Assessment of Factors Associated With Parental Perceptions of Voluntary Decisions About Child Participation in Leukemia Clinical Trials

Key Points Question Are health literacy, contextual factors, or sociodemographic characteristics associated with parental perception of voluntariness during informed consent for pediatric leukemia clinical trials? Findings In this cross-sectional study that included 97 parents of children with newly diagnosed leukemia, lower perception of voluntariness was significantly associated with lower health literacy. Meaning Low health literacy may have a role in parents not making complete and meaningful informed decisions for their child’s participation in cancer clinical trials.


Introduction
Leukemia is the most common childhood cancer in the US, 1 and survival rates have improved markedly in part owing to the successful recruitment of children into clinical trials. [2][3][4] It is an ethical imperative that parents and legal guardians fully understand their role in decision-making when giving permission for their children to participate in research. 5 Informed consent can be termed complete and meaningful if competence, information disclosure, comprehension, and voluntariness are effectively satisfied. 6,7 The process involves the consenting clinician verifying the participant's understanding of risks, benefits, and alternatives and ensuring patient's decision-making abilities. 3,5 Voluntariness is defined as the willingness to participate in research without feeling pressured or influenced. 8 Recruitment into pediatric cancer clinical trials often occurs under tremendous emotional stress due to the potentially fatal nature of cancer and the need to start treatment urgently, which may negatively influence voluntariness for participation in the clinical trial. 9 Health literacy (HL) is defined as the degree to which individuals are able to process health information to make appropriate health decisions. 10 In the US, at least 1 in 4 adults have limited HL skills, and low HL is associated with racial/ethnic minority status and poor health outcomes. 11,12 There is little research investigating the association of HL and other contextual factors and sociodemographic characteristics with the perception of voluntariness during informed consent for pediatric leukemia clinical trials, which comprise more than 50% of all pediatric cancer clinical trials.
To address these gaps, our study assessed the association of HL, other contextual factors (acculturation, decisional regret, and satisfaction with informed consent), and sociodemographic characteristics with the perception of voluntariness among parents of children with newly diagnosed leukemia who consented to have their child participate in a therapeutic clinical trial at our institution.
We hypothesized that parents with low HL have lower perception of voluntariness compared with parents with high HL.

Study Participants
Our cross-sectional study included parents or legal guardians of children with newly diagnosed leukemia receiving treatment at Rady Children's Hospital San Diego (RCHSD), a large tertiary pediatric cancer center in California. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies. The institutional review board for the University of California San Diego and RCHSD approved this study.
Written consent was obtained in a manner consistent with the Common Rule requirements from participants meeting the eligibility criteria and who voluntarily chose to participate. No one received compensation or was offered any incentive for participating in this study.
Children aged 0 to 17 years who were diagnosed as having leukemia between October 2014 and June 2017 were identified from RCHSD's Cancer Registry. Parents of these children were eligible to participate if (1) their child was eligible for a therapeutic clinical trial; (2) they had consented to participate in the therapeutic clinical trial within the previous 7 days; (3) they were a primary decision-maker; and (4) they had an understanding of written and spoken English or Spanish. We excluded parents who had given consent for any therapeutic clinical trial in the past, had a child with a second malignant neoplasm or relapsed disease, had a child previously diagnosed as having leukemia at an outside institution, or would potentially transfer cancer care to another institution.

Study Procedures
Our primary outcome of interest was voluntariness and its associations with HL and other contextual factors (acculturation, decisional regret, and satisfaction with informed consent) and sociodemographic characteristics. Participants completed questionnaires within a week of the informed consent discussion to assess perception of voluntariness, HL, acculturation (if Hispanic), decisional regret, satisfaction with informed consent, and sociodemographic characteristics (ie, age, sex, race/ethnicity, primary language spoken at home, educational level, marital status, and occupation). Questionnaires were available in both English and Spanish and administered by bilingual (English and Spanish) and bicultural (Anglo and Hispanic) research staff. [13][14][15][16][17][18] All survey questions were written with a low level of complexity and high readability 19 and did not contain health information that would require high HL skills to understand.
We collected data from the medical record for each patient including age, sex, and insurance type. Socioeconomic status was calculated using the Hollingshead index. 13 Race/ethnicity was selfreported by parents, and we used US Census Bureau categories. 20

Study Measures
We assessed HL using the Newest Vital Sign (NVS), a 6-item test that evaluates interpretation of information from a nutrition label, with an emphasis on using numeracy skills. 21 The NVS scores range from 0 to 6 and (with lower scores indicating limited HL), and the NVS has been validated in other disciplines 22 and used widely, including for parents who make health decisions for their children. [23][24][25] Perception of voluntariness was assessed using the Decision-making Control Instrument, a 9-item questionnaire with a Likert scale response format to statements such as "I was powerless in the face of this decision." 15 Decision-making Control Instrument scores range from 9 to 54 and (with lower scores indicating lower perception of voluntariness) and has been validated for parents who make decisions regarding research therapeutic protocols for serious childhood illnesses, including cancer. 26 Decisional regret was assessed using the Decision Regret Scale, a 5-item questionnaire with a Likert scale, 16 with scores ranging from 0 to 100 (lower scores indicating lower regret), and has been used for parents of children with cancer. 27 The scale has been validated in both adult oncology patients and their caregivers. 28 Acculturation was assessed using the Short Acculturation Scale for Hispanics, which evaluates factors of language use, media, and social relations among Hispanic individuals, 18 with scores ranging from 12 to 60 (lower scores indicating lower levels of acculturation). The NVS and the Short Acculturation Scale for Hispanics were validated in English and Spanish. 18 For instruments that have not been validated in Spanish, including the Decision-making Control Instrument, Decision Regret Scale, satisfaction and demographics questionnaires, we used the Brislin method, which has been extensively used in cross-cultural research. 29

Statistical Analysis
The primary outcome of perception of voluntariness was analyzed in association with sociodemographic characteristics using the Kruskal-Wallis and Mann-Whitney tests for robust analysis of group differences for each outcome. Spearman correlation analysis was used to determine significant associations between perception of voluntariness, HL, decisional regret, satisfaction with informed consent, and acculturation. A multivariable generalized estimating equation model was fit with perception of voluntariness as the outcome variable and HL, language, and acculturation level as significant risk factors (P < .10 from univariable models).
We analyzed HL and associated sociodemographic characteristics among study participants and conducted secondary analyses for decisional regret and satisfaction with informed consent to understand associations in related contextual factors. Significant associations were defined as P Յ .05. All analyses were conducted from May 2019 to May 2020 using statistical software program R, version 3.5.0 (R Project for Statistical Computing). 30

Participant Characteristics
In total, 177 patients aged 0 to 17 years were diagnosed as having leukemia during the study period, and 135 parents met inclusion criteria and were prospectively approached. Of these, 26 eligible parents (19%) declined, and 12 enrolled participants (9%) withdrew from the study within 3 months of enrollment. Reasons for declining or withdrawing included feeling overwhelmed and having no time to complete questionnaires. The final sample included 97 participants, consisting of 50 Hispanic parents (52%), and 47 non-Hispanic parents (48%). The majority of participants were women (65

JAMA Network Open | Ethics
Across all subgroups, mean decisional regret scores were overall low (x = 20.87; maximum score 100), and mean scores for satisfaction with informed consent were moderately high (x = 17.44; maximum score 21).

Discussion
We found that lower perception of voluntariness was significantly associated with lower HL in the multivariable analysis, after adjusting for contextual factors and sociodemographic characteristics.
This finding suggests that HL may play a role in perception of voluntary decisions during informed consent. Low HL is associated with poor health outcomes, having an uninsured child, higher hospitalization rates, and decreased access to primary care. 11 Prior studies have reported that lower perceived voluntariness is associated with lower educational and racial/ethnic minority status 26 and that individuals from racial/ethnic minority groups are less likely to understand voluntariness. 40 Our results did not show an independent association between Hispanic ethnicity and lower perception of voluntariness. Similarly, Spanish language and low acculturation in Hispanic participants were significantly associated with lower perception of voluntariness in univariable analyses, but these associations did not remain in the multivariable analysis, likely owing to dependence between Spanish language or acculturation and  low HL. It is plausible that our study was insufficiently powered to detect an independent association between Hispanic ethnicity and perception of voluntariness or, most likely, that HL may have a stronger association than sociodemographic characteristics, as shown in the multivariable analysis.
This finding supports the need for identification of individuals with low HL and the potential role of health literate interventions to improve voluntariness during informed consent for research, focused on concordant language, simplified medical jargon, and discussion of alternative treatments. 5,9,11 Limited English proficiency and low acculturation have been associated with both low HL and clinicians are often faced with the conflict of needing to initiate treatment urgently after the leukemia diagnosis vs giving parents optimal time for informed consent. 14  and have Medicaid insurance were associated with lower HL, these groups may benefit most from tailored informed consent interventions.

Limitations
Our study included a high percentage of Hispanic and Spanish-speaking individuals; however, our overall sample was relatively small, included more mothers than fathers, and had a small proportion of Asian/Pacific Islander and Black individuals. Hispanic participants were primarily from Mexico, and although Hispanic individuals of Mexican descent are the largest US Hispanic group (62%), 61 this may limit the generalizability of our findings to other Hispanic subgroups (eg, individuals from Puerto Rico or from Central or South America). Although we assessed voluntariness soon after the consent decision to mitigate confounding factors, we did not report comprehension of informed consent. We performed only quantitative measurements, limiting the depth of understanding of the complexity of informed consent, including "therapeutic misconception," 7 and associated contextual factors, such as patient-clinician communication, and other potentially influencing stressors. In addition, the significant association between HL and voluntariness could be alternatively explained by external influences between the health care system and individuals with low HL, in which imbalanced power dynamics may lead to perceptions that their decisions are "inappropriately influenced by others," even in health care encounters that do not involve informed consent for research. Moreover, we did not assess outcomes by leukemia types, we did not survey those who declined to participate in the therapeutic clinical trial, and we had a small proportion of participants withdraw from our study; therefore, we were unable to compare outcomes among subgroups, which may limit the generalizability of our results. Finally, we used the NVS instrument 17 to assess HL because of our past experience with this HL assessment. 62,63 The strengths of the NVS include that it may be administered in 3 minutes, has been validated in both English and Spanish, and is highly sensitive for detecting limited HL. 17,22 The NVS primarily tests listening, reading, and numeracy skills, which are key skills in cancer; however, it may not fully address verbal and reasoning skills, writing, and understanding of health terminology. Furthermore, the NVS has not been validated in pediatric oncology settings. Additional studies designed to compare the performance of the NVS with the Short Test of Functional Health Literacy in Adults, 64 considered by some researchers as the criterion standard instrument to assess HL in research settings, may help with the validation of the NVS as a rapid and valid assessment of HL in the pediatric oncology setting. Future studies should include a more diverse racial/ethnic minority group of participants, parents that have declined clinical trial participation, and characterization of cultural and linguistic concordance among clinicians and parents during informed consent.

Conclusions
Informed consent requires that study participation is completely voluntary and that all elements of Effective informed consent tailored to the individual's HL level may contribute to a reduction of disparities in clinical trial participation and equitable translation of discoveries and treatments to underserved groups.