Background
Health risk behavior significantly contributes to the burden of disease and social problems among young people globally [
1]. Health risk behavior (HRB) encompasses actions and related attitudes and perceptions that underlie people’s propensity to engage in activities associated with increased susceptibility to a specific disease or ill health as shown in epidemiological or social data [
2,
3]. Although prioritized forms of HRB may vary across geographical and demographic contexts, some of the commonly assessed forms of HRB include: sexual behaviors resulting in unintended pregnancy and sexually transmitted diseases; alcohol, tobacco and other drug use; behavior resulting in unintentional injury or violence; unhealthy dietary behavior; poor hygiene practices; and inadequate physical activity [
4‐
6].
HRB is particularly of concern during adolescence (10–19), mainly because this stage of development has been linked with increased impulsivity and propensity for risk taking that might result into disability and fatal outcomes [
7,
8]. The leading cause of disability and mortality among adolescents are HRB. More than 3000 adolescents die every day largely from preventable causes such as road traffic accidents, falls, diarrheal diseases, iron deficiency, alcohol use, violence and self-inflicted injuries; most of which are associated with HRB. These preventable causes of mortality are also ranked among the top ten causes of disability-adjusted life-years (DALYs) among adolescents globally [
1,
9]. Moreover, adolescence is a period when behaviors (either risky or protective) start or are consolidated and this has major implications for health in adulthood [
9]. Governments have therefore been called upon to prioritize action and investments in prevention, in accordance with the disease and injury risk factor profiles of their adolescent populations [
9].
Understanding HRB of adolescents especially from low resource settings such as sub-Saharan Africa (SSA) is partly hindered by the lack of culturally appropriate measures of HRB, considering that majority of the available HRB tools have been developed and utilized in high income settings [
10]. In Kenya, which is also the setting for this current study, there is a growing body of research on adolescents’ HRB [
11‐
17]. However, for most of these studies the authors do not explicitly inform the readership about the process of development, sources and psychometric qualities of the utilized HRB items. This status quo may imply that either: i) HRB items are often directly translated for use within the Kenyan context from a different context; or ii) that HRB researchers commonly develop new items and administer them without thoroughly ascertaining their cultural appropriateness and content validity. Although this approach of formulating new items, or using already existing ones with minimal or no modification, is relatively cheap and easy to implement; it is not often culturally informed and potentially may contribute to biased results [
18]. These issues can include different forms of bias, like
item bias which arises when items function differently or have different meaning in different contexts;
construct bias which may arise when an instrument partially explores the domains that constitute a construct; and
method bias that arises from differences in administration procedures or sample characteristics [
18]. Another challenge for adolescent HRB research within the Kenyan context is that many studies tend to assess HRB in isolation, hence missing out important aspects such as co-occurrence of HRB of adolescents, which has for instance been explored in a few studies and found to be an important issue [
19,
20]. Regional disparities within Kenya in the burden, forms and specific underlying factors for some behavior and related outcomes [
21] also underscore the need for culturally appropriate and comprehensive HRB tools. An example is Kilifi County at the Kenyan coast where national level data suggests a disproportionately higher burden of sexual risk behavior outcomes e.g. a highest prevalence (21%) of teenage pregnancy [
21]. The existence of such differences potentially emphasizes the presence of various context specific underlying factors [
22].
There is a growing body of evidence to guide researchers on steps towards ensuring cultural appropriateness and validity of tests on health outcomes in low resources settings where still scarcity of such tests exists [
18,
23,
24]. Adoption, adaptation and assembly of assessment instruments are three alternatives, but the latter two options are preferred compared to adoption since they are more likely to contribute contextually relevant tools [
18]. Adoption involves translating a measure and using it in another culture or context, while adaptation involves a systematic evaluation of all aspects of an existing instrument and modifying it where needed to suit context. For assembly, a new measure is either developed directly and informed by local culture and context or alternatively items and procedures are borrowed or modified from various standardized measures [
18]. Abubakar and van de Vijver [
18] propose a four-step approach to adaptation and assembly of tests which involves: i) a mixed methods approach to construct clarification; ii) item development, which comprises item translation and formulation; iii) scale development, which involves refining a scale by pretesting and piloting; and iv) psychometric and non-psychometric approaches to test evaluation. We explain in depth these procedures in the methodology section of this manuscript.
This present study is conducted in Kilifi County; a rural setting at Kenyan coast, with a main objective of assembling and psychometrically evaluating a comprehensive questionnaire for assessing health risk behavior of adolescents in this environment. Owing to the sensitivity of topics in HRB, we further aimed to ensure that the psychometrically evaluated HRB questionnaire can be answered in a manner that maximizes the adolescents’ privacy.
Results
Characteristics of participants
The characteristics (sample size, age, and sex) of the participants that took part in the various stages of the KRIBE-Q assembly and psychometric evaluation are summarized (see Table
1).
Table 1
A summary of age and sex of the participants for different stages of the study
Construct clarification (focus groups & key informants) | Adolescents (78) | 15.0 (2.4) | 53.8 |
Stakeholders (10) | 35.8 (2.9) | 40 |
Young adults (7) | 24.7 (1.1) | 43 |
Item development and refinement | Translators (2) Harmonization Panelists (6) | – | 50 |
– | 83.3 |
Test evaluation |
i. Baseline (Time 1) | Adolescents (164) | 14.8 (2.4) | 49.4 |
ii. Retest (Time 2) | Adolescents (85) | 14.3 (2.5) | 47.1 |
Non Psychometric evaluation of ACASI | Adolescents (40) | 14.1 (1.6) | 42.5 |
Construct clarification
From the systematic review, the Youth Risk Behavior Surveillance System (YRBSS) and Health Behavior in School-aged Children (HBSC) were the commonest and most comprehensive HRB assessment tools and sources of items on HRB for adolescents living with chronic conditions [
10]. Based on the findings from the review [
10], the YRBSS questionnaire [
26], informed most of our item choices for the assembly of the KRIBE-Q. We also modified the format of the items from the YRBSS to suit the local context in Kilifi and borrowed some items from the Global School –Based Student Health Survey [
4], DSM-IV-MR-J [
28], the Pittsburgh Sleep Quality Index [
29], and International Physical Activity Questionnaire (IPAQ-Short) [
30].
Our results from the focus group discussions and the key informant interviews in Kilifi showed a number of contextually relevant forms of HRB and specific local names or jargon that we integrated in to the questionnaire. The specific modifications made included:
i)
Modification of items on behavior resulting to injury and violence in order to capture injuries from motorcycles and bicycles, falls, burns and cuts; and cyber bullying (i.e. through phone text messages and social media platforms) which during the focus group discussions were found to be common among adolescents in Kilifi.
ii)
Inclusion of locally relevant examples of tobacco products like shisha, ugoro, and tobacco leaves or tumbaku; examples of locally brewed forms of alcohol like mnazi, and changaa; and local names of other drugs for example bangi, ganja and makushabu (for marijuana) and miraa or mogoka (for Khat) in the questionnaire’s sections on tobacco use, alcohol and other drug use behavior.
iii)
Specific locally relevant examples of healthy foods for example vegetables like kales (Sukuma wiki), amaranthus (mchicha), and potentially unhealthy locally or fatty available foods like fried chicken and viaza karai that were named by young people were also included under dietary behavior in the assembled KRIBE-Q [
46].
iv)
Gambling was another noteworthy form of behavior discussed by young people in Kilifi [
46] which is not captured by the YRBSS questionnaire [
26]. Therefore three items assessing gambling behavior were borrowed from the DSM-IV-MR-J [
28], with some modifications such as giving specific examples of gambling games such as card games or ‘kamare’, lottery or scratch tickets, casino games or ‘Mchina’ and sports betting that had been mentioned in the discussions.
v)
The need for assessment of quality of sleep behavior followed from engagement in social events such as parties and traditional ceremonies by adolescents, which take place in the night [
22]. An item on this aspect was borrowed from the Pittsburgh Sleep Quality Index [
29].
vi)
Personal hygiene behavior such as poor hand washing practices, poor oral hygiene and general body cleanliness were also mentioned by young people as perceived forms of HRB in Kilifi. We therefore borrowed items (with minimal modification) on personal hygiene behavior (i.e. oral hygiene, handwashing and general body hygiene), from the Global School –Based Student Health Survey (GSHS) 2013 core questionnaire modules [
4].
Item development and refinement of the questionnaire
Suggestions from the harmonization panel were that the 5 items on safety (question 8–13) from the YRBSS [
26], be replaced with the 3 items about serious injuries borrowed with some modification from the GSHS 2013 core questionnaire modules [
4]. The main reason for this was because the YRBSS questions about safety majorly focused on road traffic safety (especially motor vehicle safety), whereas other forms of injury appeared to be of higher priority in the Kilifi context, many of which are captured in the GSHS questionnaire. Addition of an extra item was also proposed to capture why the identified forms of injuries occurred, and the response options reflected issues discussed by young people: for example being under the influence of alcohol or drugs, being reckless, and having no control over the incident.
Items about violence related behavior that made reference to school property in the YRBSS were modified to reflect occurrence in a general community context in order to optimally include all potential HRB.
The harmonization panel proposed that items on alcohol use behavior from the GSHS be used since they capture various problem drinking related indicators, like number of drinks consumed in a day, being in trouble or missing school due to alcohol drinking, and being intoxicated. Additionally, the GSHS items also ask about caregivers’ alcohol use unlike those in the YRSS [
4,
26].
Consensus from the panel was that items on sexual orientation (heterosexual, gay and bisexual) as used in the YRBSS should be dropped from the assembled KRIBE-Q. Asking adolescents about their sexual orientation was thought as a culturally complex subject in the context of Kilifi.
There was preference for items on physical activity that assess vigorous and moderate forms of physical activity as well as sedentary lifestyle. Thus such items were borrowed from the international physical activity questionnaire (IPAQ-Short) [
30].
Scale development
An instruction manual for the administration of the KRIBE-Q outlining various procedures for observing privacy, seeking permission, scoring the items, handling data collection materials, managing data and reporting was developed. Following consultations with school authorities and the adolescents, we found that it was most convenient to administer the questionnaire at the start or mid-way into the academic term before too much academic work load and examinations. The sports’ time was suitable for this activity since this avoided interference with academic activities. Following the training, the research assistants and counselor felt prepared for data collection in the test evaluation phase.
Test evaluation
Of the 214 adolescents who had shown interest during the recruitment stage, 76.6% (164) completed the KRIBE-Q at Time 1, and their data was basis for the data quality and scaling evaluation. Table
2 presents the participants’ characteristics.
Table 2
Characteristics of participants involved in the HRB tool psychometric evaluation phase
Sex |
Male | 49.4 | 47.1 |
Female | 50.6 | 52.9 |
Adolescent group |
Young adolescents (10–14 yrs) | 41.5 | 50.6 |
Older adolescent (15–19 yrs) | 58.5 | 49.4 |
School level |
Primary (Class 5–8) | 45.4 | 57.1 |
Secondary (Form 1–2) | 54.6 | 42.9 |
Religious affiliation |
Christian | 82.2 | 84.7 |
Moslem | 17.2 | 15.3 |
Other | 0.6 | 0.0 |
Our results of test evaluation include 60 of a total of 69 items on health behavior from the KRIBE-Q (see Additional file
1). We excluded five multiple choice response items from the analysis; four of which were about unintentional injury and violence and one was on alcohol use. We also do not present results from four other items for which the frequency of the response options indicated that the participants had misclassified their responses. The response options of two misclassified items (use of alcohol or drugs prior to sex and use of a condom at most recent sexual intercourse) were rephrased while the other two items (attempted smoking cessation and lifetime use of prescription drugs) were dropped from the final version of the KRIBE-Q.
Data quality and scaling evaluation
None of the items in the HRB questionnaire had 5% or more missing data which indicated that acceptable data quality was obtained. Results from scaling evaluation indicated that about 20% of the items in the questionnaire had less than 60% of their response options utilized by the study participants. Table
3 summarizes the 13 items which had less than 60% of their response options utilized and the amendments that were made to these items.
Table 3
A description of items with sub-optimal results from scaling evaluation and the specific amendments undertaken
13.Been threatened or injured with a weapon during the past 12 months | 50.0 | Retained in its original format. The participants’ responses were varied and this item is important in the context of Kilifi. |
23.Attempted suicide during the past 12 months | 40.0 | Retained in its original format. The item is vital for assessing psychiatric and behavioral problems. |
26. Smoked cigarettes during the past 30 days | 42.8 | Item 26 and 27 were pooled into a new item asking about “cigarette smoking or use of any other tobacco products other than cigarettes e.g. shisha, ugoro, cigas, tobacco leaves/tumbaku” |
27. Used any other tobacco products during the past 30 days | 28.6 |
35. Has ever drank so much that they were really drunk | 50.0 | Retained in its original format. The item is crucial for assessing harmful alcohol use behavior |
38. Ever used marijuana | 40.0 | Each of these items was retained in its original format since discussions showed that marijuana is common within Kilifi. |
39. Had used Marijuana during the past 30 days | 40.0 |
40. Ever used any form of cocaine | 40.0 | Item 40, 42, 44, 47 were pooled into a multiple choice response item as opposed to asking separately about frequency of use of each drug. |
42. Ever used heroin | 40.0 |
44. Ever used methamphetamines | 40.0 |
47. Ever used a needle to inject illegal drugs into the body | 33.3 |
51. Number of sexual partners during the past 3 months. | 50.0 | Retained in its original format. Multiple sexual partnerships were mentioned as common by young people during the construct clarification. |
Test-retest reliability
The overall Gwet’s AC1 coefficient of the assembled Swahili version of KRIBE-Q was very good (ranging from 0.81 to 0.83 across sex groups) and there were no statistically significant differences in the coefficients across sexes, adolescent age categories and level of education. Across the 8 behavior categories, the Gwet’s AC1 coefficient ranged from ‘very good’ to ‘moderate’ with majority of the behavior categories (5 out of 8 categories) having very good reliability coefficients (see Table
4).
Table 4
Mean Gwet’s AC1 coefficients and their 95% confidence intervals summarized by participants’ demographic and health behavior categories
Sex |
Male | 0.81a | 0.66, 0.94 |
Female | 0.83a | 0.70, 0.95 |
Adolescent group |
Young adolescents (10–14 yrs) | 0.81a | 0.67, 0.94 |
Older adolescent (15–19 yrs) | 0.83a | 0.69, 0.95 |
School level |
Primary (Class 5–8) | 0.82a | 0.69, 0.94 |
Secondary (Form 1–2) | 0.83a | 0.67, 0.96 |
Risk behavior categories |
Behavior related to Injury and Violence | 0.85 | 0.76, 0.93 |
Tobacco Use behaviors | 0.85 | 0.77, 0.94 |
Alcohol and other drug use behavior | 0.96 | 0.91, 0.99 |
Sexual Behaviors | 0.94 | 0.88, 0.99 |
Dietary Behaviors | 0.60 | 0.43, 0.77 |
Physical Activity Behaviors | 0.74 | 0.59, 0.88 |
Gambling behavior | 0.73 | 0.59, 0.87 |
Hygiene behavior | 0.89 | 0.82, 0.96 |
At an item level, Gwet’s AC1 coefficients ranged from ‘good’ to ‘very good’ (i.e. 0.63–1.00) among 83.3% of the items in the questionnaire while 11.7% had ‘moderate’ Gwet’s AC1 coefficients ranging from 0.46 to 0.59 (See Additional file
1).
Prevalence of behavior
There were no statistically significant differences between prevalence of behavior outcomes at Time 1 and Time 2 (
n = 85) with exception of a few items on alcohol, tobacco and drug use, dietary behavior and hygiene (See Additional file
1).
The baseline (Time 1) prevalence of unintentional injury and violence related behavior was particularly high, for example, occurrence of serious injuries during the past 12 months (61.1%), feeling unsafe in the neighborhood (14.2%), engagement in physical fights during the past 12 months (39.9%), and experience of bullying (31.7%).
Noteworthy among substance and drug use related behavior were: the early initiation (at 13 years or younger) of cigarette smoking (6%); exposure to second hand smoke (54.3%); having parents or guardians who use tobacco (18.9%); lifetime alcohol use (10.4%); recent alcohol use (5%); and the use of Khat (14%).
From Time 1 data, about 11% of the adolescents were sexually active and 5.5% of the total respondents (representing about 45% of the sexually active) had their first sexual intercourse at the age of 13 years or less. Also 6.1% (equivalent to 43% of the sexually active) did not use any method for prevention of pregnancy during their most recent sexual intercourse.
Other important behavior outcomes were that close to 23% of the adolescents did not engage in vigorous or moderate physical activity within the past 7 days; and almost a quarter (24%) spent 3 or more hours engaging in sedentary activities on a typical day. Engagement in gambling behavior within the past 12 months was common (26%) and so was the lack of adequate handwashing with soap after using a latrine /toilet use among 28.8% of the adolescents.
Non-psychometric evaluation of the KRIBE-Q delivered via ACASI
We collected data on user experience of the customized audio-computer assisted self-interview (ACASI) from 40 adolescents of mean age 14.1 years (SD = 1.62), of whom 57.5% were female and 60% were young adolescents (10–14 years). Majority (95%) of the adolescents rated the interview (ACASI) as either ‘very easy’ or ‘just okay’ or 90% found it as either ‘very interesting’ or ‘a bit interesting’. Almost all (97.5%) of the adolescents ‘strongly agreed’ or ‘agreed’ that the interview was confidential and that they were comfortable during the ACASI exercise. From a test administrator’s observation, he either ‘agreed’ or ‘strongly agreed’ that 97.5% of the participants seemed comfortable and confident while using the ACASI. Majority (82.5%) did not require any assistance at all when using the ACASI while the rest needed minimal assistance especially in reminding them how to maneuver to the next item on the computer.
Discussion
Our findings from construct clarification show that adolescents in rural coastal Kenya have either experienced or are familiar with most of the constructs of HRB utilized in measures like YRBSS and GSHS which were developed in other settings. Although this demonstrates good conceptual equivalence, our experience was that in-depth consultations with the adolescents and key informants, combined with a harmonization process are vital procedures especially for identifying priority forms of behavior to focus the HRB tool development and capturing common semantics and idiomatic aspects of HRB utilized by adolescents in the study setting. We also found the harmonization process as fundamental in refining the borrowed items especially in connection to cultural appropriateness and inclusiveness of relevant behavior aspects voiced by the adolescents.
The finding that none of the items had 5% or more missing data suggests that construct clarification and item refinement significantly improved clarity and suitability of the HRB items assembled for the adolescents in the study setting. Moreover, this was further demonstrated by the fact that both young and older adolescents capably completed the assembled HRB questionnaire. The completeness of the items also potentially indicates the absence of response fatigue. However, we found that four of the sixty nine assembled items had some misclassified responses; a problem we attribute to ambiguity of the response options. For example, one of the items which asked a participant whether he or she used a condom during the last sexual intercourse initially had the following response options: “I have never had sexual intercourse”, “Yes”, “No” and we recognized a tendency for participants to select the answer option “No” while they actually meant “I have never had sexual intercourse”. For this item, we finally modified the response options as follows: “A. I have never had sexual intercourse”, “B. Yes, I used a condom”, “C. No, I had sexual intercourse but did not use a condom” so as to reduce the ambiguity. In line with our refinement of the misclassified items, there has been evidence showing that more extensive verbal labeling of response options is associated with higher reliability [
47].
The majority of the assembled HRB items demonstrate good spread of responses across options however about 20% of the assembled HRB items had sub-optimal scaling evaluation. We think this may have been due to very low prevalence of certain forms of behavior such as drug use and suicidal behavior in this setting, which meant that certain response options were often redundant. Indeed the number of response categories for all these items were within the recommended ranges of four to nine options [
48]. Rattray and colleagues [
49] advise that although redundant response options may warrant deletion of an item, it is always crucial to refer to the original research question and retain items that are thought to reflect important underlying theoretical domains. We therefore addressed such scaling problems by for example pooling items on rare forms of behavior into a single item as opposed to asking about the frequency of each single behavior. As an example, instead of asking participants how many times in their lives they had used cocaine, heroin, methamphetamines and steroid pills without a doctor’s prescription separately, we pooled them into a single multiple choice objective item asking “During your life, have you ever used the following drugs: A. Cocaine, B. Heroin, C. Methamphetamines, D. Steroid pills without a doctor’s prescription”.
Our findings support acceptable test-retest reliability of the majority of the assembled items of the KRIBE-Q for the adolescents in the rural coastal Kenyan setting. There was acceptable reliability irrespective of sex, age category and education level as shown by the overlapping confidence intervals of overall mean Gwet’s AC1 coefficient across these groups in Table
4. Overall, behavior categories of alcohol and other drug use, sexual activity, tobacco use and dietary behavior demonstrated higher test-retest reliability than the dietary behavior and physical activity categories. These findings are consistent with those reported in a study on reliability of the Youth Risk Behavior Survey questionnaire among high school students in the district of Columbia, USA [
6]. One explanation for the low reliability of dietary behavior items may be that adolescents’ diet frequently changes over a short period of time which makes it difficult to reliably recall behaviors related to their nutrition. Another explanation offered by Brener and colleagues is that behaviors surrounding substance use and sexual activity are likely to be more salient to adolescents and thus more reliably recalled compared to dietary behavior or physical activity [
6].
Although test-retest reliability was high, we suspect that due to very low prevalence of specific forms of behavior, this potentially resulted to none overlap of confidence intervals (statistically significance of differences) of prevalence rates of some behavior on first (Time 1) and second test (Time 2) especially for items assessing alcohol and drug use behavior. However, the inconsistence in prevalence may also suggest response errors for example arising from recall bias, social desirability bias or there may be actual changes in prevalence. Thus the results on prevalence of the behavior in our study should be treated with caution and may need much larger sample sizes to determine if there are actual differences.
Our findings indicate that adolescents frequently report occurrence of behavior resulting to intentional and unintentional injuries like bullying and physical fights, however such results need to be replicated in larger quantitative studies in this specific study setting. These findings are however not surprising since other larger studies also report a high occurrence of bullying (between 58 and 82%), conduct problems and involvement in physical fights among Kenyan adolescent students [
50,
51]. High occurrence of early sexual debut and unprotected sex among our sexually active study participants further highlights the need for more sexual and reproductive health research and interventions in rural coastal Kenya; where adolescent sexual health is reported as still poor [
21,
22]. Noteworthy, adolescents also endorsed other infrequently researched forms of behavior in the context of rural coastal Kenya such as gambling behavior, poor personal hygiene and physical inactivity. This potentially points to the need for more investment in adolescent health research in this setting.
This study’s findings from non-psychometric evaluation of the HRB questionnaire delivered via ACASI highlight the feasibility of using ACASI for collecting sensitive data such as information on HRB among adolescents in this rural low resource setting. In support of our findings, ACASI has previously been used among the adolescent population in Kilifi [
52], as well as among an older age-group within the Kenyan setting [
53,
54]. Overall, these studies have found the ACASI to improve quality of response to sensitive questions, to decrease socially desirable responses, and has also suitably been used for collecting data among participants with low formal education [
53,
54].
As a major strength of our study is seen that we utilized a systematic approach to tool assembly and psychometric evaluation recommended for low resource settings in SSA [
18] to develop a comprehensive culturally appropriate measure of HRB for this setting. Nonetheless, one of our study limitations is that we assembled and psychometrically evaluated the KRIBE-Q using a school attending adolescent sample and yet between 83 and 91% of Kenyan adolescents are enrolled in primary and lower secondary [
55,
56]. Although this was the case, we believe that we satisfactorily captured important HRB constructs and examples from other young people like adolescents who dropped out of school, young adults, and adolescents living with chronic illnesses during the construct clarification phase. Also given the nature of items and their scoring procedure, it was not possible to perform traditional analysis for construct validity such as factorial analysis. However, this study ascertained the face and content validity as well as reliability of the KRIBE-Q. Future studies may explore predictive or criterion validity of this tool. Another limitation stems from the self-reported nature of HRB assessment utilized in our study as self-reports on sensitive topics are at times associated with self-desirability and recall bias [
57]. Self-reports are however universally applied methods for HRB assessment and in our current study we strictly observed confidentiality to counter bias associated with self-reported behavior. Lastly, some of the results from this study, for example on prevalence of HRB, need replication utilizing a larger and more diverse adolescent sample.
Conclusions
The assembled Swahili version of the KRIBE-Q in this study can be considered as reliable and its content is valid for assessing HRB of adolescents in a rural coastal Kenyan setting. Moreover, its mode of administration via ACASI was characterized as easy to answer; private and comfortable; and an enjoyable experience by the adolescents in this setting. This assembled KRIBE-Q comprises 8 components namely: behavior related to unintentional injury and violence; tobacco use behaviors; alcohol and other drug use behavior; sexual behaviors; dietary behaviors; physical activity behaviors; gambling behavior; and hygiene behavior.
We took a systematic mixed method approach for adaptation of tests for resource poor settings. We therefore expect that this assembled questionnaire will be useful in surveys evaluating adolescents’ lifestyle; and in planning and evaluating interventions aimed at addressing adolescent health, especially in resource poor settings at the Kenyan coast where such programs are currently scarce.
Acknowledgments
The authors wish to thank the adolescents and stakeholders that took part in the focus group discussions, in-depth interviews and the psychometric evaluation of content for this HRB questionnaire. We also thank Moses Kachama Nyongesa, Kenneth Rimba, Janet Thoya and Eddy Randu for their support towards the translation and harmonization process. We would like to thank Paul Mwangi for data management and Karabu Ngombo, Richard Karisa, Beatrice Kabunda and Connie Kadenge for their role in data collection. The authors would like to thank the Director of Kenya Medical Research Institute for granting permission to publish this work.