|Home | About | Journals | Submit | Contact Us | Français|
To conduct an expert-led process for identifying research priorities for eight areas of adolescent health in low- and middle-income countries. Specific adolescent health areas included communicable diseases prevention and management, injuries and violence, mental health, noncommunicable diseases management, nutrition, physical activity, substance use, and health policy.
We used a modified version of the Child Health and Nutrition Research Initiative methodology for reaching consensus on research priorities. In a three phase process, we (1) identified research and program experts with wide-ranging backgrounds and experiences from all geographic regions through systematic searches and key informants; (2) invited these experts to propose research questions related to descriptive epidemiology, interventions (discovery, development/testing, and delivery/implementation), and health policy/systems; and (3) asked the experts to prioritize the research questions based on five criteria: clarity, answerability, importance or impact, implementation, and equity.
A total of 142 experts submitted 512 questions which were edited and reduced to 303 for scoring. Overall, the types of the top 10 research questions in each of the eight health areas included descriptive epidemiology (26%), interventions: discovery (11%), development/testing (25%), delivery (33%), and policy, health and social systems (5%). Across health areas, the top questions highlighted integration of health services, vulnerable populations, and different health platforms (such as primary care, schools, families/parents, and interactive media).
Priority questions have been identified for research in eight key areas of adolescent health in low- and middle-income countries. These expert-generated questions may be used by donors, program managers, and researchers to prioritize and stimulate research in adolescent health.
The Department of Maternal, Newborn, Child, and Adolescent Health of the World Health Organization (WHO) conducted an exercise to establish global research priorities for adolescent health in low- and middle-income countries through 2030, building on earlier work that proposed research priorities in adolescent sexual and reproductive health.
In 2014, there were 1.2 billion adolescents aged 10–19 years old, comprising 16.4% of the world's population. Adolescent mortality was estimated at 1.3 million in 2012, with the leading global causes of death being road injury, human immunodeficiency virus (HIV), suicide, lower respiratory infections, and interpersonal violence . The great majority of the world's adolescents live in low- and middle-income countries (LMICs) , and 97% of deaths among young people occur in LMICs . In the past 50 years, reductions in early child mortality have been greater than declines in adolescent mortality .
In terms of the global burden of diseases, the top three causes of disability-adjusted life years lost among adolescents are unipolar depressive disorders, road injury, and iron-deficiency anemia . However, mortality and disability-adjusted life year data will underestimate the potential disease burden among adolescents because they do not reflect conditions and behaviors that can lead to future disability and mortality later in life, such as tobacco use and dependence or physical inactivity . Health-related risk behaviors adopted or consolidated during adolescence may not always affect the adolescent's health during the second decade of life but will have a substantial effect later in life, and some will affect the health of future generations , .
Improving the health of adolescents in LMICs will be essential for the world to achieve the United Nations Sustainable Development goals , and the specific targets and goals included in the United Nations Secretary General's Global Strategy for Women's, Children's and Adolescents' Health . Although there has been an increased call for research on the health and wellbeing of adolescents and young people to guide these and other global and national initiatives, research from LMICs is still limited , .
Here, we report the findings from an exercise to identify research priorities for eight areas of adolescent health in LMICs with the aim of stimulating research on the priority questions identified. The specific areas of adolescent health selected for inclusion were communicable diseases prevention and management (including diarrhea, parasites, hepatitis, malaria, meningitis, tuberculosis, influenza, pertussis, pneumonia, and others), injuries and violence, mental health, noncommunicable diseases management (including asthma, diabetes, cancer, hypertension, heart disease, and others), nutrition, physical activity, substance use, and adolescent health: policy, health and social systems. Of note, adolescent sexual and reproductive health and related topics were not included, as they had been the subject a recent similar research prioritization exercise .
The Child Health and Nutrition Research Initiative (CHNRI) developed a method for ranking the relative importance of competing research options to help decision makers to effectively allocate limited resources to reduce morbidity and mortality . The CHNRI approach has previously been applied to more than 50 health areas , , , , , including adolescent sexual and reproductive health .
We implemented a modified version of the CHNRI priority setting method in three phases. In Phase 1, we identified research and program experts through systematic searches of published and gray literature, members of journal editorial boards, and through interviews with key informants at WHO, and invited them to participate in the exercise. In Phase 2, we asked the experts who agreed to participate to propose research questions related to descriptive epidemiology, interventions, and health and social systems research. In Phase 3, we asked the same experts to prioritize the research questions generated in Phase 2 using a scoring scheme based on five criteria.
Experts were identified through journal publications, membership of journal editorial boards, from lists of participants at WHO meetings and consultations, and by nominations from relevant WHO departments. For journal publications, we identified experts in each health area through a systematic search of PubMed and Web of Science databases from 2005 to 2015. To be included on this preliminary list, authors had to have published at least two relevant articles within a specific health area that explicitly covered adolescents (ages 10–19 years) in LMICs during the 2005–2015 period. If more than 20 experts met these criteria, then the number was reduced to a maximum of 20, based on number of publications, relevance of the titles of the articles, and the position of authorship, with discrepancies resolved through discussion by D.R. and J.F. This resulted in 116 experts.
We searched for peer-reviewed journals related to adolescent health in all six official United Nations languages. Members of the editorial boards of the two peer-reviewed journals related to adolescent health with the highest impact factor (Journal of Adolescent Health—2.75 and Journal of Research on Adolescence—2.51) based on Web of Science Journal Citation Reports for 2013  were included in the adolescent health: policy, health and social systems area. This identified an additional 69 experts.
Since the experts identified through the systematic PubMed and Web of Science search were likely to mainly be researchers, we also identified participants at WHO meetings and consultations held in 2010–2015 and that were relevant to the eight adolescent health areas through reports that were available on the WHO website and the WHO Index Medicus, a database focused on health literature produced by and within LMICs from all regions. Such meetings usually include program implementers and policymakers and researchers. The meetings included several that had participation by young persons themselves. We also invited representatives of the WHO departments relevant to each health area to review the lists and nominate any additional key experts in their respective fields. Overall, this resulted in 265 additional experts.
Combining the list of experts resulted in a total of 450 different individuals (Table 1). All these 450 experts were sent an invitation to participate in the research prioritization process, and 217 (48%) agreed to participate.
The experts identified in Phase 1 were divided into groups based on their expertise in the eight adolescent health areas. Each expert was asked to propose research questions of the greatest priority for adolescent health within their health area related to descriptive epidemiology, interventions, and health and social systems research:
Questions were submitted via a survey tool using SurveyMonkey (Palo Alto). The 512 submitted questions (Appendix A) were synthesized by removing redundancies and questions not relevant to adolescent health, as well as repositioning questions that belonged in different health areas. Some questions were rephrased in an attempt to improve clarity. This resulted in 303 questions that were included in the final scoring (Appendix B).
The same 217 experts were asked to score the final list of research questions generated in Phase 2 in their health area of expertise and in the adolescent health: policy, health and social systems area. Experts in the adolescent health: policy, health and social systems area were asked to score one additional health area of their choosing.
Experts were asked to score questions against five specific criteria:
Experts were asked to score each question for each of the criteria based on the standard CHNRI scoring system: yes, no, or undecided.
In October 2015, 15 external experts joined the authors and other WHO staff in a meeting at which the methods and preliminary findings were discussed before they were finalized.
All answers were converted to a score. A “yes” scored 100; “undecided” 50; and “no” 0 points. Rankings were based on the total Research Priority Score (RPS), which was computed as the mean of the scores for the different criteria, weighted according to published guidelines from CHNRI stakeholders  and adjusted to a 100-point scale, according to the formula:
RPS = [(answerability × .86) + (impact × 1.56) + (deliverability × .77) + (equity × .81)]/4. Although clarity has been used as a criterion in previous CHNRI exercises , its weight has not been validated by CHNRI methodological guidelines so it was not included in the final RPS . In addition, the Average Expert Agreement (AEA) scores are reported, which represent the average proportion of scorers that agreed on responses for each of the five criteria asked. This was computed as:
Characteristics of the 142 experts who submitted questions are shown in Table 2. Over half were female (57.0%) and were employed in academic institutions (63.1%). There were fewer representatives from governments and donor organizations. Most experts had a postgraduate degree (88.7%). About half of the experts described their primary role as a researcher, whereas about a fifth were program managers and a 10th were clinical health practitioners or policy makers. The experts represented 62 countries from North America (28.2%), South America (10.6%), Europe (21.1%), Africa (14.8%), Asia (13.4%), and Oceania (12.0%). The number of experts who were identified and who participated in each of the stages of the exercise is shown by health area in Table 1. The total number of experts who were approached was 450; the number varied by “health area” from 44 (substance use) to 77 (adolescent health: policy, health and social systems). From these, a total of 217 agreed to participate, 142 submitted questions, and 130 scored the questions.
Appendix C presents the full list of 303 questions that the experts were asked to score, the mean scores of each question and of each health area. The top 10–ranked research questions in each of the eight health areas are shown in Table 3. The total RPS for the top 10 questions in the eight health areas ranged from 73 to 100 out of a possible 100. The AEA score for the top 10 questions (as ranked by total RPS) ranged from 61 to 98 out of a possible 100 in the eight health areas.
For the top 10 questions in each health area, the overall mean RPS was 87, and the mean AEA was 79. In terms of scoring criteria, answerability had the highest mean score (90), followed by impact (88) and clarity (88). Equity was the criterion that had the lowest mean score (81). The types of research represented in the top 10 questions for the eight health areas were descriptive epidemiology (26%), interventions: discovery (11%), development/testing (25%), delivery (33%), and policy, health and social systems (5%; Appendix D). The top 10–ranked research questions are shown by research type in Appendix E.
Across health areas, the top-ranking research questions highlighted various themes reflecting the diversity of issues affecting adolescent health. Several questions featured delivery of interventions via different platforms, such as schools (N = 14), primary care (N = 5), families/parents (N = 5), and interactive media (i.e., novel communication technologies, mobile phones, internet, social media; N = 4). Other questions addressed integration of health services, for instance between physical health, mental health, and reproductive health services. Finally, key subpopulations of vulnerable adolescents were identified in top-ranked research questions including young sex workers, injecting drug users, refugees, and out-of-school youth. Additional themes are discussed by specific health area.
The top 10 communicable diseases prevention and management questions were dominated by tuberculosis (TB; 9 of 10), with six of nine of the TB questions also related to HIV co-infection or linking TB and HIV services (Table 3). Four of the questions were related to adherence, and three were related to retention in care. The only non-TB question that was in the top 10 communicable diseases questions was related to diarrhea and lower respiratory tract infections, whereas questions related to malaria or neglected tropical diseases were absent. This may have reflected the interests of the experts who proposed questions and scored them in this health area. For instance, the systematic searches of the literature for experts in malaria or neglected tropical diseases and adolescence yielded far fewer results than searches for experts in TB or HIV/AIDS and adolescence. Adolescence has been identified as a critical time in HIV and TB treatment and care, with recent studies demonstrating that HIV has risen to become the second-highest cause of adolescent mortality globally . Nonetheless, research in malaria and neglected tropical diseases among adolescents may be a crucial under-represented research area despite the fact that they were not featured in this priorities exercise.
The top-ranked injuries and violence question was related to barriers and facilitators of motorcycle helmet legislation. Other specific issues addressed in the top 10 questions related to drowning, bullying, partner violence, sexual violence, and burn injuries. Three of the questions related to applying or combining interventions in one area to other areas (for instance, combining brief alcohol interventions with brief violence reduction interventions or using strategies against bullying to prevent partner violence or sexual violence).
Although one question related to gender-based violence was submitted, it did not rank in the top 10 injuries and violence questions in this exercise. Of note, the previous research priorities exercise on adolescent sexual and reproductive health included an entire area on gender-based violence. Eighteen experts on gender-based violence participated, and five priority questions were featured in the published results for that exercise. Some of these questions addressed underlying issues for gender-based violence. For instance, the top-ranked question was, “how do programs that aim to keep girls in school longer through measures such as conditional cash transfers affect the prevalence of gender-based violence?”
In addition, effective strategies for a responsive health system, empathetic provider behavior, and having a single point of access to multiple different types of care may serve to promote focus on improving service delivery for sexual violence, burns, and other injuries.
The top-ranked mental health question addressed the cost-effectiveness of a package of interventions for the promotion of mental health. A question about the effectiveness of parenting programs in the prevention of mental health disorders also featured in the top 10. Three questions were related to integration of management of mental health with primary care or reproductive health care services and other strategies such as adolescent friendly health services. The third-ranked mental health question focused on suicide and self-harm behaviors in adolescent girls. Recent global reports have estimated that suicide surpassed maternal mortality as the leading cause of death among older adolescent (15–19 years old) females globally . However, although ranked third among older adolescent (15–19 years old) males, the actual mortality rates were estimated to be almost identical (11.73/100,000/year in females vs. 11.72/100,000/year in males) in 2012 . Research on self-harm and suicide in adolescent males also remains an under-researched area.
The top-ranked question on noncommunicable diseases (NCDs) management related to developing a low-cost rapid antigen test for streptococcal pharyngitis for the prevention of rheumatic heart disease. This was the only intervention: discovery question that ranked number one in a health area. In addition to rheumatic heart disease, research on other forms of heart conditions may be important during adolescence. For instance, for children born with congenital heart diseases who survive through childhood, loss to follow-up and transitions to adult medical care remain continuing health challenges. Furthermore, acquired heart-related conditions such as hypertension, high cholesterol, and coronary heart disease may begin to develop in adolescence and are an emerging research area. Four of the 10 top-ranked questions were related to diabetes, whereas other specific diseases included rheumatic heart disease and sickle cell disease. Although no questions related to cancers affecting adolescents featured in this exercise, this remains an important area of research in LMICs .
Two of the top 10 questions were related to applying existing interventions in one population group to another, for instance applying NCDs management interventions in adults to adolescents or applying NCDs management interventions in high-income settings to low-income settings. Of note, the NCDs management section did not include NCDs prevention because many of these preventive behaviors may be covered by the substance use, nutrition or physical activities areas. However, there are other areas of behaviors including sleep patterns, increased screen time with electronics and social media, social pressures, and stress related to studies, work, or earnings that may contribute to the development of NCDs and other health problems.
The top-ranked nutrition question related to the causes of adolescent anemia and how the causes vary by geographical region. Four of the top five nutrition questions were descriptive epidemiology research questions. Two of the 10 top-ranked questions related to the relationship between overnutrition and undernutrition; three related to differences in nutritional risk factors or problems by region, country, or socio-economic status; and two related to nutritional status or support for pregnant adolescent girls.
Identifying variables that predicted physical activity patterns among adolescents in LMIC was the top-scoring physical activity question. Five of the 10 top-ranked questions related to schooling or school-based physical activity interventions, whereas two related to scaling up physical activity interventions. Overall, most physical activity questions related to interventions, and particularly, their development/testing or implementation/delivery.
Three of the top 10 research questions in this health area were intervention: discovery questions, more than in any other health area. The top-ranked question on substance use was related to the most acceptable prevention and treatment services to adolescents. Two questions specifically addressed alcohol and tobacco, respectively. Three of the questions were related to community-based, parent-based, or peer-based interventions.
The top-ranked question for adolescent health: policy, health and social systems related to platforms to reach the most vulnerable adolescents. To reach the most vulnerable adolescents, research on their health status and needs may also be necessary. Three of the top questions related to primary care, including effectiveness of different models and coverage, and three related to information and communication technology, whether mobile health interventions, the internet, or social media.
Priority questions have been identified for research in eight key areas of adolescent health in LMICs through 2030, extending earlier work that proposed research priorities in adolescent sexual and reproductive health. Using a modified version of the priority setting method developed by the CHNRI, we received input from 142 experts who generated 512 research questions. These expert-generated questions may be used by donors, program managers, and researchers to stimulate and develop research in adolescent health.
A limitation of the exercise is possible nonresponse bias given that not all experts agreed to participate in the exercise (217/450 = 48%) and not all those who agreed to participate actually did so. Only 142/217 (65%) submitted questions and 130/217 (60%) scored questions. This was despite efforts to encourage responses from all. Selection bias may also affect results, as a majority of respondents were researchers from academic institutions, with less representation from program implementers, policy makers, and funders. This potential bias may reflect the lack of questions on estimating numbers for harmful practices among adolescents. Although equity had the lowest overall average score relative to the other criteria among the top questions, the mean was still 80/100. This value was consistent with previous research priority exercises with mean equity scores 84–86 , , . The creation of eight health areas, which were based on burden of adolescent mortality and morbidity as well as a life course approach to health, inevitably leads to some degree of merging and separation of topics. Furthermore, some specific disease areas may have stronger representation than others based on identified experts and their response rates, despite an effort to include a breadth of expertise by topic. The use of PubMed and Web of Science databases may have identified experts who tended to publish in English language scholarly journals although we searched for publications and experts in all languages and from all regions using the WHO website and WHO Index Medicus, a database focused on health literature produced by and within LMICs.
Furthermore, some of the questions that were submitted spanned two or more types of research question (e.g., both development and delivery types of intervention question) but, for ease of scoring and analysis, were categorized as the category that they were submitted in.
Strengths of the exercise included identification of a large number (450) of experts in adolescent health spanning a diverse range of health areas. The CHNRI methodology is a systematic and transparent process that has become the most common methodology for identifying research priorities since 2001 . It uses independent scoring by experts, avoiding situations where the most vocal or opinionated individuals affect group decisions or priorities . The range of AEA in this exercise (61–98) was consistent with, or higher than, previous research priority exercises, such as for preterm birth (62–83) , childhood pneumonia (64–76) , and newborn health and prevention of stillbirths (62–77) . Although questions were organized in vertical health areas, during the analysis questions were also classified horizontally by delivery platforms (i.e., primary care, schools, families/parents, and interactive media). Future research may consider qualitative methodology exploring themes of top questions across all adolescent health areas (including questions from the adolescent sexual and reproductive health priorities exercise).
Adolescent health is receiving increasing attention globally. For example, the updated United Nations' Global Strategy for Women's, Children's, and Adolescents' Health 2016–2030 features adolescents for the first time, and the emphasis on going beyond “survive” to “thrive” and “transform” will greatly increase the focus on adolescent health and development . Furthermore, although only one of the 17 Sustainable Development Goals is specific to health, all 17 will directly or indirectly affect adolescent health . Given the need for evidence-based policies and programs to improve adolescent health as part of these new global initiatives, priority questions for research in eight key areas of adolescent health in LMICs have been identified using a transparent process that included experts from multiple disciplines, types of institutions, and countries. These expert-generated questions may be used by donors, program managers, and researchers to stimulate and develop research in adolescent health.
The authors thank the 142 experts who actively participated in the exercise by submitting and/or scoring questions; the participants at the WHO Consultation on Adolescent Health Research Priorities who included Drs. Sulafa Ali, Margit Averdjik, Anne Buvé, Bruce Dick, Aoife Doyle, Adesegun Fatusi, Rashida Ferrand, Gwyn Hainsworth, Daniel Hale, Mark Jordans, Ana Menezes, Mahmood Nazar Mohamed, Vikram Patel, Daniel Tobon Garcia, and Daniel Wight; WHO colleagues including Drs. Annabel Baddeley, Valentina Baltag, Paul Bloem, Raschida Bouhouch, Alexander Manu, Nigel Rollins, Chiara Servili; and Claire Ory-Scharer and Margaret Kigundu for administrative support. The authors particularly thank Drs. Sachiyo Yoshida, Michelle Hindin, and Rajiv Bahl for advice related to CHNRI methodology, Joya Banerjee and Jill Kowalchuk for help with the systematic searches, and Tomas Allen for advice on electronic literature database searches.
Conflicts of Interest: The authors have no conflicts of interest or financial disclosures to report.
Disclaimer: The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of WHO.
Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jadohealth.2016.03.016.
Partial funding for this exercise was provided by the US Agency for International Development and the Mary Duke Biddle Clinical Scholars Program, Stanford University (JMN).