Introduction: Cancer is the leading cause of nonaccidental death in childhood.(Arias et al. 2003) In spite of significant scientific progress, one quarter of children with cancer still do not survive(Greenlee et al. 2000), and over 50% experience substantial suffering that could be alleviated.(Wolfe et al. 2000a) Pediatric palliative care is now an expected standard of care and research in this field has been established as a national priority by the Institute of Medicine (IOM).(Field et al. 2003) Earlier work by our group suggested that when parents and physicians recognized at an early stage that a child had no realistic chance of cure, they were more likely to integrate elements of effective palliation into the child's care. However, this early stage concordance about prognosis was rare.(Wolfe et al. 2000b) Communication and delivery of optimal care in the advanced stages of illness is especially challenging in children.(Shelton 1999) The anticipation of losing a child and physicians' own difficulties(Saunders 1979-80; Hilden et al. 2001) may hinder discussions about poor prognosis and impact on decision-making and integration of palliative care. Thus attention to symptom management and quality of life (QoL) in children with advanced cancer may be limited.
The collection of standardized symptom and QoL data has shown to enhance physician's recognition, knowledge and accuracy in the assessment and palliation of symptoms, and to increase patient-provider discussions about QoL issues and directly improve patient's QoL.(Detmar et al. 2002; Taenzer et al. 2000; Velikova et al. 2004) Whether these findings are applicable to children remains to be established.
One of the obstacles to routine symptom and QoL assessment in children is the need of developmentally adapted(Eiser et al. 2001b; Eiser et al. 2001c; Hain 1997) and parent-proxy tools(Eiser et al. 2001a), making the paper and pencil format extremely impractical. Computerized surveys offered a potential solution. Patient acceptance of such devices has been found to be high.(Buxton et al. 1998; Velikova et al. 2002; Velikova et al. 1999; Velikova et al. 2001) Our current project -the PediQUEST Study- consists of a pilot for a randomized controlled trial (RCT) aimed at assessing the effect of routinely reporting symptom and QoL data. We will use an innovative data collection strategy: a hand-held computer devise, the Pediatric Quality of Life and Evaluation of Symptoms Technology (PediQUEST) that readily administers a user-appropriate survey instrument and generates reports summarizing how the patient feels.
As far as we are aware of, this will be the first RCT conducted in children with advanced cancer to evaluate a palliative care intervention. Given the challenges with doing research in this population, we propose to simultaneously focus on several goals.
Aims: Aim 1: (Feasibility study) Assess the feasibility of conducting a randomized controlled supportive care trial (RCT) in children with advanced cancer. Aim 2: (PediQUEST Evaluation) Preliminarily assess the effect of routinely feeding back QoL and symptom data to families and primary oncologists, and alerting supportive care teams about highly distressing symptoms. Aim 3: (Descriptive study) Prospectively explore determinants of child's suffering and parental-physician discordance regarding prognosis and treatment goals in children with advanced cancer.
Design: Aim 1: Cohort survey study; Aim 2: Pilot prospective RCT; Aim 3: Cohort study embedded in an RCT.
Setting: Three large US pediatric cancer centers. Subjects: 104 children 2-years old and above with advanced cancer and their parents were enrolled between December 2004 and June 2009.
Follow-up: Patients were initially followed up for 9 months or until death whatever occurred first. However, an interim analysis of the first 29 patients showed that 75% were alive at nine months, suggesting the enrollment of a healthier cohort. We therefore changed the follow-up period to 3 months with the option of multiple re-enrollments until the end of data collection (December 2009) or death.
Methods:
Feasibility study: All subjects approached for enrollment were invited to complete a one-time paper and pencil "Consent survey" regardless of their decision about whether to participate. All enrolled subjects were asked to complete a paper and pencil "Participation survey" after completing the study or dropping-out in order to explore determinants of attrition.
PediQUEST Evaluation and Descriptive studies: Symptom and QOL scores were tracked up to once a week. Surveys were embedded in the PediQUEST system and administered through tablet computers.
Participants' recruitment: Eligible children were identified through daily review of clinic roster with clinic staff including nurses, oncologists and/or nurse practitioners and verified with the patient's primary oncologist using a checklist sent over email. If after being introduced to the study and having had the opportunity to ask questions, child and parent(s) were willing to participate, parent(s) were asked to review and sign the informed permission/assent document covering the parent(s) and child's participation. A provider's completion of the eligibility checklist will indicate informed consent of providers.
Study Instruments:
1. PediQUEST survey (PQ-Survey): assessed symptoms, quality of life, and overall sickness. Nine survey versions were developed to allow for consistent measurement across the age ranges and respondents. Self-report started at 5 years of age; children aged 5 to 12 years answered shorter, age-adapted questionnaires which were complemented with additional parental questionnaires; teens answered a full self-report questionnaire. Parents of children between 2 and 4 years old answered a full survey on their child's behalf. If a child did not feel well, parents could complete the full questionnaire on their behalf. PQ-Surveys were pilot tested in 15 patients showing good acceptability and comprehensibility. PQ-surveys used four tools: 1) PQ-Memorial Symptom Assessment Scale (MSAS), adapted from previously validated child(Collins et al. 2000; Collins et al. 2002) and proxy(Drake et al. 2003) MSAS, measures frequency, severity and distress for 24 symptoms; 2) Pediatric Quality of Life Inventory (PedsQL), a 22-item health related quality of life instrument;(Seid et al. 1999; Varni et al. 2002; Varni et al. 2003) 3) Faces Pain Scale Revised (a faces pain scale, used to measure current pain intensity);(Hicks et al. 2001) 4) Sickness VAS scale (overall sickness measure created ad hoc).
2. Survey about Caring for Children with Cancer (SCCC): In addition, parent and physician perspectives regarding prognosis and treatment goals were collected at baseline and then every 3 months using a paper-and-pencil survey SCCC, adapted from our survey for bereaved parents.(Wolfe et al. 2000b)
3. Post death survey: If patients died within the study period parents were invited to complete a post-death survey at least four months following the death to assess distress during the last month of life and retrospective perspectives regarding the overall quality of care.
4. Satisfaction Surveys: Child and parent surveys were adapted from existing questionnaires.(Grogan et al. 1995; McCusker 1984; Weaver et al. 1993) Satisfaction with PQ technology was assessed using seven items for children eight years old and above and 11 items for parents. Surveys were embedded in the PediQUEST system and administered at fourth and eighth PediQUEST administration. A survey evaluating participant's satisfaction with the PediQUEST intervention (see below) was administered concurrently (eight items (parents), five items (children eight years old and above)) to all intervention participants. Provider satisfaction with the PQ intervention, and perceived usefulness of the PQ-system was assessed through an online survey administered at the end of data collection.
Intervention: Patients were randomly allocated to receive routine feedback about symptoms and QOL. The feedback intervention had two components: (i) PQ-reports: printed reports were made available to providers (inserted in charts before the visit) and families (handed immediately after survey completion at clinic or ward). PQ-reports consisted of bar plots of PedsQL and MSAS symptom scores from current and four prior administrations, a summary highlighting changes since last report, and a list of available resources for symptom control (for families), or generic pain management recommendations (for providers). Training on how to interpret PQ-reports was offered to families at enrollment and on an annual basis to providers; (ii) PQ emails: were automatically generated if PQ-MSAS individual symptom scores were ≥70, or if on two consecutive administrations the patient or parent reported a score of ≥70 for any frequency, severity or distress item, or a PedsQL total score ≤ 40. E-mails were sent to primary providers (oncologist, nurse, and psychosocial clinician), the local palliative care service, and, when pain was reported, the pain service (or equivalent). No instructions were imparted on how to respond to emails. Families and providers of patients in the control arm of the study did not receive feedback reports. The allocation process was embedded in PediQUEST and blinded to investigators until the patient completed the first survey. Because of the nature of the intervention this was an open label study. As a token of our appreciation, small non-monetary incentives were provided to children, parents, and providers.
Outcomes: This study explores the feasibility of conducting an RCT assessing a supportive care intervention in children with advanced cancer (aim 1). It also provides estimates of consent and attrition rates, its determinants, and the burden associated with repeated measurements. In addition, this pilot study will preliminarily assess the impact of feedback on child's distress and quality of life. Main outcomes of this efficacy study (aim 2) are presented below. This pilot study will render sufficient data for accurate sample size calculation for future studies and serve to explore determinants of suffering in children with advanced cancer (aim 3).
Statistical Analysis:
PediQUEST Evaluation (Aim 2): To assess the effect of PediQUEST, the proportion of patients with unrelieved symptoms in each arm will be compared using odds ratios and associated chi-square tests. To assess the progression of reported levels of child's distress and QOL trends, we will run a mixed effects model that assumes a linear trend in the outcome over 20 weeks of follow up for each subject. The model will examine the difference in the average trend over time (slope) between the intervention and control group by including a fixed effect term that represents the interaction of treatment status and time since study entry. Lack of difference between slope coefficients will support the null hypothesis of no effect by the intervention. Single point secondary outcomes will be analyzed by comparing both arms with a t test or Wilcoxon rank-sum test for continuous outcomes and odds ratios and associated chi-square tests for proportions.
Sample Size and Power Calculations A formal sample size estimation was not possible given the lack of data on distribution of scores in an advanced cancer population or their variation over time. For this reason we originally proposed to recruit 120 patients and their parents, a sample size driven by practical rather than statistical considerations that would serve as a basis for future sample size calculations. We estimated that 60 subjects per arm would allow for an adequate balance between arms and likely provide sufficient power to detect large clinical effects for single point measurements.
