Prevalence and correlates of the unmet supportive care needs of individuals diagnosed with a haematological malignancy

Abstract

Background. Information about the unmet supportive care needs of haematological cancer patients is needed for service planning and to inform clinical practice. This study described the prevalence of, and factors associated with, unmet supportive care needs among haematological cancer patients.

Material and methods. A total of 380 adults diagnosed with haematological cancer were recruited from the outpatient department at three comprehensive cancer treatment centres in Australia. Of these, 311 completed a self-report questionnaire. Unmet needs were assessed using the 34-item Supportive Care Needs Survey (SCNS-SF34). The data were examined using descriptive statistics and logistic regression.

Results. Overall, 51% (95% CI 45–57%) of participants reported having at least one ‘moderate to high’ level unmet need, while 25% (95% CI 20–30%) reported ‘no need’ for help with any items. Unmet needs were most commonly reported in the psychological (35%; 95% CI 30–41%) and physical aspects of daily living (35%; 95% CI 30–41%) domains. The three most frequently endorsed items of ‘moderate to high’ unmet need were: lack of energy/tiredness (24%; 95% CI 20–30%), not being able to do the things you used to do (21%; 95% CI 17–26%), and uncertainty about the future (21%; 95% CI 16–25%). Patients’ sociodemographic characteristics influenced unmet needs more than disease characteristics. Patients who were female, aged less than 55 years or not in the labour force had higher odds of reporting ‘moderate to high’ level unmet supportive care needs.

Conclusion. Unmet supportive care needs are prevalent among haematological cancer patients, particularly in the psychological and physical aspects of daily living domains. These findings provide valuable insight about the range of resources, multidisciplinary linkages and referral pathways needed to address haematological cancer patients’ unmet needs.

Background

Haematological cancers represent a wide range of diseases with distinct illness trajectories, treatment pathways and outcomes [1]. These cancers accounted for 7% of all new cancers diagnosed worldwide in 2008 [2]. While the clinical course of haematological cancer is highly variable, for many patients it is a chronic condition characterised by prolonged duration, multiple acute episodes, functional impairment or disability, and incurability. Studies of the quality of life of haematological cancer patients have identified that many experience debilitating symptoms including fatigue and pain; reduced role and physical functioning; and decreased emotional wellbeing [3–5]. Compared with patients diagnosed with other common cancer types, haematological cancer patients experience higher levels of psychological morbidity with 27–48% reporting symptoms of anxiety and 17–38% reporting symptoms of depression [6].

The assessment of unmet needs is one of several approaches for determining the psychosocial impact of cancer. Unmet needs are the result of a mismatch between the actual services or care that an individual receives, and those they perceive necessary to achieve optimal wellbeing [7]. In contrast to assessments of quality of life and psychological morbidity, needs assessment does not assume that the presence of a side effect or deficit in functioning means that the patient wants help with that issue. In the context of the Supportive Care Framework [7] needs assessment involves patients explicitly identifying the physical, informational, emotional, psychological, social, spiritual and practical issues they perceive that they need help with, and the magnitude of their need for help. A strength of this approach is that it allows the delivery of care and the allocation of resources to focus on the issues that patients themselves have prioritised [8,9].

Reviews of the literature have identified that, compared with other cancer types, relatively few studies have examined the unmet needs of people diagnosed with haematological cancer [10–12]. Findings from the handful of studies conducted with haematological cancer patients suggest that unmet needs may be highest in the physical, psychological and practical domains. Feeling tired, fear of cancer recurrence, uncertainty about the future, and access to car parking are among the most prevalent unmet needs [12–15]. Younger patients appear more likely to report unmet needs than older patients [14,16,17]. Those reporting unmet emotional needs also appear more likely to report unmet needs in other domains, including financial, physical, sexual, social and family needs [17].

While these findings provide some insight, a recent review by Hall and colleagues [18] highlighted that previous studies of unmet needs with haematological cancer patients are typically characterised by small sample sizes, homogenous haematological cancer types, low response rates, and lack of standardised or multidimensional needs assessment measures. These limitations compromise the confidence with which inferences can be drawn from the study findings and thus the utility of the data. Robust information about the unmet supportive care needs of haematological cancer patients is necessary to inform clinical practice and service planning that supports patient-centred care. Therefore, the aims of this study were to:

1)	describe the prevalence of supportive care needs among adults diagnosed with haematological cancer;
2)	identify the most prevalent items of ‘moderate to high’ level unmet need; and
3)	identify sociodemographic and disease characteristics associated with haematological cancer patients reporting ‘moderate to high’ levels of unmet need in the psychological, health system and information, physical aspects of daily living, patient care and support, and sexuality domains.

Material and methods

Design and setting

A cross-sectional survey of patients attending outpatient haematology clinics at three comprehensive cancer treatment centres in Australia was undertaken. Each treatment centre was located in a major city of a different state, and treated more than 300 adult haematological cancer patients per year.

Sample

Patients were eligible to participate in the study if they had a diagnosis of haematological cancer, were attending the outpatient haematology department for their second or subsequent appointment between October 2012 and March 2013, were aged 18 years or older, and able to read and understand English adequately to provide informed consent and complete a questionnaire. Patients who were deemed by clinical staff to be physically or psychologically incapable of completing the questionnaire were excluded. These criteria meant that a diversity of haematological cancer patients in terms of haematological cancer type and length of time since diagnosis were eligible to participate.

Procedure

A haematologist or nurse identified potentially eligible patients from the daily clinic list. Potentially eligible patients were approached in the clinic waiting room by a trained Research Assistant and provided with written information about the study. Patients who consented to participate were asked to complete a self-administered pen-and-paper questionnaire while waiting for their appointment. Patients who were unable to complete the questionnaire prior to their appointment were provided with a reply-paid envelope to return their completed questionnaire to the research team. One written reminder was sent to patients who had not returned their completed survey within two weeks of consenting to participate. The gender and age of non-consenting patients was recorded. The research was conducted in accordance with the protocol approved by the Human Research Ethics Committee of the University of Newcastle and each treatment centre.

Measures

Supportive Care Needs Survey – Short Form 34 (SCNS-SF34) is a 34-item self-report measure which assesses cancer-specific perceived needs across five domains: psychological (10 items); health system and information (11 items); physical aspects of daily living (5 items); patient care and support (5 items); and sexuality (3 items). For each item, respondents indicate their level of need for help in the last month, as a result of cancer, on a five point scale where 1 = no need, not applicable; 2 = no need, satisfied; 3 = low need; 4 = moderate need; and 5 = high need. The SCNS-SF34 was developed with a heterogeneous sample of cancer patients in terms of cancer type, extent of disease and time since diagnosis. The domains were confirmed using factor analysis, and psychometric properties demonstrate acceptable internal consistency, convergent validity and discriminant validity [19].

Sociodemographic and disease characteristics including sex, age, Indigenous status, country of birth, marital status, living arrangements, highest education level, employment status, private health insurance status, concession card holder status, smoking status, type of haematological cancer at diagnosis, extent of disease, time since diagnosis, treatments received and relocation for treatment were assessed by self-report items. Residential postcode was obtained to assess geographical remoteness according to the Accessibility Remoteness Index of Australia [20].

Statistical analysis

If less than half of the items on a SCNS-SF34 subscale were missing, the mean of the remaining subscale items was imputed [21]. The proportion of survivors who reported either ‘no’ need (i.e. selected response option 1 or 2 for all 34 items), ‘low’ need (i.e. selected response option 3 for at least one item, but did not select response option 4 or 5 for any items) and ‘moderate to high’ need (i.e. selected response option 4 or 5 for at least one item) was calculated overall and by domain, with 95% confidence intervals. For each item, the proportion of participants who reported a ‘moderate to high’ level of unmet need (i.e. selected response option 4 or 5) was calculated with 95% confidence intervals, and the corresponding domain was identified. The 10 items with the highest prevalence were then identified. For each domain, individual associations between participant sociodemographic and disease characteristics and reporting a ‘moderate to high’ need versus ‘no or low’ need were examined using simple logistic regression. Variables with a p-value ≤ 0.2 on the Wald test were included in a multiple logistic regression model for each domain. Variables with an adjusted p-value ≤ 0.05 on the Wald test were considered statistically significant. All models also included treatment centre to account for the sampling [22]. Adjusted odds ratios and 95% CIs are reported for the final model for each domain.

Results

Sample

Of the 422 eligible haematological cancer patients approached, 380 (90%) consented to participate in the study and 311 (74%) returned a completed questionnaire. As shown in Table I, more than half of participants were male (58%), aged 55–74 years (55%), resided in a major city (68%), were married or partnered (67%) and not in the labour force (60%). The most common types of haematological cancer diagnosed were lymphoma (40%), leukaemia (27%) and myeloma (21%). Just over one half of participants were diagnosed more than two years ago (53%) and had received chemotherapy plus some other cancer treatment (58%). A significantly higher proportion of patients aged less than 55 years declined to participate compared with patients 55 years and older (85% vs. 15%; p < 0.001).

Table I. Characteristics of study sample (n = 311).

Characteristic	Grouping	n	(%)
Sex	Male	179	(58)
	Female	132	(42)
Age	Less than 55 years	76	(25)
	55–74 years	171	(55)
	75 years and over	62	(20)
Indigenous status	Non-indigenous	302	(99)
	Indigenous	3	(1)
Geographical location	Major city	208	(68)
	Regional or remote	96	(32)
Marital status	Married or partner	205	(67)
	Single, divorced, separated or widowed	103	(33)
Education completed	High school or below	150	(49)
	Vocational training or other	91	(30)
	University	66	(21)
Living arrangements	With others	238	(78)
	Alone	66	(22)
Country of birth	Australia	212	(69)
	other English speaking	47	(15)
	non-English speaking	50	(16)
Employment status	Not in labour force	184	(60)
	Paid employment	113	(37)
	Unemployed	10	(3)
Private health insurance	Yes	177	(57)
	No	133	(43)
Concession card holder	Yes	173	(56)
	No	137	(44)
Smoking status	Current smoker	28	(9)
	Former smoker	114	(37)
	Never smoked	168	(54)
Cancer type	Lymphoma	132	(43)
	Leukaemia	84	(27)
	Myeloma	66	(21)
	Other blood cancers	25	(8)
Extent of disease	Early or in remission	121	(40)
	Advanced	73	(24)
	Don't know	110	(36)
Time since diagnosis	12 months or less	84	(27)
	13–24 months	60	(19)
	More than 24 months	165	(53)
Treatment received	Chemotherapy only	66	(22)
	Chemotherapy and other treatment*	175	(58)
	Other treatment only*	30	(10)
	No treatment	31	(10)
Relocate for treatment	No	277	(90)
	Yes	32	(10)
Place of treatment	Centre 1	56	(18)
	Centre 2	112	(36)
	Centre 3	143	(46)

*Other treatment included radiotherapy, hormone therapy, biological treatments, bone marrow or stem cell transplant.

Prevalence of supportive care needs

Table II shows that overall, half (51%, 95% CI 45–57%) of participants reported having at least one ‘moderate to high’ level unmet supportive care need. One quarter (25%; 95% CI 20–30%) of participants reported ‘no need’ for help for all of the 34 items assessed. ‘Moderate to high’ level unmet needs were most common in the psychological (35%; 95% CI 30–41%) and physical aspects of daily living (35%; 95% CI 30–41%) domains, followed by the health system and information (20%; 95% CI 15–24%) domain. Relatively few participants reported a ‘moderate to high’ level unmet need for help with sexuality (12%; 95% CI 8–16%) or patient care and support (10%; 95% CI 6–13%) issues.

Table II. Prevalence of supportive care needs overall and by domain.

	No needs^†		Low needs^‡		Moderate/high needs^§
	n	% (95%CI)	n	% (95% CI)	n	% (95% CI)
Overall	75	25 (20–30)	71	24 (19–29)	151	51 (45–57)
Psychological	121	40 (34–45)	76	25 (20–30)	108	35 (30–41)
Physical aspects of daily living	121	40 (34–45)	75	25 (20–29)	108	35 (30–41)
Health system and information	204	67 (62–72)	40	13 (9–17)	61	20 (15–24)
Sexuality	224	74 (69–79)	44	14 (10–18)	35	12 (8–16)
Patient care and support	242	80 (75–84)	31	10 (7–14)	30	10 (6–13)

^†selected ‘no’ need for help for all items; ^‡selected ‘low’ level need for help for at least one item, but did not select ‘moderate’ or ‘high’ need for any item; ^§selected ‘moderate’ or ‘high’ level need for help for at least one item.

Most prevalent unmet supportive care needs at the ‘moderate to high’ level of need

Table III shows the 10 highest ranked items of ‘moderate to high’ level unmet need reported by participants. The three highest ranked items were lack of energy/tiredness (24%; 95% CI 20–30%), not being able to do the things you used to do (21%; 95% CI 17–26%), and uncertainty about the future (21%; 95% CI 16–25%). Half of the top 10 needs items were from the physical aspects of daily living and the other half were from the psychological domain.

Table III. Ten most prevalent ‘moderate to high’ level unmet supportive care needs.

Rank	Item	n	% (95% CI)	Domain
1	Lack of energy/tiredness	74	24 (20–30)	Daily living
2	Not being able to do the things you used to do	65	21 (17–26)	Daily living
3	Uncertainty about the future	63	21 (16–25)	Psychological
4	Concerns about the worries of those closest to you	59	19 (15–24)	Psychological
5	Work around the home	53	17 (13–22)	Daily living
6	Fears about the cancer spreading	51	17 (13–21)	Psychological
7	Worry that the results of treatment are beyond your control	42	14 (10–18)	Psychological
8	Feeling unwell a lot of the time	38	13 (9–17)	Daily living
9	Pain	38	12 (9–17)	Daily living
10	Anxiety	37	12 (9–16)	Psychological

Total number of observations for each item ranges from 297 to 307 due to missing values.

Factors associated with ‘moderate to high’ level unmet need

Tables IV–VIII show the sociodemographic and disease characteristics associated with patients reporting ‘moderate to high’ level unmet needs for each domain. Females (OR = 1.86; 95% CI 1.12–3.08) had almost twice the odds of reporting psychological needs compared with males. The odds of reporting health system and information needs and patient care and support needs decreased with increasing age. Patients not in the labour force (retired, pensioner, home duties, carer) had four times the odds (OR = 4.5; 95% CI 1.23–16.52) of reporting patient care and support needs compared with those who were in paid employment. The association between type of haematological cancer and physical aspects of daily living needs was marginally significant. Compared with patients diagnosed with lymphoma, those diagnosed with myeloma (OR = 2.05; 95% CI 1.02–4.13) or other types of blood cancer (OR = 2.94; 95% CI 1.07–8.04) had at least twice the odds of reporting physical aspects of daily living needs.

Table IV. Factors associated with ‘moderate to high’ level unmet psychological needs.

Variable	Adjusted OR* (95% CI)	Adjusted p-value^†
Sociodemographic
Sex		0.016
Male	1.0
Female	1.86 (1.12–3.08)
Health insurance		0.106
Yes	1.0
No	1.53 (0.91–2.57)
Concession card		0.470
Yes	1.0
No	0.82 (0.49–1.40)
Disease
Time since diagnosis		0.083
< 12 months	1.0
13–24 months	0.56 (0.27–1.16)
> 24 months	0.52 (0.29–0.94)
Treatments received		0.118
Chemotherapy only	1.0
Other treatment only	2.85 (1.10–7.38)
Chemotherapy and other	1.76 (0.90–3.44)
No treatment	1.11 (0.41–3.00)

Variables with a crude p-value ≤ 0.2 on the Wald test were included in the final model; those with an adjusted p-value ≤ 0.05 were considered statistically significant.

*adjusted for all variables in the final model; ^†p-value from the Wald test.

Table V. Factors associated with ‘moderate to high’ level unmet health system and information needs.

Variable	Adjusted OR* (95% CI)	Adjusted p-value^†
Sociodemographic
Age group		0.005
Less than 55	1.0
55–74	0.34 (0.16–0.71)
75 and older	0.17 (0.05–0.58)
Employment		0.603
Paid employment	1.0
Not in labour force	1.26 (0.59–2.68)
Unemployed	1.94 (0.48–7.85)
Disease
Extent of cancer		0.257
Early or in remission	1.0
Advanced	1.64 (0.78–3.44)
Don't know	0.89 (0.43–1.85)
No treatment

Variables with a crude p-value ≤ 0.2 on the Wald test were included in the final model; those with an adjusted p-value ≤ 0.05 were considered statistically significant.

*adjusted for all variables in the final model; ^†p-value from the Wald test.

Table VI. Factors associated with ‘moderate to high’ level unmet patient care and support needs.

Variable	Adjusted OR* (95% CI)	Adjusted p-value^†
Sociodemographic
Age group		0.014
Less than 55	1.0
55–74	0.20 (0.07–0.61)
75 and older	0.16 (0.03–0.87)
Employment		0.037
Paid employment	1.0
Not in labour force	4.50 (1.23–16.52)
Unemployed	4.61 (0.83–25.77)
Concession card		0.075
Yes	1.0
No	2.66 (0.91–7.80)
Disease
Extent of cancer		0.057
Early or in remission	1.0
Advanced	1.67 (0.66–4.23)
Don't know	0.39 (0.12–1.27)
Relocate for treatment		0.057
No	1.0
Yes	3.08 (0.97–9.80)

Variables with a crude p–value ≤ 0.2 on the Wald test were included in the final model; those with an adjusted p–value ≤ 0.05 were considered statistically significant.

*adjusted for all variables in the final model; ^†p-value from the Wald test.

Table VII. Factors associated with ‘moderate to high’ level unmet physical and daily living needs.

Variable	Adjusted OR* (95% CI)	Adjusted p-value^†
Sociodemographic
Age group		0.619
Less than 55	1.0
55–74	1.27 (0.60–2.68)
75 and older	1.63 (0.61–4.35)
Employment		0.325
Paid employment	1.0
Not in labour force	1.63 (0.74–3.63)
Unemployed	2.4 (0.55–10.55)
Health insurance		0.087
Yes	1.0
No	1.66 (0.93–2.97)
Concession card		0.469
Yes	1.0
No	0.76 (0.37–1.59)
Disease
Cancer type		0.059
Lymphoma	1.0
Leukaemia	1.11 (0.52–2.35)
Myeloma	2.05 (1.02–4.13)
Other blood	2.94 (1.07–8.04)
Extent of cancer		0.221
Early or in remission	1.0
Advanced	1.50 (0.74–3.04)
Don't know	0.80 (0.40–1.58)
Time since diagnosis		0.086
< 12 months	1.0
13–24 months	0.39 (0.17–0.91)
> 24 months	0.77 (0.4–1.48)
Treatments received		0.079
Chemotherapy only	1.0
Other treatment only	2.25 (0.82–6.15)
Chemotherapy and other	1.50 (0.73–3.07)
No treatment	0.47 (0.14–1.55)

Variables with a crude p-value ≤ 0.2 on the Wald test were included in the final model; those with an adjusted p-value ≤ 0.05 were considered statistically significant.

*adjusted for all variables in the final model; ^†p-value from the Wald test.

Table VIII. Factors associated with ‘moderate to high’ level unmet sexuality needs.

Variable	Adjusted OR* (95% CI)	Adjusted p-value^†
Sociodemographic
Living arrangement		0.126
With others	1.0
Alone	0.42 (0.14–1.27)
Disease
Extent of cancer		0.100
Early or in remission	1.0
Advanced	1.52 (0.68–3.40)
Don't know	0.52 (0.2–1.35)

Variables with a crude p-value ≤ 0.2 on the Wald test were included in the final model; those with an adjusted p-value ≤ 0.05 were considered statistically significant.

*adjusted for all variables in the final model; ^†p-value from the Wald test.

Discussion

Prevalence of unmet needs

This multi-site cross-sectional study is one of the largest investigations of the unmet needs of haematological cancer patients conducted to date using a validated, multidimensional, cancer-specific needs assessment measure. While it is not unexpected for some patients to have unmet needs following a cancer diagnosis, this study found that half (51%) of haematological cancer patients receiving outpatient care at three comprehensive cancer centres reported having moderate to high level unmet supportive care needs. Only one quarter of haematological cancer patients reported having no unmet supportive needs in the last month. Both the prevalence of unmet needs by domain and by individual item showed that haematological cancer patients most often reported moderate to high level unmet needs in the psychological and the physical aspects of daily living domains.

There is limited opportunity for comparing the results of the current study to previous findings as few studies have examined the supportive care needs of individuals diagnosed with haematological cancer, either as discrete samples or as a distinct sub-group within mixed cancer samples. Furthermore, the few studies that do exist vary in sample characteristics, the measures used to assess unmet needs, the criteria used to classify unmet needs, and the reporting of outcomes [18]. Nevertheless, compared with the small number of previous studies which have used a validated, multidimensional cancer-specific needs assessment measure, the overall prevalence of unmet need (75%) identified in the current study is similar to that reported in two previous studies involving samples with mixed haematological cancer types [12,23]. However, this prevalence rate is almost three times higher than that reported in a study of individuals diagnosed with myeloma alone [15]. Compared with a previous study which also used the SCNS-SF34, the prevalence of unmet need reported by haematological cancer patients in the current study is considerably higher than that reported by individuals diagnosed with prostate, colorectal, head and neck cancer or melanoma, and similar to that reported by individuals diagnosed with breast or lung cancer [23].

While the proportion and rank order of domains and specific items of highest unmet need varies between studies, the current findings are consistent with the general pattern of unmet needs reported by haematological cancer patients’ and individuals with solid tumours. That is, needs are typically highest in the physical and psychological domains, and tiredness, uncertainty about the future, and fear of cancer spread are among the top ranked items of unmet need [12,13,15,23]. Focusing clinical attention and limited healthcare resources on these issues may best assist patients to achieve optimal wellbeing.

Factors associated with unmet needs

Compared with other needs assessments conducted with oncology populations, relatively few of the socio-demographic and disease characteristics examined in the current study were associated with domains of unmet need. This may reflect the distinctive nature of haematological cancers compared to solid cancers. It could also be a consequence of the limited array of characteristics examined, the use of patient self-reported disease information wherein the extent of disease was unknown for 36% of participants, or the limited power of the study to detect moderate to small associations. Nevertheless, the findings that unmet needs had higher odds of being experienced by haematological patients who were female, younger and not in the labour force are consistent with results from needs assessments conducted with general oncology populations [10,24]. The current findings also concur with those of Fielding and colleagues [25] who found that individual characteristics appear to have a stronger influence than medical characteristics in cancer patients’ unmet needs. These characteristics may serve as ‘red flags’ for identifying which patients may be at increased risk of poor psychosocial outcomes and require monitoring or early intervention.

Clinical implications of study findings

The findings from this study highlight the range of resources, multidisciplinary linkages and referral pathways that health care providers need to have in place in order to address haematological cancer patients’ unmet supportive care needs. In particular, these data suggest that health care providers need to be able to link patients with occupational therapists and physiotherapists to assist with physical aspects of daily living, and psychologists and social workers to assist with psychological issues. Ideally, pathways of care would ensure that individuals reporting unmet needs were followed up by a healthcare provider to plan and initiate interventions to reduce those needs.

Strengths and limitations

The identification of study participants from three comprehensive cancer centres located in three major cities enabled a relatively large heterogeneous sample of haematological cancer patients to be recruited and increased the generalisability of the study findings. The 74% response rate is higher than that achieved by several previous studies examining unmet needs in this population [14,16,17]. However, the significantly higher rates of non-consent among younger patients, and the exclusion of patients who were not proficient in English or who were physically or psychologically unwell may have resulted in the prevalence of unmet needs being underestimated. A well-validated, multidimensional, cancer-specific measure of unmet needs which is applicable to the diversity of patients in terms of cancer type, extent of disease and phase of the cancer trajectory, was used in this study. Despite this, it is possible that the SCNS-SF34 may not encompass needs which are unique to haematological cancer patients. Gaining consensus about which unmet needs instrument to administer in other similar studies would help to more rapidly build in-depth knowledge regarding the unmet needs of this vulnerable population. Although data about extent of disease, time since diagnosis and treatments received were obtained, participants’ current treatment status was not assessed and therefore the proportion of participants who were in watchful waiting or in active treatment is unknown. A relatively limited range of individual and disease characteristics were examined in relation to moderate to high levels of unmet needs. Lifestyle, social support, coping strategies and health systems characteristics have also been found to be strong correlates of unmet need among general oncology populations [6,9,25] and should be investigated in future studies with haematological cancer patients.

Conclusion

This study found that current care fails to meet the supportive care needs of a large proportion of haematological cancer patients, particularly in relation to their psychological and physical aspects of daily living needs. Patients who are female, younger than 55 years of age and are not in the labour force appear to have the highest odds of reporting unmet supportive care needs. Within a limited knowledge base about the nature and extent of the supportive care needs of haematological cancer patients, these findings not only inform healthcare providers about the services and resources which may be required by this population, but also those aspects of care which may require improvement.

Acknowledgements

This research was supported by a Strategic Research Partnership Grant (ID CSR 11–02) from Cancer Council New South Wales to the Newcastle Cancer Control Collaborative (New–3C), and infrastructure funding from the Hunter Medical Research Institute (HMRI). Dr Allison Boyes is supported by a National Health & Medical Research Council Early Career Fellowship (APP1073317) and Cancer Institute New South Wales Early Career Fellowship (13/ECF/1–37). Dr Tara Clinton-McHarg was supported by a Leukaemia Foundation of Queensland Post-Doctoral Fellowship. The authors are grateful to Dr Heidi Turon for research support and Ms Alessandra Bisquera for statistical support.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.