Patient population
The details of the study participants and baseline characteristics are comprehensively described elsewhere [12]. Briefly, between January 1, 2012 and January 1, 2014 a cohort of 161 pediatric inpatients and outpatients between 6 and 18 years of age were enrolled, visiting hospitals in Amsterdam, Rotterdam and The Hague (the Netherlands). Inclusion criteria were eligibility for clinical research participation and speaking Dutch. The clinical research projects at offer were 10 randomized controlled trials and three observational studies at departments of allergology, oncology, pulmonology, ophthalmology and gastroenterology. The study protocol was approved by the institutional review boards at each site. Prior written informed consent was obtained from the parents or guardians of the children who served as participants, and of participants 12 years or older, and assent form participants under 12 years of age.
Methods
Children’s competence was assessed by the MacArthur Competence Assessment Tool for Clinical Research (MacCAT-CR) interview. The MacCAT-CR guides clinicians and patients through the process of information disclosure required for informed consent, combined with an assessment of the patient’s capacities, in approximately fifteen to twenty minutes. The MacCAT-CR offers a structured overview of patients’ capacities on four subscales (understanding, appreciation, reasoning, expressing a choice) to base a competence judgment on. The Dutch version of MacCAT-CR was modified for children by using simple language to be understood by children of elementary school age, and adding questions on the influence of social relationships [2]. The version used will be available after arrangement of proprietary issues. We demonstrated that children’s competence to consent to clinical research can reliably and validly be assessed by using the MacCAT-CR [12].
A MacCAT-CR competent classification was considered present when at least two out of three of the experts rated the MacCAT-CR interview positive for competence, in other cases patients were classified incompetent. Additional patient data were collected, demographic characteristics included ethnicity. Number of trials previously participated in and duration of disease were measures used to express disease experience. Disease experience was arbitrarily categorized as low (no prior trial participation and duration of disease less than one month), moderate (no prior trial participation and duration of disease more or equal to one month) or high (prior trial experience). The level of education of the highest educated parent served as an indicator of SES, which we categorized: low (no primary school, primary school, special primary school, special secondary school); middle (preparatory secondary vocational education, secondary vocational education, senior general secondary school, preparatory scientific education); high (college, university). Complexity of the decision was categorized into subgroups by consensus between three researchers (LG, IH, PT): low (open trial or randomized trial without blinding), or high (randomized trial with either the use of placebo, or blinding, or both). Risk was categorized using the same manner: low (no risk), moderate (little risk) or high (possible risk). Cognitive capacities were expressed as intelligence quotient (IQ) and assessed by the Wechsler Nonverbal Scale of Ability short version (WNV). The WNV was administered by trained certified professionals (special education or psychology graduates) under supervision of a senior professional. Scores on the WNV could be categorized into three IQ categories: low (under 90), average (90–110) or high (110 or higher). Ethnicity was classified as Western European, Middle East, Surinam/Antillean or other.
Parent(s) were asked if they judged their child had the capacities to make a well-considered decision on giving informed consent, in other words, if they considered their child decision-making competent. We classified if the child decided to participate in the research project at offer or not, or if he/she had not decided yet.
Data analysis
Effects of the child variables and contextual variables on a competence judgment were expressed with odds ratios (OR) and their 95 % confidence intervals obtained by simple logistic regression. ORs > 1 indicate higher odds of a competent judgment when a characteristic is present, ORs <1 a lower odds. We considered the following child variables that may “cause” a competent classification: age, gender, intelligence, disease experience, SES, ethnicity. Contextual variables considered were complexity, risk, parental competence judgment and decision to participate in a study. First, we entered all child variables simultaneously into a multiple logistic model to examine their association with a competent judgment as expressed by the Wald-test statistic and associated p-value. Then, we entered the child characteristics one by one into a new logistic regression model, the variable with the largest Wald statistic from the full model first, then the variable with the second largest Wald statistic and so on. To evaluate the independent contribution to the total explained variance in competent classifications for a child variable, we examined the increase in Nagelkerke R-square explained variance after entering a variable into the model. The influence of the contextual variables, was examined by the extra increase in R-square explained variance after adding them to the model that already included the child characteristics.