and mixed models (GLLAMM) (either logistic or linear as appropriate), which included study arm, a categorical indicator of time and the interaction between group and time were generated. Likelihood ratio tests examined the significance of the interaction and thus the effectiveness of the intervention. GLLAMM models utilised data from all participants completing a questionnaire at each time point. Chi-square analyses, t-tests and GLLAMM models were used to examine intervention delivery for those with and without priority needs and for any follow-up cross-sectional analyses of differences in outcome measures between groups. GLLAMM models were used to examine whether receipt of the intervention improved perceptions of social support as indicated by the total MOS scale scores. Robust standard errors of estimates were calculated to account for clustering of participants at the volunteer level. For all analyses, two-sided p-values are reported.

Sample size calculations

Statistical power analyses indicated that we needed to retain 275 participants per group at study end to allow a 12% difference between groups in the change in proportion with any unmet needs and elevated levels of anxiety and depression to be detected with 85% power, 5% significance level, a design effect due to clustering of participants within Pathfinders of 1.24 and assuming a correlation between follow-up measurements of 0.5 or less.

Results

Participant clinical and demographic details

Figure 2 shows the number of eligible patients identified and participants in the study. There were 717 (79% of those agreeing to researcher contact) who consented to study participation, with 343 randomised to usual care and 306 to the intervention. Of these, 93% participated at Time 2 (93% usual care and 92% intervention), 87% at Time 3 (87% usual care and 87% intervention) and 82% at Time 4 (84% usual care and 80% intervention) (Figure 2). Over the course of the study, 100 participants had scores 11 or above on either HADS anxiety or depression scales. Of these, three were considered not suitable for further study participation (Figure 2), were withdrawn from that point and were offered other services.

Figure 2.

The Consolidated Standards of Reporting Trials (CONSORT) diagram of recruitment, attrition and delivery of intervention. #Percent of number randomised to that group

Demographic characteristics and clinical details of the two groups were similar (Table 1), although slightly more intervention participants received chemotherapy.

Table 1. Demographic and clinical details of participants in the intervention and usual care groups at study entry and treatment details over life of project
Patient characteristicsUsual care (n = 343)Intervention (n = 305)
Mean(SD)Mean(SD)
Age64.28(9.26)64.86(9.23)
Days between diagnosis and referral to researchers99.37(34.78)96.26(31.74)
 n%n%
Sex
Male1665919260
Female1174112740
Marital status
Married/defacto2217824978
Separated/divorced279268
Widowed176227
Not married/single124165
Highest education
Primary2593511
Secondary1435115348
Certificate/diploma60217724
University degree45163812
Currently employed79289731
Private health insurance1846519462
Cancer stage
I52186821
II1314616652
IIIA/IIIB100358527
Treatment (past and current)
Baseline(n = 343) (n = 306) 
Surgery3028827690
Chemotherapy1554516353
Radiotherapy34102910
Time 2(n = 319) (n = 283) 
Surgery2778725590
Chemotherapy1504714953
Radiotherapy33102910
Time 3(n = 299) (n = 266) 
Surgery2658922886
Chemotherapy1434813952
Radiotherapy31102811
Time 4(n = 288) (n = 245) 
Surgery2578921989
Chemotherapy1294513254
Radiotherapy32112912

Delivery of the intervention

Table 2 shows the proportion of participants in the two study arms with any priority needs and the three most common priority needs at each time point. Slightly more intervention participants than usual care participants indicated a priority need at baseline. However, at Times 2 and 3, the proportion reporting a priority need was similar in the two groups. The most common priority needs were the same across groups.

Table 2. Percentage of participants in the intervention and usual care group with any priority unmet needs at Times 1, 2 and 3 and the top three priority needs among those with a priority need
SurveyDomainTotalUsual careInterventionp-value
Baseline n = 649n = 343n = 306 
 Any priority needs45%41%48%0.06
 Top 3 priority needs among those with priority needs    
 Fear of cancer spreading26%21%31% 
 Fear of cancer returning26%20%32% 
 Tiredness17%20%14% 
Time 2 n = 602n = 319n = 283 
 Any priority needs41%40%41%0.83
 Top 3 priority needs among those with priority needs    
 Fear of cancer spreading30%32%28% 
 Fear of cancer returning28%30%25% 
 Tiredness18%19%17% 
Time 3 n = 565n = 299n = 266 
 Any priority needs32%34%31%0.35
 Top 3 priority needs among those with priority needs    
 Fear of cancer returning27%29%25% 
 Do things used to do22%18%27% 
 Fear of cancer spreading21%26%15% 
Time 4 n = 532n = 287n = 245 
 Any priority needs29%27%31%0.28
 Top 3 priority needs among those with priority needs    
 Fear of cancer returning30%36%25% 
 Not sleeping well22%24%20% 
 Do things used to do17%15%20% 

The proportion of intervention participants engaging in all elements of the intervention except contact with Pathfinder reduced between each time point (all p < 0.01), as did the time Pathfinders spent talking with participants (p < 0.01) (Table 3). Contact with Pathfinders reduced significantly between baseline and Time 2 but did not change significantly between Times 2 and 3. Although around half of the participants wanted literature sent at baseline, this reduced to around 20% by Time 3. At each time point, participants with at least one priority need were more likely to request a written needs plan (all p < 0.01) and have longer contact with Pathfinders (all p < 0.01).

Table 3. Proportion of all intervention participants and those with and without priority needs receiving each of the different components of the intervention after the first three surveys
 BaselineTime 2Time 3
  • When data for intervention participants were missing because of a contact report not being returned, it was assumed that scores for the different variables were 0. Has priority needs, participants indicating any priority needs at each time point; No priority needs, participants not indicating a priority need at each time point.

  • *

    Difference between priority needs and no priority needs significant at p < 0.05.

  • **

    Difference between priority needs and no priority needs significant at p < 0.01.

Intervention elementAllNo priority needs≥1 priority needAllNo priority needs≥1 priority needAllNo priority needs≥1 priority need
n%n%n%n%n%n%n%n%n%
Total N306 158 148 283 167 116 266 185 81 
Accepted call2909514994141952408514386978421179147806479
Tailored letter to doctor1434768437551551926162925411525141620
Had at least one strategy discussed1685574479464**983541255750**572127153037**
Requested a typed needs plan1153851326443*461617102925**2491051417*
Literature sent1575167429061**612222133934**331219101417
 Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)Mean (SD)
Number of Pathfinder contacts1.7 (0.8)1.7 (0.8)1.8 (0.8)1.4 (1.1)1.3 (0.9)1.6 (1.3)**1.2 (1.0)1.2 (0.9)1.4 (1.0)
Duration of total Pathfinder contact (minutes)41.6 (30.8)36.7 (30.2)46.8 (30.7)**27.3 (30.5)20.5 (21.0)37.0 (38.5)**21.5 (22.9)19.3 (21.7)26.7 (24.7)**
Patient report of number of strategies used2.0 (3.8)1.7 (3.3)2.4 (4.2)0.8 (2.9)0.4 (1.5)1.4 (4.0)**0.4 (1.9)0.2 (1.2)1.0 (2.8)**

There was a significant interaction between study arm and time for MOS social support (p = 0.04). Perceived social support increased among the intervention group (mean scores, baseline through Time 4: 74.8, 77.1, 78.4, 79.7), whereas, it remained constant among the usual care group (mean scores, baseline through Time 4: 76.9, 76.3, 76.5, 77.0).

Primary outcomes

Supportive care needs

The prevalence of moderate to high unmet needs reduced over the study period for both groups, and there was no significant interaction between time and study arm (χ2 = 2.85, df = 3, p = 0.41) (Table 4).

Table 4. Proportion of usual care and intervention participants with any moderate to high unmet needs, elevated HADS anxiety or depression scores at each time point
Domain Usual careInterventionχ2 statistic and p-value for interactiona
n%n%
  1. HADS: Hospital Anxiety and Depression Scale, scores ≥ 8 on the scales indicate elevated levels.

  2. a

    Likelihood ratio test used to determine significance of interaction between time and study group. GLLAMM models adjusted for clustering at Pathfinder level.

% moderate or high unmet needsBaseline1674915350 
 Time 2124369832 
 Time 399297725 
 Time 472216321χ2 = 2.85, df = 3, p = 0.41
% elevated HADSBaseline53166020 
AnxietyTime 250163211 
 Time 352173513 
 Time 444152912χ2 = 13.5, df = 3, p < 0.01
% elevated HADS DepressionBaseline44134013 
 Time 242133312 
 Time 330102911 
 Time 430113213χ2 = 1.40, df = 3, p = 0.71

Anxiety and depression

There was a significant interaction between study arm and time for anxiety (χ2 = 13.5, df = 3, p < 0.01), with the prevalence of elevated anxiety decreasing over time among the intervention group, while remaining fairly constant among the usual care group. The main reduction in the intervention group occurred between baseline and Time 2. However, the prevalence of anxiety was similar for intervention and usual care groups at each follow-up. Prevalence of elevated depression did not change over the study period for either group (interaction: χ2 = 1.40, df = 3, p = 0.71).

Secondary outcomes

Colorectal cancer symptoms

There was a greater reduction in the average number of colorectal symptoms over time among the intervention group relative to the usual care group (interaction: χ2 = 10.43, df = 3, p = 0.02) (Table 5). This was mainly due to a greater decrease in colorectal needs between baseline and Time 2 for the intervention group relative to the usual care group. However, despite this greater decrease, the average number of colorectal symptoms reported was similar for the intervention and usual care groups at each follow-up time.

Table 5. Mean scores and standard deviations (SD) for intervention and usual care groups on number of colorectal symptoms reported and use of services at each time point
  Usual care mean (SD)Intervention mean (SD)χ2 statistic and p-value for interactiona
  1. a

    Likelihood ratio test used to determine significance of interaction between time and study group. GLLAMM models adjusted for clustering at pathfinder level.

Colorectal symptomsBaseline2.7 (2.1)3.0 (2.0) 
 Time 22.3 (1.9)2.2 (1.9) 
 Time 31.9 (1.8)1.8 (1.8) 
 Time 41.6 (1.7)1.6 (1.7)χ2 = 10.43, df = 3, p = 0.02
Use of servicesBaseline1.0 (1.2)1.1 (1.2) 
 Time 20.5 (0.9)1.1 (1.3) 
 Time 30.5 (0.9)0.9 (1.2) 
 Time 40.4 (0.8)0.8 (1.1)χ2 = 22.9, df = 3, p < 0.001

Use of support services

Although there was a decline in the reported use of support services over time for both groups, the decline was greater among usual care participants (interaction χ2 = 22.9, df = 3, p < 0.001) (Table 5). The greater decline in the usual care group resulted in the intervention group reporting, on average, a greater use of services at all three follow-up points (all ps < 0.01).

Discussion

We examined whether a needs-based volunteer-delivered intervention for people recently diagnosed with stages I–IIIb colorectal cancer resulted in lower prevalence of supportive care needs and elevated anxiety and depression over a 9-month period compared with usual care. No evidence of an intervention effect on these measures was found. A reduction in the prevalence of elevated anxiety between baseline and Time 2 was observed for the intervention but not for the usual care group. Although this might suggest an intervention effect, as the prevalence of elevated levels of anxiety was slightly higher among the intervention than the usual care group at baseline and as the prevalence of elevated anxiety was similar for both groups at each follow-up, we cannot rule out the possibility that this effect was due to ‘regression to the mean’. The average number of colorectal cancer symptoms reduced more, and the use of services reduced less among the intervention than the usual care group, again mainly because of changes between baseline and Time 2.

A number of factors may have limited the impact of our intervention. First, unlike other studies in this area, we used volunteers rather than health professionals to deliver the intervention. Volunteers were attached to a population-focused cancer charity with the remit to support people with cancer. Half of our volunteers had a background in nursing or psychology, and all undertook training in manual use and in supporting people with cancer. Our feasibility study [32] suggested that the use of volunteers was acceptable to participants. The high proportion of intervention participants accepting Pathfinder calls supports this. Volunteers can deliver complex interventions [17, 33]. In addition, our findings suggest that our volunteers and intervention were effective at enhancing perceptions of social support. As previous RCTs aiming to reduce unmet needs of people with cancer using nurses or doctors for intervention delivery have not found an intervention effect, we do not believe our results were because of our use of volunteers.

Second, the intervention was delivered to people with early-stage colorectal cancer who were about 3–4 months post-diagnosis at study entry. A recent review of the unmet needs literature found that the degree of need was highest during treatment [2]. In our study, around 50% of participants were receiving chemotherapy at study entry. As the greatest decrease in the prevalence of moderate to high needs among our study participants occurred between our baseline and first follow-up questionnaires, our intervention may be more effective if delivered to people undergoing treatment. A recent pilot study of a nurse-delivered, telephone-based intervention for people with colorectal cancer that delivered the first three intervention points in the first month after hospital discharge provides tentative support for this suggestion [34].

Third, although the intervention was tailored to participant's level of need, it may be that contact between Pathfinder and participants was not frequent enough. However, the study attempted to examine what a volunteer-based cancer organisation might be able to do within its existing resource base. Providing a higher resource intervention would be difficult for such institutions to deliver.

Fourth, we took a population approach to intervention delivery with all intervention participants having contact with a Pathfinder. This follows a supportive care model that suggests that all people with cancer should be screened for needs [27]. Around 40% of participants indicating no priority needs had literature sent, and around one-third developed a needs plan, supporting our population approach. Tailoring our intervention meant volunteers spent longer time with participants reporting priority supportive care needs, and these participants were more likely to receive literature, discuss needs with their Pathfinder and use suggested strategies. However, despite our finding that volunteers could tailor the intervention appropriately, it is possible that the programme's effectiveness may be improved if it was targeted at patients with a certain level of unmet needs. This might also help to retain volunteers as some found contacting people with no needs difficult.

Fifth, although it seems intuitive that needs should reduce if people are informed of strategies to address them, ours is now the third sufficiently powered RCT to not find an intervention effect. Studies conducted to date [12-15] have employed a range of cancer types and disease stages, and their interventions have been relatively intense. All studies have used the SCNS or its earlier version to both tailor the intervention and assess its outcome. The SCNS assesses the presence of a problem in the previous month and the level of help wanted. The time frame for assessing needs (i.e., previous month) may not be appropriate for assessing the impact on an intervention delivered 3 months earlier. Our finding that the use of supportive care services reduced more among the usual care than intervention group may suggest that the intervention enables participants to use or continue their use of services. As others have suggested [35], interventions like the one tested here may impact on participants' self-confidence in accessing and using services. Future research might benefit from including these measures among their outcomes.

Sixth, methodological inadequacies in the current study may have hindered its success. As participants were not blind to whether they received the intervention, ‘demand characteristics’ including a desire to suggest their Pathfinder was performing well may have influenced intervention participants' follow-up questionnaire responses. However, as the change in supportive care needs was similar for the intervention and control groups, we do not believe this to be the case. Although it is possible that participants in the usual care group may have actively sought help to address needs identified by completing the questionnaire, as the average number of services used by this group decreased over time, this seems unlikely.

This study suggests that a volunteer-delivered, telephone-based intervention, run from a cancer charity, is plausible and acceptable to people with cancer. However, as the intervention was not effective at producing a greater decrease in the prevalence of supportive care needs or in lowering depression levels, the results suggest that this programme in its current format does not warrant roll-out. Although we cannot rule out the possibility that the significantly greater reduction in anxiety among intervention than usual care participants was due to ‘regression to the mean’ rather than our intervention, this finding is promising. As the reduction was mainly due to change between baseline and the first follow-up, our results suggest that work investigating whether intervening closer to diagnosis is effective is warranted. Including measures reflecting participants' confidence in using support services as outcome measures may also benefit work in this area.

Acknowledgements

We thank Dr Haley Matic for her work managing the Pathfinder Programme and Ms Sarah Pratt for developing the Pathfinder manual and training programme. We also thank the participants and Pathfinder volunteers for their time and commitment to this project.

This project was funded by a project grant from the National Health and Medical Research Council of Australia (Project #300759) and by the Cancer Council Victoria. There are no financial disclosures to report for any author.

Ancillary