Overcoming cultural barriers to implementing oral HIV self-testing with high fidelity among Tanzanian youth

In Tanzania, few youth seek Human Immunodeficiency Virus testing. Youth were better able to test themselves after receiving culture-specific picture and video instructions.

Abstract

Tanzanian youth have high levels of HIV risk and poor access to HIV-testing. Oral self-implemented testing (Oral-SIT) provides an alternative that reduces barriers to HIV-testing. We examined adaptations to Oral-SIT training components in a randomized experiment to evaluate a “train-the-trainer” strategy for improving comprehension of graphic training materials. Participants (N = 257, age = 14−19 years) were randomly assigned to one of two self-training conditions: graphic instruction book (GIB) or Video-GIB. Outcomes included behavioral performance fidelity, self-reported comprehension, and intentions to seek treatment. Video-GIB participants, relative to GIB-only participants, had higher performance fidelity scores, made fewer performance errors, had better instruction comprehension, and were more likely to intend to seek treatment. Oral-SIT timing errors were significantly more common among GIB-only participants.

Graphic training materials in conjunction with a “train-the-trainer” video has significant potential for increasing Oral-SIT’s reach by overcoming technological and literacy barriers.

Implications.

Practice: This study provides practical knowledge on the adaptation of graphic training methods to youth in impoverished environments.

Policy: The results provide new data to aid in the development of policies directed at increasing the reach of current HIV-testing systems.

Research: This study expands implementation theory to include self-implemented health practices.

INTRODUCTION

Tanzanian youth: HIV risk and testing

Tanzanian adolescents are ranked fifth in HIV prevalence among sub-Saharan countries [1], with the highest prevalence observed among youth age 15–24 years [2]. This is a substantial population segment since more than half of Tanzanians, approximately 30.4 million, are under the age of 20 [3] with a median population age of 17.4 years [4]. Low-income Tanzanian youth have poor access to HIV-testing and care [2, 3, 5–9]; with testing estimates ranging widely for 15–24 year olds [1, 10]. Limited access to HIV-testing leads to delays in receiving care and increases in HIV mortality among adolescents [11, 12].

Oral HIV-testing

HIV-testing is traditionally offered in clinic settings. Barriers to clinic-based testing include social stigma, discrimination, and geographic distance [8, 13–15]. Oral self-implemented testing (Oral-SIT) offers a supplemental strategy for increasing the reach of HIV-testing services for youth. Oral-SIT allows people to self-test under conditions of privacy that avoid stigma and discrimination, increases repeat and partner testing, and, with appropriate dissemination strategies, reduces transportation barriers [16, 17]. The OraQuick test by OraSure is preferred by consumers over other self-implemented tests since the oral swab method is less painful than finger stick methods [17]. Prior studies assess a wide range of Oral-SIT performance outcomes [18, 19]. Findings show that higher levels of education and receiving the training materials in one’s native language are associated with better performance [18, 19]. Research in Kenya and the US illustrate that current “high literacy” variations of Oral-SIT self-training have limitations [19, 20].

Self-implementation fidelity

Oral-SIT requires the person to self-implement a set of training and testing components, and take appropriate follow-up actions. In this regard, Oral-SIT can be understood within a broader implementation science framework (i.e., a special case of program implementation involving self-implementation) [18]. Self-implemented programs place a heavier burden on the consumer because consumers must train themselves. Consequently, self-training materials must be delivered and communicated in a manner that has high levels of program-consumer fit (e.g., cultural, linguistic, cognitive fit [18, 21–24]).

Current, commercially available Oral-SIT training materials include written instructions (i.e., flip charts, inserts, package labels) and an internet-based video. These materials require consumers to process and comprehend potentially unfamiliar and challenging concepts or actions. Additionally, the video requires access to reliable internet technology. Consequently, consumers with low levels of literacy and poor internet access may have challenges using the current Oral-SIT training materials even when materials are in the appropriate language. That is, there may be poor program-consumer fit.

In order to address these challenges, the present study developed graphic training materials (i.e., graphic instruction book [GIB]) that do not require an understanding of the written language or access to internet technology. Graphic training materials have been used successfully in a variety of contexts [25, 26]. Since graphic training materials may be challenging for some youth [19], we developed a video (in Swahili) to train youth on how to use the GIB to perform the Oral-SIT. In addition to instructions on the test itself, the graphic instructions and the supplemental video address linkage-to-care. Linkage-to-care materials included sequences illustrating the importance of seeking a confirmatory test and appropriate treatment when necessary. Motivational material linked seeking confirmatory testing and treatment to positive health and developmental outcomes (e.g., marriage, children). Using a randomized controlled design, we contrasted participants using the GIB-only with those whose employment of the GIB was preceded by the video. Our goal was to develop a training method that had potential to be generalizable across consumer groups and which could serve to increase the reach of Oral-SIT in high-priority communities.

METHODS

Participants and recruitment

We conducted a series of pilot investigations with local youth to inform the development of recruitment materials, procedures and events in the study locale (Tandika area of the Temeke District of Dar es Salaam; see Appendix A). Participants (n = 257) were recruited from a large NGO that served low-income, orphaned, and homeless youth. Over 90% of participants had two or more poverty indicators as reflected in measures of food insecurity, household employment, and transportation access (see Table 1). Eight percent of youth reported not living with family. Overall, Tanzania falls in the bottom 10% with respect to food security in Africa based on the Global Food Security Index [27]. None of the participants had access to at-home or personal computers, and those with a cell phone did not have internet access. Eighteen percent reported ever being HIV tested, and 20% reported being sexually active in the past year. Study procedures, instruments, and materials were approved by the Oregon State University IRB, and the National Health Research Ethics Review Committee of the National Institute of Medical Research (Dar es Salaam, Tanzania).

Table 1.

Background characteristics (total sample)

Demographic (n)a		Testing history and risk behavior (n)
Age/years (259) (n)	Total %d		Total %d
14 (1)	<1%	HIV-testingc (256)
15 (67)	26	Yes (49)	18%
16 (71)	28
17 (58)	23	Sex partnersc (257)
18 (43)	17	0 (205)	80
19 (16)	6	1 (38)	15
		2+ (14)	5
Education (253)
Primary (31)	12	Condom usec (52)
Secondary (217)	86	None (11)	21
Advanced (5)	2	Some (23)	44
		All (18)	35
Gender
Male (132)	51	Needle injectionc (257)
Female (127)	49	Yes (15)	6
		Clean needlesc (15)
Poverty indexb		Yes (15)	100
0 (5)	2
1 (14)	9
2 (54)	26
3 (90)	43
4 (46)	22

Note: age was estimated by converting birth year to years lived, education (in school, Yes/No; class level completed: primary, secondary, advanced secondary), gender (male, female); sexual orientation based on gender of sexual partners/past year]. HIV-testing (ever), numbers of sexual partners (past 12 months), condom use (condom use/past 12 months: none, some, all of the time with all sex partners), use of needles to inject drugs or medications, and needle cleaning were assessed (needle use/past 12 months: [Yes, No]; [If Yes] clean needles: [Yes, No]).

^aSample sizes vary due to nonresponse or audio recording problems.

^bPoverty index: total sum of 4 poverty indicator items (food insecurity, unemployed household, household lacks motorized transportation, participant lacks transportation options). The poverty index included two household-family indicators which excluded participants who are not living with a family (n = 21). One family question asking about family employment had 12% nonresponse. Most indicators were relatively complete, for example, food insecurity was assessed for 98.9% of all participants (66% reported food insecurity). Missing data/non-response did not differ significantly between conditions.

^cPrimary: completed/currently attending primary school; secondary: completed/currently attending secondary school; advanced: completed/currently attending advanced school; HIV-testing, sex partners, condom use, use of needles to inject drugs or medications, always clean needles or use of a needle from sterile package were assessed for the past 12 months. Condom data are reported only for participants sexually active in the past year.

^d% do not add to 100 due to rounding.

GIB and video development

The GIB and video (see Appendix A) were developed with input from Tanzanian youth and local personnel, and featured a young-adult female Tanzanian actor. The video was presented in Swahili and designed to enhance the understanding of the GIB. Video-GIB participants viewed the video and, subsequently, used the GIB to perform the Oral-SIT. GIB and video content covered all key training steps, and information or motivational material.

Motivating linkage-to-care

The GIB and video illustrated that those who tested HIV-positive should go to a doctor to obtain medication, and take the medication daily in order to grow to adulthood, engage in adult roles (e.g., work, marriage, parenthood), and become an elder. Based on the perspectives of local youth and NGO personnel, these are life-goals of general importance to Tanzanian youth. These life goals were identified by Nalkur [28] and through interviews with youth and young adults in a 2014 pilot study [29, 30].

Study design and procedures

Primary data collection took place in April and May, 2016. Participants (n = 257) were randomized (14–19 years; see Table 1) to two conditions (GIB condition, n = 125; GIB and Video condition n = 132) (see Appendix A). Four trained Tanzanian male and female interviewers were gender matched to the participants. Study procedures were conducted in private rooms provided by the study NGO; interviews were audio-recorded and simulations evaluated by direct observation. The OraQuick testing solution was desensitized by OraSure for use in the present study so the tests would not provide results on HIV status. We used a desensitized test to control for the effects of HIV-testing anxiety on learning and to facilitate recruitment of individuals who might be anxious about HIV-testing. After obtaining verbal consent, participants were asked to summarize what participation entailed to ensure their understanding of study procedures, particularly that they would not be receiving an actual HIV test.

Following consent, all participants completed (a) a brief pre-simulation interview, (b) the Oral-SIT simulation (see Appendix A), and (c) a post-simulation interview and debriefing session. Under the Video-GIB condition, participants viewed the video on a laptop, were offered an opportunity to see it again (16% watched more than one time), and then used the GIB, without further access to the video, to complete the Oral-SIT. Under the GIB-only condition, participants were introduced to the GIB, which they used to complete the Oral-SIT. After participants completed the main study, interviewers responded to questions, corrected misperceptions, and offered referrals. If participants wanted immediate counseling, they were referred to staff personnel who were trained HIV test counselors. The simulation was constructed to (a) ensure that participants were exposed to all training components, and (b) mirror self-paced training and performance aspects of real-world self-testing. Similar procedures have been employed previously [31]. The design of the present study allows for more evaluative-control than naturalistic study designs (e.g., naturalistic designs allow participants to skip some or all of the training components; e.g., MacGowan et al. [31]).

Measures

Comprehension of Oral-SIT test results and concepts

Because the study used a desensitized Oral-SIT, we assessed ability to read the test results by presenting respondents (post-simulation interview) with flashcards depicting the different types of test results as they would appear on an Oral-SIT specimen swab (i.e., HIV-negative, HIV-positive, and an indeterminant results). Presentation of the test result stimulus cards was randomized (see Appendix B). These procedures were successfully used by MacGowan et al. [31]. We assessed if participants understood the meaning of the technical terms HIV-positive and HIV-negative correctly (e.g., “Do you know what it means to get a positive HIV result?”). We avoided using technical terms in our training materials but instead introduced concepts derived from pilot research such as, “this result means you should go to a doctor” or “this result means you do not need to go to a doctor.”

Oral-SIT performance fidelity

The fidelity measure was based on interviewers’ observations of correct/incorrect behaviors in performing the Oral-SIT (see Appendix B), plus participants post-simulation evaluations of the test result stimulus cards (total correct: range = 0–8 points).

GIB misunderstandings

Participants were asked in the post-simulation interview to review each picture or picture series in the GIB and indicate if they found that picture or picture series to be confusing or difficult to understand. We summed the number of pictures or picture series that participants found to be confusing/difficult for a total score.

Confirmatory testing and linkage-to-care

Participants were asked a series of questions to determine whether they would seek confirmatory testing and treatment. We asked (confirmatory testing), “These next questions ask about what you would do if you found out you had an HIV test result that showed you should go the doctor”; (linkage-to-care) “If you took an HIV test like the one you took today and thought you had the HIV virus, how soon would you go to a nearby clinic?” (would not go at all = 0, would go right away to a month later = 1); “If you were told at the testing clinic you had the AIDS virus, how long would you wait before going to get the [AIDS treatment] pills from this location?” (would not go at all = 0, would go right away to a month later = 1).

Data management and analysis

Audio recordings were transcribed from Swahili, with responses being translated into English and inserted into a parallel English version protocol. Initial transcriptions were conducted by two Swahili-English Tanzanians with consulting from an English–Swahili study investigator (N.F.). Reliability checks were conducted on the full transcription process. All quantitative data were entered into Excel files and uploaded into Statistical Package for the Social Sciences for data management purposes, and Stata for final computational purposes.

RESULTS

Randomization and engagement

The randomization procedures successfully balanced conditions on key background variables. Treatment groups did not differ significantly on demographic/risk variables, in their understanding of the concepts of HIV-positive and HIV-negative, their knowledge of HIV transmission routes (sexual contact, blood transmission, mother-to-child), or the need to seek medical care if HIV-positive (data available from first author). No significant differences were found between conditions in participants’ engagement of the GIB with respect to their ability to carefully review the GIB prior to and during the test (GIB vs. Video-GIB: 98% vs. 100%; and 94% vs. 91%, respectively; p = .12). All Video-GIB participants were judged to have reviewed the video carefully.

Performance fidelity

Video-GIB participants had significantly higher scores, relative to GIB-only participants, in terms of total performance fidelity scores, and made significantly fewer errors with regards to tube spillage and correct swabbing, use of the timer, and identification of the test result cards (see Table 2). Conditions did not differ with respect to correctly opening the test tube, and correctly placing the specimen stick in the tube. These findings suggest that the video component improved comprehension of more complex behaviors and concepts (e.g., using the timer correctly).

Table 2.

Comparison of overall Oral-SIT performance fidelity scores and individual Oral-SIT behaviors by condition

Variable	GIB-only	Video-GIB	GIB vs. Video-GIB
Total fidelity M (SD) (n)	5.73 (1.49) (124)	7.59 (0.83) (131)	F = 52.402**
Oral-SIT behaviors
Open tube correctly (259)	95%	97%	NS
Didn’t spill tube contents (259)	95	99	χ  2 = 3.957*
Swabbed correctly (259)	87	100	χ  2 = 18.001**
Spec. stick in tube correctly (259)	94	97	NS
Used timer correctly (259)	45	93	χ  2 = 70.817**
Reading test results
Positive result (255)	60	91	χ  2 = 33.616**
Negative result (257)	41	88	χ  2 = 62.533**
Indeterminate result (257)	56	92	χ  2 = 45.080**

*p = .05, **p = .0001.

Note: total fidelity = a sum of all correct Oral-SIT behaviors; test results were read from photographs showing the specimen stick with the relevant test strips indicating a positive, negative, or indeterminate result. As described in the methods the test results were described in terms of “a test result that means you have to go to a doctor” or “a test result that means you do not have to go to a doctor” or “a test result that means you need to take the test again.”

GIB misunderstandings

Participants evaluated GIB picture sequences and indicated which pictures were confusing/difficult (see Table 3). Video-GIB participants (vs. GIB participants) reported significantly fewer pictures to be confusing/difficult. Post-hoc tests revealed that for pictures with a high prevalence of misunderstandings (>5% misunderstood), Video-GIB participants reported significantly fewer challenging pictures/sequences than GIB-only participants (see Table 3).

Table 3.

Differences between conditions in confusion/misunderstandings of GIB pictures

Variable (n)	GIB-only	Video-GIB	Significance
Total pictures (257): M (SD)	9.30 (7.38)	4.29 (4.89)	F = 21.11****
Specific pictures (257)
Spec. applicator	10%	4%	χ  2 = 4.31*
Swab upper gum	11	5	χ  2 = 2.98†
Timer	13	6	χ  2 = 5.59**
Setting timer: 20 min	39	11	χ  2 = 28.37****
Start timer	26	10	χ  2 = 11.03***
Waiting for time to expire	26	7	χ  2 = 16.89****
Timer stops: remove spec. stick	10	2	χ  2 = 7.26***
Examine spec. results window	7	2	χ  2 = 3.50†
Negative result	19	9	χ  2 = 5.45*
Positive result	23	11	χ  2 = 6.34**
Indeterminate result	10	4	χ  2 = 3.51†
If HIV-positive: LTC	19	3	χ  2 = 17.29****

*p = .05, **p = .01, ***p = .001, ****p = .0001; †p ≤ .10 and p > .06.

Note: total pictures: M number of pictures reported as confusing or very difficult to understand; spec. applicator: picture shows a close up of the specimen stick applicator; swab upper gum: actor places a specimen stick onto her upper gums (there were no significant differences between groups in confusion of movement of the specimen stick or application of the specimen stick to the lower gum); timer: the actor displays kitchen timer for timing the test; setting timer: 20 min.: the actor adds 20 min to the timer; start timer: the actor presses the timer start button; waiting for time to expire: the actor is seen waiting while the timer ticks down to zero; timer stops: remove spec. stick: the timer shows zero and the actor removing the specimen stick from the reagent tube; examine spec. results window: the actor removes the specimen stick and examines the applicator window; negative result: a close up of the applicator window showing a negative HIV test result and a picture indicating that you do not need to see a doctor; positive result: a close up of the applicator window showing a positive HIV test result and the actor visiting a doctor’s office; indeterminate result: a close up of the applicator window showing an indeterminate result and a series of pictures showing that you need to retake the test; if HIV-positive: LTC: a close up of the applicator window showing a positive result and visiting a doctor’s office and receiving a bottle of medication from the doctor.

Confirmatory testing and LTC

Approximately half of the participants in both conditions understood the need for obtaining a confirmatory HIV test result if they tested HIV-positive on the Oral-SIT (GIB: 52%, Video-GIB: 49%; p > .10). However, participants in the Video-GIB condition were significantly more likely to indicate that they would seek treatment if they tested HIV-positive on an Oral-SIT (66%) relative to the GIB-only condition (46%) (χ² = 9.75**, p = .01).

DISCUSSION

Intervention effects

In general, participants in the Video-GIB condition had significantly superior comprehension and fidelity relative to those in the GIB-only condition. Improved comprehension of the GIB in the Video-GIB condition, relative to the GIB-only condition, supports the contention that the video enhanced understanding of the GIB. Further, the results suggest that the impact of graphic training materials on performance fidelity can be enhanced through the use of “train-the-trainer” techniques, such as those employed in our video. Differences between conditions varied somewhat by Oral-SIT procedures. The largest differences were observed for correct timer use and test result interpretations. For instance, 91% of individuals in the Video-GIB condition correctly identified an HIV-positive test result, while only 60% in the GIB condition correctly identified this result. Overall, these results suggest that the video was successful in training participants to better understand and execute GIB instructions. In both conditions the most frequent errors were with timing the test. There is a need to improve the timing procedures beyond the current method of using watches, clocks, or timers and recording the start and stop times. Problems with timing the Oral-SIT test have been observed in other populations [32, 33].

We sought to motivate intentions to seek confirmatory testing and linkage-to-care by connecting these health actions to positive, culturally relevant developmental images associating treatment with longer life and achieving adult roles (e.g., getting married, finding work). Our results suggest that these images had a significantly more powerful effect on the Video-GIB condition than the GIB condition with regard to linkage-to-care (Video-GIB: ≈66%; GIB-only: ≈46% would seek HIV-treatment). Conditions did not differ, however, with regard to intentions to seek confirmatory testing (≈50%/condition). Because people seeking treatment would typically be re-tested, we speculate that the proportion seeking confirmatory testing in the Video-GIB condition would be larger than the 50% figure (i.e., 66%; 50% ≈ 16% larger). Additional research should be directed at increasing youths’ understanding of and intentions to seek confirmatory testing/treatment.

Recent studies [34–37] have employed a Swahili version of the English–Spanish Oral-SIT training components produced by OraSure Corporation. These training components include a combination pictorial-print fold-out instructions and an online video. Only one study, to date, included an evaluation of these adapted training components [36], but the evaluation was limited to video recordings of a sub-sample of 20 respondents. In addition, none of the studies report that participants used the video training component, so the value of the Swahili video is unknown. Moreover, participants in these studies have typically been adults who have previously received HIV-testing (i.e., a history of HIV-testing has been found to influence Oral-SIT fidelity [32]). Rigorous evaluations of Oral-SIT adaptations are needed. Future evaluations should include a wider range of populations, including studies of impoverished youth.

Oral-SIT dissemination

The training techniques examined in the current investigation may be generalizable to large-scale dissemination efforts. For instance, NGOs disseminating Oral-SITs to youth might employ a single video display to educate small groups of youth on how to use the GIB. We would anticipate that such an approach does not require individual users to have computer access. Subsequently, additional Oral-SIT and GIB dissemination might take place through youths’ informal social networks. This process is facilitated by the fact that the GIB is portable and does not require technology or the ability to read. A limitation to disseminating Oral-SIT in warm climates is that the kits can become unreliable when stored at temperatures over 80°C. Research addressing distribution of Oral-SIT in warm climates is needed.

Study limitations

Our opportunistic sample limits generalizability. Additionally, the simulation does not fully reflect naturalistic circumstances. Although the simulation included some elements that mirror what occurs in a natural setting (e.g., self-pacing), the present study used direct observation which may induce demand characteristics resulting in better performance than what might occur in real-world settings [31]. Nevertheless, alternative methods such as video recordings and self-report assessments also have limitations [31].

In the present study, we controlled for HIV test-related anxiety (participants were knowingly using a desensitized HIV test). These procedures were intended to allow us to examine the effects of participant comprehension on performance of the Oral-SIT without being confounded by anxiety, and reduce potential barriers to recruitment (i.e., to reduce nonparticipation related to fears of being tested). In this regard, we recognize that Oral-SIT training and performance may have occurred under more optimal effective conditions. Lastly, the study design does not include all potential control groups (e.g., a video-only control group). We cannot, for instance, separate the unique effects of the video on fidelity independently of the effects of the GIB on fidelity. Nevertheless, we believe the improved comprehension of the GIB in the Video-GIB condition suggests that the video functioned as designed (i.e., to improve GIB comprehension).

Despite these limitations, simulation studies provide a basis for identifying problems and conducting preliminary tests of solutions. Our perspective is that simulation studies and naturalistic studies should be conducted in tandem. That is, new training strategies identified through simulation studies should be field tested under naturalistic circumstances. Naturalistic studies [31] are useful for identifying performance deficits under real-world circumstances, though they may not provide direct insight into potential solutions for addressing these deficits.

Recommendations and conclusions

The current study evaluated alternative training procedures that have significant potential to increase the reach of Oral-SIT by overcoming technological and literacy barriers. More generally, Oral-SIT offers a potential solution to the challenges to venue-based HIV-testing for adolescents in Tanzania (e.g., privacy [7], social stigma [13, 14], age biases [8], transportation problems [15]). Oral-SIT provides an opportunity to increase privacy, reduce HIV-testing related social stigma, and increase the understanding of early treatment. The current Video-GIB self-training adaptations are at an early stage of development (T1-Studies: Investigations of promising interventions and intervention components; [38]) and additional research is needed to further refine and test these adaptations.

Public health implications

Increasing the reach of current HIV-testing services in Tanzania among adolescent and young adult populations is important to reducing the transmission of HIV and HIV-related morbidity and mortality in this population. Oral-HIV self-testing offers an avenue for increasing the reach of HIV-testing services, but it is crucial that adolescent and young adults be able to perform this test with high fidelity. The current study provides important data on strategies that may be used to enhance fidelity in performing the Oral-HIV self-test.

APPENDIX A

Participant exclusions:

Of 274 participants, 15 were ineligible (i.e., incapable of understanding the informed consent instructions, demonstrated significant cognitive impairment or were under the influence of drugs/alcohol); 2 participants were in sessions experiencing equipment failure.

Recruitment:

Recruitment posters and flyers were distributed throughout the Tandika area of the Temeke District. Eligible participants were offered a variety of incentives identified by local youth as appropriate including food packets, T-shirts, soccer balls, and phone cards. A large recruitment event was conducted in Mwembe Yanga Park approximately a week before study onset.

GIB and video development:

The GIB and video are available in electronic form from the first author. The video demonstrated use of the GIB while correspondingly demonstrating the Oral-SIT procedures. The intention was to enhance understanding of the GIB so that participants in this condition could view the video and subsequently use only the GIB to perform the Oral-SIT. This would be similar to, for instance, showing the video to a small group of youth and providing them with picture instruction books and kits to use at another time.

Design and procedures:

A total of 136 females and 129 males took part in the main study. All data were collected at a single NGO (FASDO) in the Tandika area of the Temeke District (Dar es Salaam). The NGO provides recreational and arts opportunities and instruction for youth in the district, and is well-known and respected. During the verbal consent process, the interviewer read a script that explained study procedures and participant rights in detail. To ensure that each individual was comprehending all of this information, the interviewers stopped after each topic to allow the individuals to ask questions and to provide any clarification. After reading the verbal consent document, the interviewers asked the individuals to summarize what they would be doing if they participated in the study and ensured that the individual understood that they were not taking an actual HIV test. If the individual had difficulty understanding the information given in the verbal consent process, the interviewer could read it again or summarize the information. The simulation followed the procedures described below.

Following the introduction to the training materials, participants reviewed either the GIB or the video and then the GIB depending on condition assignment:

GIB-only:

After a brief warm-up period participants in the GIB-only condition received a brief summary of the purpose of the study and the fact that the HIV test they were about to take would not tell them if they actually have HIV. Following the summary participants were introduced to the GIB and instructed to look through the book, take as much time as they wanted, and hold any questions until the end of the study period. The participant was then instructed to use the GIB to guide them through the testing process and the participant was given the test kit and a kitchen-style timer. Participants were again reminded that they could use the GIB to guide them while they went through the test procedures. Our pilot research demonstrated that many of the youth did not have a method of timing the test, and they would be willing and capable of using a kitchen-style timer if they were shown how to use the timer; both the GIB and the video demonstrated timer use, how to set 20 min on the timer, and how to start and stop the timer. The timer and instructions on how to use it were described in the GIB. The GIB directed participants to remove test materials from a small drawer at the bottom of the test kit containing a tube filled with a testing reagent and a specimen swab. The GIB further instructed participants to perform each step in the oral testing process from opening the packet containing the reagent tube to timing the specimen stick in the reagent tube for 20 min using the timer provided. During the 20-min wait period, participants were given either computer games or playing cards to pass the time. The interviewer remained in the room during the entire oral test process, making observations, and conducting the post-simulation interview and debriefing period.

• Participants were free to engage the GIB training component to whatever extent they liked. Consequently, the simulation allowed for real-world conditions with regard to exposure to a training video at a kit dissemination site followed by open-ended self-implementation of the GIB when administering the test.

• Participants followed the GIB instructions without assistance in completing the Oral-SIT.

• While the participants were using the Oral-SIT kit, interviewers made behavioral observations and notes on how engaged participants were in the instruction materials, and noted any challenges participants were having. Similar procedures have been employed previously, and are considered to be superior to self-reports, and potentially superior to video recording [31]. Specifically, the interviewer observed if participants opened the tube and transferred it appropriately without spilling a significant amount of the reagent, and if participants removed the oral swab from the packaging, and were careful not to touch the specimen swab with their fingers or to place it on the countertop. The interviewer then observed if participants obtained an oral specimen by first swabbing the upper gum and then, after reversing the swab, from the lower gums. Next the interviewer observed if participants placed the swab into the tube containing the reagent such that the specimen portion of the stick was fully submerged in the testing solution. The interviewer observed if respondents correctly set the timer for the initial 20-min time interval, and started the timer.

• Since participants understood that this was not a real HIV test and the reagent had been desensitized (we verified this understanding during the consent process), we assessed ability to read the test results by presenting respondents with randomly ordered flashcards (post-simulation interview) depicting the different types of test results as they would appear on an Oral-SIT specimen swab (see below). These procedures were successfully used by MacGowan et al. [31].

Video and GIB:

participants in the video and GIB condition received a brief summary as in the GIB condition and then were introduced to the video (presented on a laptop) and instructed to watch the video and upon completing this task participants were offered the opportunity to watch the video or portions of it again as needed. The laptop was then put away and participants were given the GIB and told this was the book that they saw in the video and they were to use this book as a guide to completing the Oral-SIT, and to hold questions to the end of the simulation. The participant was then given the test kit and a kitchen-style timer. The timer and instructions on how to use it were described in the GIB. Participants were reminded that they could use the GIB to guide them while they went through the test procedures. All subsequent instructions and procedures paralleled those in the GIB condition.

APPENDIX B

Instrument translation and development:

the translation process included forward and backward translation by bi-lingual (Swahili/English) native Tanzanians (two translators) with consultation and mediation from a study investigator who is English–Swahili fluent (N.F.). This work was also guided by input from field staff and focus groups/pilot studies with Tanzanian youth. A number of sexual behavioral items were obtained from Sommer et al. [39] and adapted for use in the current study.

Interviewers and interviewer training:

Interviewers were trained over a three-week intensive process and then participated in a series of pilot interviews with observation and feedback. Observation and feedback was provided at regular intervals during data collection.

HIV test cards and Oral-SIT performance fidelity observations:

the procedures for administering the test result stimulus cards (cards depicted HIV-negative, HIV-positive, and an indeterminant results) were as follows: Interviewer: “Here are three pictures of the different test results.” [Interviewer: lay out the three test result images]; “Which test result means that you do not have to go to the doctor?,” “Which test result picture means the test didn’t work, and you need to take another test?,” “Which test result means that you should go to the doctor?.” The stimulus cards were presented in a random order to each participant. The performance fidelity measure is composed of interviewer observations of the following,

• R correctly opened the test tube and placed it in the kit holder.

• R did not spill the contents of the tube between opening and transferring to the holder.

• R swabbed her/his gums correctly (without contaminating the test swab).

• R placed the specimen stick correctly into the test tube.

• R correctly set the timer to 20 min and started the timer.

• R correctly evaluated the negative test result card.

• R correctly evaluated the positive test result card.

• R correctly evaluated the indeterminant test result card.

Participants who spilled only a small amount of liquid reagent, in the interviewers’ judgment (as addressed in training), were not coded as having made an error since only a small amount of spillage will still result in an accurate result; swabbing correctly included variants judged to produce an accurate sample (e.g., gentle brushing of the gums, circular movements on different areas of the gum).

Compliance with Ethical Standards

Conflict of Interest: none declared.

Authors’ Contribution: Tori Geter, Kulindwa Mhandagani (data management), Johnson Chiganga, Grace Nanyaro (field interviewers), Tedvan Chande, Sylvia Mamkwe, MD (NGO and public health liaisons).

Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed Consent: Informed consent was obtained from all individual participants included in the study. This article does not contain any studies with animals performed by any of the authors.