Although the National Commission for the Protection of Human Subjects (1) defines assent as a child's affirmative agreement to participate in a research study, there remains a wide variation in the interpretation and requirements for assent by IRBs (2). Under the National Commission's definition, assent is limited to a preference for participation, and thus does not require the same level of understanding as parental permission (3). Nevertheless, in many cases, it would seem important that in giving their assent, child subjects have some understanding of what will happen to them and the risks and benefits of participation. Several studies, however, suggest that many children have poor understanding of both observational and randomized controlled trials (4–6). In one study, only 50.5% and 41.9% of children had complete understanding of the risks and the protocol, respectively (6), and in another, 36% of children in a trial were unable recall any aspects of the study information (4).
Although there is no consensus regarding the use of a separate assent document for research (2), many IRBs require their use when presenting study information to children; therefore, it is important that, if required, assent forms be written and presented in a manner that optimizes understanding. Unfortunately, consent and assent forms are often written such that many adults and children have difficulty in understanding even the basic elements of a study. Although various strategies have been used to enhance research subjects' understanding of study information, none have proven universally effective (7). Furthermore, there are very little data that address children's understanding of research information. This study was designed, therefore, to examine the effect of improved readability and processability (a function of format and layout) on children's understanding of written study information. In addition, this study sought to identify characteristics that predict understanding among child research participants.
The University of Michigan's IRB approved this study and a waiver of documentation of informed consent and assent was granted. The study population included pediatric patients aged 7–17 yr who had been hospitalized after an elective surgical procedure (ASA I–III) or medical condition. Children who could not read, who were cognitively impaired, or those with emergent or serious illnesses were excluded. Parents and children were approached for consent and assent in the child's inpatient room after surgery or treatment. Each was told that we were examining ways in which to improve the manner in which study information is presented to children. The children were told that they would be required to read some information about a study and then answer some questions regarding the material.
Children were randomized to receive information about a study of postoperative nausea and vomiting (PONV) formerly conducted in our department, in one of two ways: 1) standard informational form; 2) standard form modified by a reading expert to reflect the average reading ability of this cohort of children and to conform to the federal literacy guidelines for readability and processability, i.e., seventh grade reading level and a processability score of >60 of 100 (8). Processability was measured using the format described by Irwin and Davis and modified by Philipson et al. for use with consent documents (9,10). Processability scores are based on the presence or absence of criteria for the five components of comprehension, i.e., integrative processing, microprocessing, metacognitive processing, elaborative processing, and macroprocessing. A processability score of >60 of 100 is considered acceptable (8).
The methods used to modify the standard form including the use of bullets, bolding, increased font size, and pictures have been described in detail elsewhere (11). The original standard form had been written at above the eighth grade level and with a processability score of 52 of 100. With the help of a reading expert, the standard form was rewritten to reflect a seventh grade reading level and a processability score of 70 of 100. Although we acknowledge that this reading level was above that of some of the children, particularly the younger ones, we chose seventh grade as the average reading level of this cohort of children. One should note, however, that despite simplification of the standard form, care was taken to retain all the important elements. An excerpt of the modified form is shown in Figure 1.
In an attempt to measure understanding in the context of the “process” of assent, children were also given the same information verbally using a standard script in age-appropriate language. Children were informed that they would not be participating in the actual PONV study but that they should consider the information as if the study were real.
Measurement of Child Understanding
Children were given as much time as they needed to read the information and were then interviewed to determine their understanding of eight informational elements. These included the purpose of the study, the protocol, risks, benefits to themselves (direct) and to others (indirect), the alternatives, voluntariness, and freedom to withdraw. The interview was presented in a semistructured fashion and the children's open-ended responses were transcribed verbatim by trained research assistants. Interviewers were allowed to prompt the children for further information but were not allowed to guide their responses. The children's levels of understanding of each of the informational elements were scored independently by two assessors who were knowledgeable about the PONV study but blinded to the child's group assignment. Further details of the interview process and method of scoring are described elsewhere (12). Scores of 10, 5, or 0 were assigned to responses indicating complete, partial, or no understanding of each element, respectively. In addition to the assessors' measures of understanding, children were asked to rate their own level of understanding using a 0–10 numbers scale (where 10 = complete understanding).
Measurement of Child Reading Ability
In addition to the questions related to understanding, children were asked their perceptions of the amount, clarity, and understanding of the information using Likert type scales. Demographic information was obtained from both parents and children. At the end, children were shown both forms and asked which they preferred and why. Children also completed the shortened version of the Slosson Oral Reading Test (SORT-R3) (13). This test has been validated for use in children 4-yr-of-age through adulthood and takes 3–5 min to complete (14,15).
Statistical analyses were performed using SPSS® statistical software (SPSS, Chicago, IL). Sample size determination was based on a previous study that showed that children's overall understanding of a standard informational form was 5.3 ± 2.7 [0–10 scale (where 10 indicates complete understanding)] (12). Accepting a 20% improvement in overall understanding as clinically important, we required a sample size of 95 children per group (α = 0.05, β = 0.20, two-sided). Comparisons of parametric data between groups were analyzed using unpaired t-tests. Nonparametric data were analyzed using Mann-Whitney U, χ2, and Fisher exact test, as appropriate. When appropriate, the Bonferroni correction was used for multiple comparisons. Correlations between ordinal variables were performed using Spearman's rho (ρ) correlation coefficient. In comparing the assessors' scores of understanding (Likert scale) with the children's perceived understanding (0–10 numbers scale), z scores were first generated to control for the different metrics.
Factors that were found to be significantly associated with understanding by univariate analysis were entered into a multiple regression model. Inter-rater reliability between the assessors' scores were performed using Cohen's kappa (κ) statistic. κ values of ≥0.4 were considered to represent acceptable levels of agreement and correlation. For age comparison purposes we divided children into “younger” (7–10 yr) and “older” (11–18 yr) age groups. This age cut-off was based on a previous study that suggested that understanding of study information improved substantially after the age of 10 yr (6).
Three-hundred-ninety-two children were approached to participate in this study, of whom 202 either declined participation or did not meet the study's inclusion criteria (Fig. 2). Data are thus presented for 190 children (95 per group). The average age of those who declined was 11 yr and 54% were male. Eighty-three percent of those who declined were white, 10.9% were African American and 2.5% were Hispanic. These demographics were very similar to those of the final study sample.
There were no differences in the demographics of the children between the two groups. Furthermore, there were no demographic differences between these groups by age (Table 1). The average grade reading level of children in the study was seventh grade (range, grades 1–12.5) and was similar between the groups. The average grade reading level of the younger and older age group children was fifth (range, grades 1–12.5) and ninth grade (range, grade 2–12.5), respectively. Reading ability was significantly correlated with understanding (ρ = 0.50, P < 0.001). There was excellent inter-rater reliability for scores of understanding between the two assessors. κ Statistics for each of the elements ranged from 0.90 to 0.98 (P < 0.001).
There was a trend towards greater understanding of all the informational elements among those who received the modified form (Table 2). Specifically, children who were assigned to the modified form had greater overall understanding of the study and a greater understanding of the protocol and benefits to others.
Not surprisingly, the older children (11–17 yr) had, overall, greater understanding than the younger children (7–10 yr). However, the differences between age groups were far more marked among children who had received the standard form (Table 3). Overall, the older children's perceptions of their understanding were significantly greater than the younger children (7.6 ± 2.0 vs 5.9 ± 2.5, P < 0.001); however, both groups over-estimated their understanding when compared to the assessors' measured levels of understanding [older 7.6 ± 2.1 vs 7.2 ± 1.9 (P < 0.05), younger 5.9 ± 2.5 vs 5.2 ± 2.4, (P < 0.05)].
Twenty-one percent and 13.7% of children who received the standard form reported that it contained “too much” information and was difficult to read compared to 9.5% and 5.3% in the modified group, respectively (P < 0.05). This perception was particularly evident among the younger age group. For example, 30.6% and 34.3% of younger children reported that the standard form was difficult to read and had “too much” information compared with 3.4% (P < 0.001) and 13.6% (P < 0.05) among the older group, respectively. When shown both forms, 81.3% of children chose the modified form. Preference for a particular form was independent of reading ability, understanding, and demographics, although more girls than boys tended to prefer the modified form (86.5% vs 75.3%, P = 0.052). Children who preferred the modified form stated that it was “easier to read,” “friendlier,” and “more appealing.” These children also liked the larger font and use of pictures. Some of the children who preferred the standard form found the pictures distracting whereas others thought that it was “more professional” and contained more information.
A number of factors were found by univariate analysis to be associated with improved understanding. These factors included older age (ρ = 0.39, P < 0.001), superior reading ability (ρ = 0.50, P < 0.001), the modified form (ρ = 0.20, P < 0.01), and low anxiety (ρ = 0.15, P < 0.05). Improved understanding was also associated with children who perceived that the amount of information was “just right” (ρ = 0.19, P < 0.05) and “easy to read” (ρ = 0.42, P < 0.001), together with those who had read the form “completely” (ρ = 0.24, P < 0.01) and believed that they had “enough time” to read the form (ρ = 0.22, P < 0.01). These factors were subsequently entered into a multiple regression model with stepwise selection. Multivariate analysis of these factors yielded several independent predictors of understanding (Table 4).
Given that the basis of assent in children is simply to establish a preference for participation in a research study, the informational requirements are less defined and regulated than for parental permission (3,16). Despite this, it would appear important that children have some understanding of what will be done to them, the potential risks and benefits, and the voluntary nature of their participation. Results from this study, however, support the findings of others that many children have poor understanding of study information (4–6,17). This appears, in part, due to the fact that study information is often written and presented with little consideration for the literacy, cognitive abilities, and preferences of children.
Although several studies have examined methods to improve the understanding of written treatment and research information, none has proven universally effective (7). The techniques used have included modifying layout, simplifying vocabulary, highlighting important points, increasing font size, and incorporating illustrations or multimedia effects (7,10,18–23). Many of these studies, however, have focused on adult populations, and there are very little data addressing interventions to improve children's understanding of research information.
Previously, we showed that parents' understanding of consent information was increased by improving the readability and processability of the consent document (11). Consistent with the present study, parents randomized to receive a modified form also had significantly greater overall understanding of the material and greater understanding of the protocol, study duration, risks, and direct benefits. Similarly, a large majority of parents preferred a modified form (81.2%). Results from the present study support these findings and reinforce the importance of improving both the readability and processability of written study information.
Although child assent cannot be equated with parental consent, it nevertheless serves to remind us that children deserve to be treated with dignity and respect (24,25). Furthermore, allowing children to evidence a preference for research participation may play an important role in their development as autonomous individuals and in enhancing self-esteem (25,26). Although there is no mandate for the age of assent, recommendations by the Committee for Bioethics suggest that assent be sought from children over the age of 7 yr (1). Wendler and Shah, on the other hand, believe that assent should only be sought from children over the age of 14 yr, arguing that by this age children can appreciate the concept of altruism and are capable of making autonomous decisions (27).
Although the older children in this study had greater overall understanding of the information than the younger ones, the observation that the modified form appeared to close the gap in understanding between the younger and older children is extremely compelling. This suggests that use of a modified format will be important in providing younger children with developmentally appropriate information that can enhance their decision-making abilities.
A few points regarding the limitations of the study design are warranted. Despite simplification, the modified form was still written at a reading level above that of some children, particularly the younger ones. Thus, one would naturally expect those with poor reading skills to have poorer understanding. We should note, however, that several of the younger children were able to read at the 12th grade level and, conversely, some of the older children had very poor reading skills. This observation reinforces the concept that age as a sole criterion for assent may be discriminatory in that it does not consider the child's maturity or reading ability. Therefore, unless mandated by an IRB, the decision regarding the ability of children to provide assent should be individualized based on a number of factors including the parents' perception of their child's ability to provide meaningful assent, the child's reading ability, and their overall maturity.
On the basis of the wide range of reading abilities of school-aged children, we attempted to prepare a document that conformed to the average reading ability of this age-cohort. Although this approach may not be optimal, it is somewhat impractical in the clinical setting to provide multiple assent forms written to satisfy all ages and/or reading abilities. Thus, although we fully acknowledge that further refinements of the standard form may have produced even greater improvements in understanding, we were, nevertheless, able to demonstrate that a single assent form written with improved readability and processability can result in improved understanding, regardless of age or reading ability. Future studies designed to refine this process and further improve children's understanding of study information are, therefore, needed.
Another potential limitation was that although we measured children's understanding of a real study, it was, nevertheless, presented in a simulated format. This approach obviated any concerns regarding the ethics of recruiting children to a study using an untested form. Although one could argue that simulated methodology may not fully reflect the real life situation, several studies have confirmed the ability of simulated studies to approximate real behaviors (27–29).
Given that a large number of children declined participation, there is a potential for selection bias. For example, approximately half of the decliners reported that they were simply not interested in participating and, thus, one might be concerned that only the more motivated children were recruited. Although this may simply reflect the vagaries of research involving children for a study that offered no apparent benefit, we should also note that the demographics of the children in this study were almost identical to those of a previous study of child assent that had a higher participation rate (>75%) (6). We should also note that although only 9% of children declined due to acknowledged poor reading skills, others may have given other reasons rather than admit that they had reading difficulties. Thus, although the demographics of the children who declined were very similar to those of the study sample, the potential for a selection bias cannot be discounted.
The sample size for this study was calculated based on an expected difference in understanding between children receiving the standard form and those receiving the modified form; therefore, the sample may not have been sufficiently powered to detect all univariate and multivariate comparisons, and, thus, the potential for a β error cannot be discounted. Finally, we should note that the standard form used in this study represented one study at one institution and, as such, may not be generalizable to all situations.
Results of this study demonstrate that improving the readability and processability of written assent information increases the understanding and acceptance of the material by adolescents and young children. The observation that the modified form appeared to close the gap in understanding between the younger and older children is extremely important and suggests a simple strategy to enhance understanding among these vulnerable research participants. Although more studies are needed, these data may be important in suggesting a more child-friendly approach that takes into consideration children's reading and health literacy, learning abilities, and preferences.
The authors are indebted to Sandra J. Philipson, MA, CAGS, for her expertise in creating the modified informational form. We also wish to thank Catherine Burger, BS, Yasmin Elsayed, Heather Bowman, and Christopher Joseph, BA, for help with the study.
1. National Commission for the Protection of Human Subjects in Biomedical and Biobehavioral Research. Research involving children: report and recommendations. DHEW publication 77-0004. Washington, DC: US Government Printing Office, 1977
2. Kimberly M, Hoehn K, Feudtner C, Nelson RM, Schreiner M. Variation in standards of research compensation and child assent practices: a comparison of 69 institutional review board-approved informed permission and assent forms for 3 multicenter pediatric clinical trials. Pediatrics 2006;117: 1706–11
3. Erb T, Schulman S, Sugarman J. Permission and assent for clinical research in pediatric anesthesia. Anesth Analg 2002;94:1155–60
4. Postlethwaite R, Reynolds JM, Wood A, Evans J, Lewis M, Eminson D. Recruiting patients to clinical trials: lessons from studies of growth hormone treatment in renal failure. Arch Dis Child 1995;73:30–5
5. Susman E, Dorn L, Fletcher J. Participation in biomedical research: the consent process as viewed by children, adolescents, young adults, and physicians. J Pediatr 1992;121:547–52
6. Tait AR, Voepel-Lewis T, Malviya S. Do they understand? II. Assent of children participating in clinical anesthesia and surgical research. Anesthesiology 2003;98:609–14
7. Flory J, Emanuel E. Interventions to improve research participants' understanding in informed consent for research. JAMA 2004;202:1593–601
8. Kirsch I, Jungelblut A, Jenkins L, Kalstad A. Adult literacy in America: a first look at the results of the adult national adult literacy survey Washington, DC: National Center for Education Statistics, 1993
9. Irwin J, Davis C. Assessing readability: a checklist approach. J Reading 1980;24:124–30
10. Philipson SJ, Doyle MA, Gabram SG, Nightingale C, Philipson EH. Informed consent for research: a study to evaluate readability and processability to effect change. J Investig Med 1995;43: 459–67
11. Tait AR, Voepel-Lewis T, Malviya S, Philipson S. Improving the readability and processability of a pediatric informed consent document: effects on parents' understanding. Arch Pediatr Adolesc Med 2005;159:347–52
12. Tait AR, Voepel-Lewis T, Malviya S. Do they understand? I. Parental consent for children participating in clinical anesthesia and surgery research. Anesthesiology 2003;98:603–8
13. Slosson R. Slosson oral reading test-revised. East Aurora, NY: Slosson Educational Publications, 1990
14. Davis T, Long S, Jackson R, Mayeaux EJ, George RB, Murphy PW, Crouch MA. Rapid estimate of adult literacy in medicine: a shortened screening instrument. Fam Med 1993;25:391–5
15. Jastak S, Wilkinson G. Wide range achievement test-revisited. Wilmington, Del: Jastak Associates, 1987
16. Denham E, Nelson R. Self-determination is not an appropriate model for understanding parental permission and child assent. Anesth Analg 2002;94:1049–52
17. Kodish E, Pentz R, Noll R, Ruccione K, Buckley J, Lange B. Informed consent in the Children's Cancer Group: results of preliminary research. Cancer 1998;82:2467–81
18. Meade CD, Howser DM. Consent forms: how to determine and improve their readability. Oncol Nurs Forum 1992;19: 1523–8
19. Tarnowski KJ, Allen DM, Mayhall C, Kelly PA. Readability of pediatric biomedical research informed consent forms. Pediatrics 1990;85:58–62
20. Silva M, Sorrell J. Enhancing comprehension of information for informed consent: a review of empirical research. IRB 1988;10:1–5
21. Hochhauser M. The informed consent form: document development and evaluation. Drug Inf J 2000;34:1309–17
22. Coyne C, Ronghui X, Raich P, Plomer K, Dignan M, Wenzel L, Fairclough D, Habermann T, Schnell L, Cella D. Randomized, controlled trial of an easy-to-read informed consent statement for clinical trial participation: a study of the Eastern Cooperative Oncology Group. J Clin Oncol 2003;21:836–42
23. Campbell F, Goldman B, Boccia M, Skinner M. The effect of format modifications and reading comprehension on recall of informed consent information by low-income parents: a comparison of print, video, and computer-based presentations. Patient Educ Couns 2004;53:205–16
24. Bartholome W. Hearing children's voices. Bioethics Forum 1995;11:3–6
25. Diekema D. Taking children seriously: what's so important about assent? Am J Bioeth 2003;3:25–6
26. Strong C. Respecting the health care decision-making capacity of minors. Bioethics Forum 1995;11:7–12
27. Drapkin R, Wing R. Responses to hypothetical high risk situations: do they predict weight loss in a behavioral treatment program or the context of dietary lapses. Health Psychol 1995;14:427–34
28. Robinson M, Clore G. Simulation, scenarios, and emotional appraisal: testing the convergence of real and imagined reactions to emotional stimuli. Pers Soc Psychol Bull 2001;27:1520–32
29. Jago A, Vroom V. Predicting leader behavior from a measure of behavioral intent. Acad Manage J 1978;21:715–21