Abstract
Background The long-term impact of childhood cancer treatment on dietary intake is likely to be complex and the length of time dietary behaviours are affected after childhood cancer treatment is unknown. Aim The aim of this study was to determine the diet quality in childhood cancer survivors recently off treatment and identify possible contributing factors that may affect diet quality in this population. MethodsParticipants were 65 parents and/or carers of childhood cancer survivors (CCS) (aged 2-18 years), recently off treatment and 81 age-matched controls. Methods Participants completed two self-administered dietary intake and eating behaviour questionnaires. Study data was explored to determine between group differences, bivariate analysis using Spearman’s correlations was used to determine the relationship between diet quality and identified variables, and hierarchical cluster analysis was completed to characterise specific variables into clusters.Results CCS had a significantly poorer diet quality score than the age-matched controls (t=-2.226, p=0.028). Childhood cancer survivors had significantly higher parent-reported rates of ‘picky eating’ behaviour than the control group (t=0.106 p=0.044). Factors such as picky eating, emotional overeating and Body Mass Index z-score appeared to drive diet quality in survivors. Conclusions A CCS with picky eating behaviours could avoid complete food groups, have strong food preferences/aversions and over- consume high energy foods to maintain their energy intake, possibly affecting diet quality. The outcomes highlighted the need for a tailored intervention aimed at improving healthy eating behaviours in CCS after treatment for cancer.
Background
Many children with cancer are diagnosed in the first 10 years of life(1). This developmental period includes several key phases of eating and cognitive development which contribute to food preferences, diet intake, energy intake and eating patterns as children grow older (2). The development of food preferences occurs through the interaction of foods, sensory cues (olfactory, tactile and visual sensory systems) and the consequence of ingestion, which may elicit a positive or negative response to the food(2).
Eating is an essential human activity, yet evidence from the general paediatric population suggests that up to 40% of children aged 3-6 years old experience mealtime difficulties(3) and 25% have feeding disorders, which can affect a child’s diet quality(4). Feeding difficulties in young children can range from chronic feeding disorders to more common behaviours such as food refusal, food aversion, excessive mealtime length, inappropriate behaviours and picky eating(4). A child with picky eating behaviours can avoid complete food groups, have strong food preferences/aversions, over-consume high energy foods to maintain energy intake and demonstrate an unwillingness to try new foods(3).
As in the general paediatric population, children receiving cancer treatment can also have feeding difficulties and poor diet quality during treatment (5), with the consequences having the potential to be magnified in this population. Weight loss during treatment is associated with poorer clinical outcomes such as impaired immune competence, increased susceptibility to infection, reduced treatment tolerance leading to dose reductions and treatment delays, increased side-effects, decreased wound healing and reduced quality of life (6-9). Despite the impact of cancer treatment toxicities, which can result in the child feeling nauseated and unwell, parents often pressure their child to eat during this stressful time to maintain weight during childhood cancer treatment(10). Parent pressure, coupled with food becoming associated with treatment side effects of vomiting/nausea and taste/smell changes can lead to learned food aversions during treatment, restricting adequate intake and compromising diet quality(10, 11). During cancer treatment, young children are unable to differentiate between the impact of the disease, treatment and side effects on their eating, placing patients at a high risk of ongoing adverse eating behaviours(12).
With significant therapeutic success and a resulting increase in survival outcomes, the medical challenge has become the minimisation of the secondary effects of cancer treatments (13). One of the most significant findings of childhood cancer survivors (CCS) is the increased risk of developing metabolic syndrome (MetS) and cardiovascular disease (CVD)(1 4). Poor dietary intake is associated with the development and progression of MetS in the general population as well as adult survivors of childhood cancer (14). What is not known are the long-term implications that interrupted feeding development and changes in eating behaviour during cancer treatment has on dietary intake in childhood cancer survivors and the longer-term impact on health including MetS.
When looking specifically at macro- and micronutrient intake, Cohen et al (2015) found that children who had recently completed cancer treatment had an inadequate micro-nutrient intake and were consuming an excessive energy intake(15). An increase in the intake of high fat, high sugar and refined convenience foods has also been found for cancer survivors (15-17). The impact of childhood cancer treatment on dietary intake in young childhood cancer survivors of is likely to be complex. The combination of treatment toxicities (nausea, food aversions and taste and smell changes) with maladaptive parent eating behaviours all potentially influence dietary intake during and after cancer treatment. While recent research shows that adult survivors of childhood cancer are displaying frequent cravings for junk food (18) little is known about the eating behaviours of young survivors of childhood cancer. It is also unknown how the diet quality of young childhood cancer survivors compares with the diet of the general paediatric population.
The primary aim of this study was to compare diet quality of childhood cancer survivors with age-matched children from the general community. Secondly, the study aimed to identify possible contributing factors that may affect diet quality in this population.
Methods
Participants
Participants were parents and/or carers of CCS aged between 2-18 years, who had received anti-cancer treatment at Kids Cancer Centre (KCC), Sydney Children’s Hospital (SCH), and John Hunter Children’s Hospital (JHCH), Newcastle, Australia. Parents and/or carers were eligible if they were the mother, father or primary caregiver of a child who: 1) had undergone treatment for any type of childhood cancer; 2) had completed their cancer treatment protocol; 3) were less than five years’ post-diagnosis; and 4) had no cognitive or other mental impairments. Control group participants were parents and/or carers of a child aged between 2-18 years with no co-morbidities that affected nutritional intake. Participants were recruited between September 2012 and March 2014. The study protocol had full ethical approval from SCHN HREC (12/SCHN/29) and UNSW HREC (HC12659).
Recruitment
Study information packages were posted to eligible participants. To recruit parents of the control group, staff of participating childcare centres, community groups and volunteer community members were contacted to identify eligible participants and to ensure that: a) it was appropriate to approach each family identified; and b) they met the study eligibility criteria. Potential participants were then provided with a study package. Parents returned their completed questionnaires and a signed consent form in the supplied reply-paid envelope.
Data collection
Parents/carers were asked to complete a separate self-report questionnaire for their child’s demographic, treatment (if applicable) and eating behaviour information followed by a dietary intake questionnaire. The first questionnaire included information about the child’s geographic location (i.e. postcode of residence) and demographic data (i.e. gender, date of birth, weight and height, and current medications). Clinical information included cancer diagnosis, date of diagnosis, date of completion of cancer treatment, cancer treatment regimen (surgery, radiation, chemotherapy and hematopoietic stem cell transplant) and nutrition interventions (education, oral nutrition support (ONS), Enteral Nutrition (EN), Total Parenteral Nutrition (TPN) received during cancer treatment. Parent demographic information collected included: socioeconomic status, gender, date of birth, employment status, education and perceived health status, weight and height.
The child’s eating behaviour was measured using the validated parent report Children’s Eating Behaviour Questionnaire (CEBQ), which assesses variation in eating style between children. The measure assesses the child’s eating style though 35 items, comprising eight subscales(19) . An objective measure of the child’s food pickiness was determined using the Child Feeding Questionnaire (CFQ), ‘picky eating’ subscale. The CFQ is a validated seven factor parent self-report measure designed to assess parent attitudes, beliefs and practices regarding their child’s eating behaviour(20).
The second questionnaire was the Australian Child and Adolescent Eating Survey (ACAES) which assessed the child’s dietary intake. The ACAES is a validated parent reported 120 question self-administered semi-quantitative Food Frequency Questionnaire (FFQ) that requests specific information about foods and beverages consumed by the child to assess their dietary intake(21).
Analysis
Child and parent weight and height were used to calculate body mass index (BMI), using the formula: weight in kilograms divided by height in metres squared. For each child aged > 2 years, a BMI z-score was calculated using Epi Info™ (Version 3.5.1, 2008; Centres for Disease Control and Prevention, USA). This allowed for equal comparison across all age groups (22). Weight and height were also used to determine the weight range category of the child (thinness, healthy, overweight and obese) to define the child’s nutritional status(37). The dietary intake, proportion of energy intake and serve size data were calculated using the ACAES(21).
The Australian Child and Adolescent Recommended Food Score (ACARFS) was generated from the ACAES to determine each child’s diet quality. This score reflected the relationship between the child’s dietary diversity and nutrient adequacy by measuring their adherence to the Australian dietary guidelines (24).
Study data were explored using univariate data analysis from the statistical software package SPSS (version 22, 2014; SPSS Inc, Chicago, IL, USA) to assess whether there were significant differences between the survivor and control groups. Chi-squared and t-tests (two-tailed, α = 0.05) were used to compare parent/child demographics, BMI z-scores and Socio-Economic Indexes for Areas (SEIFA) index. Age and sex adjusted means of the ACARFS were compared using analysis of covariance to determine the difference between age and gender for diet quality scores. Differences between the parent/child eating behaviours sub-scales determined in the CEBQ and CFQ were compared using t-tests.
Bivariate analysis using Spearman’s correlations was used to determine the relative contribution of child treatment regimen side effects, parent feeding behaviour and child eating behaviour on the child’s diet quality. Differences and associations were considered significant at p < .05 (2-tailed) for all tests.
For the quantitative data variables, hierarchical cluster analysis was completed using the software ‘R’ (version 3.0.1; Packages: ’mclust’ version 4.3; ’FactoMineR’ version 1.27). This analysis grouped participants characterised by specific variables into clusters based on the distance between each observation or variable. In this study, four composite scores derived from the questionnaires (BMI z score, child picky eating behaviour, child emotional overeating behaviour and diet quality) were used after scaling and a logit transformation.
Results
Demographics
A total of 146 parents of children aged 2-18 years completed the study: 65 participants were parents of CCS (response rate 26%) and the control group of 81 participants were parents of children without a cancer diagnosis (response rate 64%).
Significant differences between control and cancer groups were for parent age, education level, and employment, with control parents working more hours per week. There was no significant difference in BMI z-scores for participating children, although the CCS had a higher proportion of participants classified as being overweight (21% vs 6%) and having grade 1 thinness (12% vs 2%) (TABLE 1).
In the CCS group, mean age at diagnosis was 6.2 (±4.5) years, mean time since diagnosis was 50.3 (±15.6) months and mean time since treatment completion was 24.9 (±17.3) months. CCS had a heterogeneous representation of cancer diagnoses, with the most common being Acute Lymphoblastic Leukaemia (ALL) 29% (n=19) (TABLE 2).