Transparent reporting of recruitment and informed consent approaches in clinical trials recruiting children with minor parents in sub-Saharan Africa: a secondary analysis based on a systematic review

We searched PubMed/MEDLINE, Embase, CINAHL, and Google Scholar without any time limitation. The search strategy included the elements of IC, decision-making, CTs, minors, and SSA. We performed the first search in July 2017 and updated it in March 2019 based on a revised and improved search strategy [9]. The references of included articles were not systematically searched. However, for publications that explicitly stated that some methodology details were published in other articles, we considered secondary sources for the analysis. We did not list the secondary articles separately in the results, but included them as supplementary data accompanying primary CT publications.

Articles were included if they were publications of CTs involving children in SSA whose parents were potentially minors. Minor parents were broadly defined as adolescents between the age of 12 and the respective age of majority in each country, who are the parents of a CT child participant. If information on the parental age was not given, we considered studies in which child participants were < 5 years, because for this age group the probability of minor parents is higher [23]. We referred to CTs as prospective health-related interventions in persons [24]. Health-related interventions included drug, vaccine, diagnostic, medical device, surgical, emergency research, and dietary supplements trials. CTs had to have taken place in at least one SSA country. We only included publications in English or French.

The search results were imported into the reference management software Endnote X7. After removing duplicates, we extracted information (Author, Year, Journal/Publisher, Title, Abstract, Keywords, ISBN/ISSN, DOI, and URL) into an MS Excel table for screening. The identified articles were screened in two steps: First, two independent reviewers (ADP and DOB) screened titles and abstracts for potentially eligible articles. Second, if information about an inclusion criterion was missing, one reviewer (ADP) screened additionally full-texts. The reviewers reached a moderate (Cohen’s Kappa κ = 0.47) and substantial (Cohen’s Kappa κ = 0.71) agreement for the title and abstract screening of the initial and updated search, respectively [25]. Disagreements were mostly systematic, mainly concerning the distinction of study types, and were all resolved through discussion. One researcher (ADP) then assessed the full-texts of potentially included articles for final eligibility and extracted data. The second reviewer (DOB) crosschecked 10% of the full-text articles for eligibility and the extracted information.

Extracted data of resulting full-text articles included CT characteristics and the following elements for analysis: Study location, health condition, medical intervention, population size and type, CT design, ethics committee (EC)/institutional review board (IRB) approval, information sections addressing eligibility criteria, IC approach, and proxy decision-makers. Whenever applicable, we accessed the referenced regulatory and ethical guidance, CT registration details, supplementary files, and referenced protocols. We considered specific sections of the CONSORT statement [17], and its adaptation for children [19] to identify information concerning the recruitment and IC approach for children with minor parents. The considered CONSORT sections included the eligibility criteria, participant flow diagram (exclusions), baseline data, ethical considerations, and access to protocol and registry information. We performed a descriptive analysis using MAXQDA (VERBI GmbH) and MS Excel.

In order to assess transparency on reported proxy decision-makers, we defined three levels of transparency according to the level of detail of their description: A basic level of transparency was assigned when neither the type nor the number of proxy decision-makers was specified. A first level of extended transparency was defined when the type or number of proxy decision-makers was specified. A second level of extended transparency was attributed when also the proxy decision-makers’ age or competence was taken into account.

Table 2 summarises general reporting characteristics relating to the research design, EC/IRB approval, implementation of regulatory or ethical guidance, supplementary material (i.e. CT registrations, protocols and supplementary files), and the section addressing informed consent (IC).

Information on maternal age could potentially be found in publications among the description of the eligibility criteria, the effective exclusion reasons in study flow diagrams, and the baseline data about study participants. While in most CTs, children were directly recruited, in four CTs, recruitment was first based on the eligibility of pregnant women [31, 33, 67, 68]. One of these four publications [31] referred to a CT conducted in Malawi recruiting mothers and children explicitly as a dyad [71]. It was the only CT publication addressing minor mothers in the eligibility criteria. Mothers from 15 years of age were considered eligible, while the legal age of majority in Malawi was 18 at the time [72]. However, there is no evidence in any section of the CT publications that any mother meeting this criterion actually was recruited and participated in the CT: Neither the exclusion reason in the study flow diagram, which stated “underage”, nor the way the maternal age was described in the baseline data (i.e. only mentioning the mean age, including standard deviation without explicitly stating the minimal age of included mothers) were sufficient to allow such a confirmation. Two more CT publications described maternal age in the same way in the baseline data [40, 67], and no other CT publication provided details about parental age in the main text.

Proxy decision-makers providing IC for children’s CT participation were mentioned 77 times across the CT publications or supplementary materials (i.e., CT registration, protocol, or referenced articles further detailing the methodology) (Table 3). We found the terminology used to describe proxy decision-makers to be variable and identified four main types: “parents” (39.0%); “parents or guardians” (36.4%); “caregivers” (13.0%); and “guardians” (11.7%). Further details were provided in some CT publications or supplementary material specifying a subtype and number of proxy decision-makers. Subtypes were specified in a third (32.5%), and the number was specified in about half (50.6%) of all cases. In 40.3% of all cases, neither the subtype nor the number was specified.

The terms used to describe proxy decision-makers were sometimes inconsistent within publications (n = 20/44, 45.5%), as well as between publications and registrations (n = 16/21, 76.2%) or protocols (n = 6/6, 100%). Most inconsistencies within CT publications included a reduction from “parents or guardians” to “parental” consent, or from “legal” or “authorised” guardians to “guardians” only (n = 12/20, 60%). However, some CT publications (n = 8/20, 40%) also applied the terms interchangeably by, e.g. first using “caretaker” then “guardian”, or by switching between “parents” and “primary caregivers” or “mothers” [see Additional file 2, Table S2]. Inconsistencies between publications and registrations or protocols mostly included more or less specification of proxy decision-maker types and some included an interchangeable use of terms.

Table 4 shows the transparency levels of reported proxy decision-makers in publications and supplementary material. When considering only the CT publications, without supplementary material, and focusing on the most detailed description of proxy decision-makers within the main text of each publication, in half of all CT publications (50.0%), neither the type nor the number of proxy decision-makers was specified. In less than half CT publications (45.5%) transparency on the type of proxy decision-makers was extended, and only one publication provided a second level of extended transparency [31].

Publications of CTs recruiting children and their mothers as dyad may list maternal age among the eligibility criteria and the effective reasons for exclusion [31]. In other CT publications, however, maternal eligibility requirements were uncommon, perhaps because research focused on children independently from their mothers. Several CT publications included maternal data among the baseline data. These data contained maternal age (mean and standard deviation), years of education, or literacy levels [31, 40, 49, 60, 66‐69]. The reporting of these aspects might aim at acknowledging some confounding factors pertinent for the study results. Two of the included CT publications also discussed possible relations between maternal data and study cooperation or the generalizability of study results [40, 68], which emphasizes the importance of including information on special population groups in CT publications. Hence, standard reporting of maternal age when including small children as CT participants and the additional provision of the maternal age thresholds (i.e. minimum and maximum age) is a straightforward way for researchers to increase transparency on the inclusion of children with minor mothers who are also the primary caregiver.

In half of all CT publications, only a blanket statement was provided in the Methods section that IC was granted by four main proxy decision-maker types: “parents”, “parents or guardians”, “guardians”, or “caregivers”. The other half of the CT publications additionally specified subtypes of proxy decision-makers (e.g., mother, legal guardian, and primary caregiver) and their number (at least one, each, and one). In 11% of the CT publications (n = 5/44), CT registrations (n = 5/21, 23.8%) and corresponding protocols (n = 2/6, 33.3%) providing more transparency about the possible involvement of children with underage parents and the requirements for proxy decision-makers were referenced. Therein, consent was required from parents or guardians who were adults (or of legal age) or able (or capable) to understand study procedures and give consent [32, 33, 43, 63, 69]. However, since the corresponding CT publications contained only blanket statements that the consent was given by “parents or guardians”, the final procedure remained unclear.

Overall, the information on proxy decision-makers in CT publications was very brief and overlooked any difficulties or exceptions. This is surprising since previous literature shows that, for example, identifying suitable proxy decision-makers in paediatric CTs in LMICs can be challenging [73]. Moreover, depending on the policy, context or law applied, the meaning of the main proxy decision-maker types may vary and leave room for interpretation [see Additional file 2, Table S2]. Hence, there may be differences in whether falling into the category of “parent” or “caregiver” qualifies a person to be the proxy decision-maker, as they may still lack legal capacity. For example, in Malawi, biological fathers who are not married to a child’s mother do not automatically have direct “parental responsibility” rights in relation to the child [74]. These aspects require special attention in developing and reporting on the IC process. In the included CT publications, it remains uncertain if and how far such ethico-legal variation was considered. Confusion could also arise from applying the terms inconsistently throughout and between publications and their registrations or protocols, as found sometimes in this analysis. We, therefore, believe it is useful to establish standard operational definitions of possible proxy decision-maker types (e.g. guardians, legal guardians, legally acceptable representatives, caregivers) in international CT guidelines to facilitate a correct and coherent use of terminology.

Some of the publications referred to national CT guidelines they followed. However, again we could not find any particular statement on the case of children with minor parents. The general absence of information on children with minor parents in CT publications may suggest, first, that this case did not occur in these CTs, and, second, that researchers are not sensitised to the possibility of children having minor parents. In contrast, for example, several CT publications included additional details on other exceptional factors related to IC in LMICs, such as high illiteracy rates and providing oral consent with thumbprints, or the need for prior community consent. One publication also discussed the impact of the social and cultural background of the population on consent [57]. Third, authors of CT publications may focus more on reporting that IC was granted, regardless of how this was achieved.

Previous research showed that a detailed description of the consent process is uncommon in CT publications [75]. Consequently, the need to provide more details was debated [76]. Some critics question the practicality of adding succinct IC process descriptions, implying that it can be assumed that the adequacy of IC was evaluated and established by a competent ethics committee that has approved the CT [77]. While we agree that extensive IC descriptions in CT publications are not practical, given journals’ limited word count, it can also not be ignored that consent processes carry not only ethical but also scientific implications. IC procedures must be appropriate and tailored to the risks involved in each trial and may contribute to selection bias [78]. A description of how the IC procedure is handled for specific groups may, therefore, be useful for the overall picture of the CT.

For ethics committees to approve the appropriateness of IC procedures, these must be described in the CT protocol or reference must be made to applicable guidelines. Only publications can confirm what processes were actually applied. Publishing specific IC processes would not only help ensure the consent’s validity, but would also provide guidance for future researchers, prevent protocol deviations, and possibly highlight areas that need new guidance. Also, it can incentivise researchers to consider possible practical challenges early on, mitigating recruitment delays and consent withdrawals, and to strengthen reflections on risk-benefit analyses. We, therefore, recommend the general presentation of IC processes in CT publications and argue for the inclusion of evidence on exceptional IC cases. At a minimum, this should be described in an appendix to the publication in case of space limitations.

Ethical considerations, including consent, are currently not part of the CONSORT checklist [17]. Although the explanation and elaboration of CONSORT refers to IC and mentions that obtaining consent should be reported, it defers to journal instructions for specific ethical requirements [79]. The CONSORT adaptations for children, however, deem IC in paediatrics more complicated than for adults and propose consent related considerations to the checklist, such as reporting if assent was provided [19, 20]. This view reinforces our recommendations, as similar to paediatric CTs, consent procedures in CTs conducted in LMICs may require specific considerations meriting additional clarification in publications. In case a new CONSORT extension for paediatric CTs is developed, it should address ethical requirements and ask for an explicit description of exceptional IC situations for special population groups and of the solutions implemented.

The SPIRIT protocol development checklist already contains a requirement to specify how the IC should be obtained from participants [16]. Our review found that CT protocols that were likely developed after the implementation of SPIRIT provided increased transparency about the proxy decision-maker, emphasising the possible benefit of including consent requirements also in checklists for CT publications.

Furthermore, our study found that only a small number of publications provided information on the CT phase. However, a clear indication of the phase is useful to ascertain IC validity, as it is an indicator of research borne risks. It potentially has an impact on the operationalisation of the IC, including the selection and number of proxy decision-makers. The CONSORT explanation and elaboration paper also recognises that specifying the phase may be relevant in drug trials [79].

To our knowledge, this is the first systematic analysis of the recruitment and consent of children with minor parents in CT publications. Previous studies have shown that a small percentage of CT publications do not report the provision of consent [80, 81]. The search strategy was designed to identify articles that relate to consent and proxy decision-makers based on the information in titles and abstracts. If articles did not include such information in their titles and abstracts and were not otherwise linked to our topics of interest, they were probably omitted by the search.

In addition, CT publications that referred to other literature for more details on the research methodology, but did not specifically refer to “consent” or “permission” within their text, may have been missed. Our search strategy may have also overlooked some specific terms. For example, we searched for “adolescent pregnancy” but did not explicitly include the term “pregnant women”, which may have inadvertently excluded studies in which children of underage parents could be the focus of CTs with pregnant women.

Given these limitations, the extent of IC information provided in our sample of CT publications may not be generalizable. However, as young children in LMICs are particularly vulnerable research participants, this vulnerability may have led the respective researchers to report more cautiously than for typical research participants [80]. Therefore, it is also possible that our search strategy identified CT publications that provide above-average information on consent and the management of children with minor parents in research might be even less transparent than shown in our analysis, which underlines the importance of increasing standards in reporting.

Nevertheless, to follow-up on this research, we would recommend collecting and analysing a random sample of CT publications published from 2014 (after the CONSORT and SPIRIT guidelines were both introduced). This would also give the opportunity to extend the review to LMICs beyond SSA. Further strengths and limitations of the search strategy were previously published [9].

The authors acknowledge that directly involving collaborators based and working in research in the SSA region may have enriched the analysis and strengthen the conclusions of this work. The limitation of the study’s design is partially mitigated by the knowledge available from the researchers’ institution being involved in over 150 CTs executed in partnership and direct team discussions held with clinical research colleagues from the SSA region.

We recognize the general added value that engagement of stakeholders from the region can bring and recommend it to be included in future studies. Also, efforts to improve criteria for IC procedures, such as the CONSORT checklist, should include regional and cultural consultation, to avoid standardizing processes while overlooking their impracticality for specific situations.