Testing strategies for couple engagement in prevention of mother-to-child transmission of HIV and family health in Kenya: study protocol for a randomized controlled trial

Background: HIV-related maternal deaths and HIV infection among infants remain unacceptably high across sub-Saharan Africa despite increased antenatal care attendance and provision of antiretroviral therapy to pregnant women. In the Jamii Bora (“Better Family” in Swahili) Study, we seek to test the efficacy of an interdependence theory-based couple intervention. The intervention reaches pregnant women and male partners through home visits by male-female pairs of lay health workers. The aim is to increase access to home-based couples’ HIV testing and counseling services to improve family health. Methods: This is a three-arm randomized control trial among 1080 pregnant women 15 years of age or older, living with their male partners, and who have not undergone couples’ HIV testing and counseling in Kisumu and Migori Counties in Kenya. Couples will be randomized into three groups: home-based couple visits, HIV self-testing kits for couple use, or standard care (male partner clinic invitation letters). Participants will be followed up to 18 months postpartum. The study has three aims: in aim 1, we will determine the effects of the intervention on our primary outcome of couple HIV testing, compared to HIV self-testing kits and standard care; in aim 2, we will examine the intervention impact on HIV prevention behaviors, facility delivery, and postnatal healthcare utilization, as well as secondary health outcomes of maternal viral suppression and HIV-free child survival up to 18 months for couples living with HIV; and in aim 3, we will compare the cost-effectiveness of the home-based couple intervention to the less resource-intensive strategies used in the other two study arms. Assessments with couples are conducted at baseline, late pregnancy, and at months 3, 6, 12, and 18 after birth. Discussion: The results from this study will inform decision-makers about the cost-effective strategies to engage pregnant couples in the prevention of mother-to-child transmission and family health, with important downstream benefits for maternal, paternal, and infant health. Trial registration: ClinicalTrials.gov NCT03547739. Registered on May 9, 2018

This study will be conducted in 18 ANC health facilities, half in Kisumu County and the other half in Migori County. Kisumu County (19.3%) and Migori County (14.7%) are among the top four highest HIV-burden counties for adults in Kenya [33]. Maternal mortality in the counties is high with Kisumu County at 465 [34] and Migori County at 673 per 100,000 live births [35]. The two counties account for 13% of new infant HIV infections in the country and 15.13% of the need for PMTCT services [4]. Although the vast majority of women who present at ANC clinics in Kisumu and Migori counties are tested for HIV, the number of pregnant women testing for HIV at the national level has been reducing in recent years to stand at 56.8% in 2017 with a high rate of MTCT at 11.5% [3]. The two study counties report lower than national rates of MTCT at 8.7% for Kisumu and 7.2% for Migori County, and there is a significant dropout of women and infants along the PMTCT cascade [36]. The clinics in this setting provide integrated antenatal care/maternal and child health (ANC/MCH) and HIV services and are implementing the Option B+ strategy, in which all pregnant and breastfeeding women are immediately initiated on life-long ART, regardless of CD4 count or HIV disease stage. We adapted the Interdependence Model of Health Behavior Change to understand the mechanisms through which this intervention may impact health outcomes (Fig. 2 [37]. This model extends beyond an individually based understanding of health behavior change (e.g., health beliefs, self-efficacy) by positing that both partners influence one another’s health decisions and behaviors and emphasizing that positive relationship dynamics, e.g., communication and commitment, are inherent to a couple’s ability to engage in joint decision-making aimed to improve health outcomes for the couple [38]. Detailed study design The following are the inclusion criteria: The following are the exclusion criteria: Women who report recent severe intimate partner violence as assessed by the baseline questionnaire responses to the dyadic version of the WHO intimate partner violence measure [39] are included in the study assessments, but their male partners are not contacted for enrollment, and the couple is not included in the randomized part of the study. Pregnant women presenting at participating ANC clinics in Kisumu and Migori Counties who meet study the inclusion criteria are asked if they would like to participate in a study about approaches for supporting pregnant couples on family health issues (including HIV) during pregnancy and postpartum. If interested, a lay health worker obtains informed consent for study participation, conducts the baseline questionnaire, and obtains permission for contacting her male partner. With the woman’s permission, we subsequently contact her primary male partner, arrange to meet with him in a community location, obtain informed consent, and conduct his baseline questionnaire. The home-based couple intervention includes five home visits (3 main and 2 booster visits) laid out as follows: two during pregnancy, one at 6 weeks after the birth; two booster sessions, one at 6 months after the birth and one at 12 months after the birth. At each home visit, pairs of lay health workers (one male and one female) meet with the woman and her partner together at their home (or another preferred community site) and deliver key intervention elements as shown in Table 1. The key elements include discussing family health promotion, HIV-related services including offering CHTC, couple communication skills such as active listening, and active linkage to nearby clinic-based services for family health and HIV prevention and treatment services (including PrEP for discordant couples). Intervention content at each couple home visit • ANC visits • Nutrition • Malaria • Male partner support during pregnancy • IPV • Mental health • Birth plan for HF delivery • Danger signs • Infant feeding • Male partner support for birth • What to expect • Infant health visits and immunizations • Family planning • Male partner support after birth • Postpartum checkups • Infant health • Infant feeding • Family planning • Male health • IPV • Mental health • Infant feeding • Infant visits and immunizations • Family planning • Couple HIV testing • PMTCTb • PrEPa • Linkage to careb • Repeat testing • PMTCTb • PrEPa • Linkage to careb • Infant HIV testingb • PMTCTb • PrEPa • Linkage to careb • Repeat testing • PMTCTb • PrEPa • Linkage to careb • Infant HIV testingb • PMTCTb • PrEPa • Linkage to careb • Introduction of CHTC • Linkage to HIV care/PrEP • Offer of CHTCc • Linkage to HIV care/PrEP • Offer of CHTCc • Linkage to HIV care/PrEP • Offer of CHTCc • Linkage to HIV care/PrEP • Offer of CHTCc • Linkage to HIV care/PrEP aFor discordant couples bFor couples living with HIV cFor those who have not done CHTC yet or who need to repeat testing Over 87% of pregnant women in this region live with their male partner [40], making home visits an optimal approach to reach the couples. Recognizing that some pregnant women may live in extended family households where privacy is difficult to maintain [41], we also identify a location in each community for couple sessions that participants may choose if privacy cannot be assured in the home. The use of lay health workers as couples’ counselors contributes to sustainability, given the inadequate availability of professional counselors in this setting. The comparison groups are the provision of HIV self-testing kits (HIVST) for the couple (pregnant woman and male partner) and standard of care. We chose HIVST in order to compare our home-based couple intervention with a promising, but less resource-intensive, approach for encouraging couple and male partner testing. Standard of care was chosen as a comparison group in order to contrast our intervention with the approach currently being used at Kenyan ANC clinics. We have two comparison arms: HIV self-testing kits and standard of care. In the HIV self-testing kits arm, pregnant women receive pairs of HIVST for themselves and their male partners. Trained study staff will provide four oral fluid-based rapid HIV test kits (OraQuickRapid HIV-1/2 Antibody Test, OraSure Technologies) during pregnancy and up to four additional kits after birth. Each test is accompanied with an instruction sheet that describes the step-by-step self-testing procedures in multiple languages. The study staff provide participants with a brief demonstration of how to use the tests and encourage them to offer test kits to their partners to undertake the testing together if they feel comfortable doing so. They are also counseled on how to talk to their partners about HIV testing and the possibility of any adverse partner reactions. Women are free not to accept the test kits or to not introduce them to the male partner if they do not feel comfortable doing so. Following Kenya’s HIV testing services guidelines, participants are instructed to seek clinic-based confirmatory testing if a reactive self-test result is obtained, and an invitation for confirmatory testing at a clinic is included with each test. Participants are also encouraged to report the results of the couples’ HIVST use to the study staff. All women in the standard care group are encouraged by the clinic staff to come to the clinic with their male partners and are routinely asked to undergo CHTC. The clinic staff use letters and other methods to invite the male partner to ANC at the next visit. Couples coming together to ANC are given priority to receive services so that they can avoid queues and rapidly be seen by clinic staff. Our initial assessments indicated that less than 25% of ANC clients participate in CHTC during ANC visits. The primary outcome measure is change in couple HIV testing uptake from baseline to 12 months postpartum. This is assessed in the questionnaires for all couples regardless of HIV status at baseline and each follow-up until 12 months postpartum and confirmed through home visitors’ reports and medical records. The change in couples’ HIV testing uptake during the observation period (from baseline to 12 months postpartum) is coded as Y/N assessed through direct observation in the intervention arm and surveys in the comparison arms. Secondary outcomes include HIV-free child survival at 18 months and maternal viral load suppression at 18 months assessed from medical records (for HIV-positive women only); the number of new male HIV-positive diagnoses coded as Y/N assessed during the observation period and confirmed from medical records; family health service utilization; men and women’s linkage to HIV care, enrollment, and adherence to care assessed from medical records; and PrEP uptake for discordant partners assessed in follow-up questionnaires and medical records. Our target populations are pregnant women identified in the ANC facilities and their male partners. We include women at 36 weeks of pregnancy or less, to have time to provide at least one home visit during pregnancy. The couple testing uptake outcome (primary outcome) will be assessed including both HIV-positive and HIV-negative women/couples. All three arms will be compared with each other (3 comparisons), so our type I error rate is 0.05/3 = 0.017 (two-sided). With three repeated measurements (baseline, 3 months, and 12 months) and compound symmetry covariance structure, the correlation between the observations on the same subject was assumed to be 0.50 as a mid-way between the range of 0.00 to 1.00. With N = 300 couples in each arm, our study will have > 80% power to detect statistically significant differences in couple testing uptake of 30–40% in either intervention arm and 23% in the control arm, based on CHTC rates obtained in the Jamii Bora pilot study [42] (Table 2). The secondary outcomes are HIV-free child survival at 18 months and maternal viral load suppression at 18 months. HIV-free child survival up to 18 months will be assessed only in HIV-positive women with live births, with approximately N = 200 in each arm. Based on prior estimates from sub-Saharan Africa [43, 44], we expect proportions ≥ 90% of HIV-free survivors in each arm. As there are two comparisons (each intervention arm vs the control arm), we set our type I error rate to 0.05/2 = 0.025 (two-sided) when calculating power. If arm 3 (standard care) has a survival rate of 91%, the study has a power of > 80% when arms 1 (home-based couple intervention) and 2 (HIVST) have survival rates of 98% or higher. The other secondary outcome of maternal viral load suppression at 18 months is assessed only in HIV-positive women with around N = 200 in each arm. In Table 2, if arm 3 has a maternal viral load suppression rate of 85%, the study has a power of > 80% when arms 1 and 2 have maternal viral load suppression rates of 95% or higher. Statistical power for comparison of outcomes among study arms We estimated that approximately 120 HIV-positive pregnant women could be recruited from each of the 18 ANC clinics in 24 months (n = 960) in Kisumu and Migori Counties, Western Kenya. Using a stratified randomized design, we are recruiting HIV-negative women (n = 480) in balanced numbers to HIV-positive women (1:2) each month (total HIV-positive and HIV-negative N = 1440). Based on the rates achieved in the pilot study, we conservatively estimate a 75% participation rate for male partners in the study resulting in a total sample size of 1080 couples and estimate up to 17% loss-to-follow-up of couples by final follow-up resulting in approximately 900 couples (300 couples in each randomized group) for analysis. We recruit pregnant women attending ANC clinics to participate in the study until we have achieved a sample size of 1080 randomized couples—1080 women of which two thirds will be HIV-positive at baseline and 1080 male partners. Couples completing baseline interviews are randomized to one of the three approaches to increase couple engagement in HIV prevention and maternal and child health. Couples receive a sealed envelope labeled with their newly assigned study ID numbers, which contain their random assignments that are computer-generated and stratified by clinic and couple HIV status. Blocked randomization with randomly permuted block sizes is used to assure approximately equal numbers in each study arm and in each HIV status group in any given time period. We do not randomize women or men reporting severe intimate partner violence (IPV) at baseline, indicated by responding yes to six items about severe physical or sexual violence during pregnancy as measured through the WHO multi-country study instrument [45]. Instead of randomization, these participants are referred directly to local support services and are invited to take part in the study follow-up assessments to ensure their continued safety and well-being. Participants reporting severe depression, indicated by a score of 20 or greater on the Patient Health Questionnaire (PHQ-9) [46], are randomized and referred to support services. There is limited empirical evidence that couple-based interventions could exacerbate violence or depression, but we opted to err on the conservative side (i.e., excluding high-risk women from the intervention part of the study). All participants will be asked to provide informed consent for data abstraction from their medical records. After randomization, a lay health worker obtains detailed locator information, including cellphone contacts. If randomized to the home visit intervention arm, the worker consults with them about optimal times for a home visit. The HIV self-testing kits arm consists of the distribution of pairs of self-test kits to women at up to 4 time points (twice during pregnancy and twice after the birth). The standard of care arm offers standard clinic-based services, including the standard practice of giving the pregnant woman a letter (or other modes of communication) to invite her male partner to the clinic, and the option for women and partners to return to the clinic for male partner HIV testing or CHTC. We will collect data from multiple sources, including the following: Each participant is reimbursed for each assessment visit (questionnaires), but not for intervention activities, at the rate of approximately 500 Kenyan shillings (roughly 5 US dollars) paid in cash per assessment. This reimbursement is in accordance with other studies being conducted in the area. Participants in the home visit study arm receive a small gift of approximately 200 KSh value ($2 US) at each home visit, which is a cultural expectation for persons visiting the home of pregnant/postpartum couples. Table 3 shows in detail the variables and the measurements that are assessed at various stages in the course of the study. Factors and measures to be assessed in data collection Secondary outcome measures and assessment are listed in the table above: 1. HIV re-testing [time frame: baseline, 3 months postpartum, 12 months postpartum]. Re-testing for HIV during pregnancy and postpartum during observation period assessed in the questionnaire and confirmed through medical records, completed by all participants at baseline and each follow-up until 12 months postpartum 2. Number of new male HIV-positive diagnoses [time frame: baseline, 3 months postpartum, 12 months postpartum]. Number of new HIV-positive test results of male partners during the observation period, coded as Y/N. This is assessed for all male participants at baseline and each follow-up until 12 months postpartum in the questionnaires and confirmed through medical records 3. Number of new discordant couples [time frame: baseline, 3 months postpartum, 12 months postpartum]. Number of new HIV serodiscordant couples identified during the observation period. This is assessed for all couples at baseline and each follow-up until 12 months postpartum in the questionnaires and confirmed through medical records 4. Use of PMTCT interventions (for HIV-positive women only) [time frame: 3 months postpartum, 12 months postpartum, 18 months postpartum]. Composite variable including mothers use of antiretrovirals (ARVs) (Y/N), prophylactic ARVs given to the infant (Y/N), and appropriate infant feeding practices. These are assessed in the questionnaires completed at each follow-up up to 18 months postpartum 5. Utilization of maternal and child health (MCH) services (all couples in the study) [time frame: 3 months postpartum, 12 months postpartum, 18 months postpartum]. Composite variable including having completed at least four antenatal care (ANC) visits during pregnancy (Y/N), childbirth with a skilled attendant (Y/N), and postnatal check-ups for woman (Y/N) and infant (Y/N). These are assessed in the follow-up questionnaires completed up to 18 months postpartum 6. Pre-exposure prophylaxis (PrEP) uptake (discordant partners at follow-ups) [time frame: 3 months postpartum, 12 months postpartum, 18 months postpartum]. Initiation of PrEP by discordant partners assessed at each follow-up in the questionnaires and confirmed through medical records up to 18 months after the baby’s birth 7. Woman’s linkage to HIV care (HIV-positive women) [time frame: baseline, 3 months postpartum, 12 months postpartum, 18 months postpartum]. Time to linkage to HIV care assessed for HIV-positive women at baseline and in the follow-up questionnaires completed up to 18 months postpartum and confirmed through medical records 8. Woman’s enrollment in HIV care (HIV-positive women) [time frame: baseline, 3 months postpartum, 12 months postpartum, 18 months postpartum]. Enrollment of HIV-positive women in HIV care (Y/N) assessed at baseline and in the follow-up questionnaires completed up to 18 months postpartum and confirmed through medical records 9. Woman’s adherence to HIV care (HIV-positive women) [time frame: baseline, 3 months postpartum, 12 months postpartum, 18 months postpartum]. Self-reported adherence to HIV care assessed for HIV-positive women at baseline and in the follow-up questionnaires completed up to 18 months postpartum and confirmed through medical records 10. Number of HIV care visits (HIV-positive women) [time frame: baseline, 3 months postpartum, 12 months postpartum, 18 months postpartum]. Number of HIV care visits assessed for HIV-positive women at baseline and in the follow-up questionnaires completed up to 18 months postpartum and confirmed through medical records 11. Man’s linkage to HIV care (HIV-positive men) [time frame: baseline, 3 months after baby’s birth, 12 months after baby’s birth, 18 months after baby’s birth]. Time to linkage to HIV care assessed for HIV-positive men at baseline and in the follow-up questionnaires completed up to 18 months postpartum and confirmed through medical records 12. Man’s enrollment in HIV care (HIV-positive men) [time frame: baseline, 3 months after baby’s birth, 12 months after baby’s birth, 18 months after baby’s birth]. Enrollment of HIV-positive men in HIV care (Y/N) assessed at baseline and in the follow-up questionnaires completed up to 18 months postpartum and confirmed through medical records 13. Man’s adherence to HIV care (HIV-positive men) [time frame: baseline, 3 months after baby’s birth, 12 months after baby’s birth, 18 months after baby’s birth]. Self-reported adherence to HIV care assessed for HIV-positive men at baseline and in the follow-up questionnaires completed up to 18 months postpartum and confirmed through medical records 14. Number of HIV care visits (HIV-positive men) [time frame: baseline, 3 months after baby’s birth, 12 months after baby’s birth, 18 months after baby’s birth]. Number of HIV care visits assessed for HIV-positive men at baseline and in the follow-up questionnaires completed up to 18 months postpartum and confirmed through medical records 15. Maternal HIV viral suppression (HIV-positive women) [time frame: baseline, 3 months postpartum, 12 months postpartum, 18 months postpartum]. Viral load < 200 copies (undetectable) for all HIV-positive women at baseline and 18 months postpartum through medical records 16. HIV-free child survival [time frame: 18 months after the birth]. Child alive and HIV-free at 18 months after the birth. This is assessed in a brief interview and confirmed through medical records 17. Result of infant HIV test [time frame: 3 months after birth, 12 months after birth, 18 months after birth]. Result of infant HIV test based on medical records 18. Date of infant HIV test [time frame: 3 months after birth, 12 months after birth, 18 months after birth]. Date of infant HIV test based on medical records Data will be shared in agreement with funder (US National Institutes of Health) data sharing policies. Results will be disseminated to local stakeholders (including participants) through local presentations, regional/national/international conferences, and publications. Data collected on android tablets through ODK Collect (Open Data Kit 2019 Creative Commons 4.0 International License) will be aggregated for analysis using the SAS statistical software (Cary, NC, USA), version 9.4. Our primary outcome is the uptake of CHTC. We will use all longitudinal measures of couple testing in a marginal model to compare the proportions among the three study arms. We use this model because our primary interest is to estimate the population average effect of intervention participation on each outcome rather than the effect for a hypothetical average subject or couple. Moreover, within-subject and within-couple correlations among the outcomes are nuisance parameters, not quantities of interest to be modeled explicitly. Our models will include a dummy variable indicating the study group (arm 1 vs arm 3; arm 2 vs arm 3), as well as our stratifying variables and other additional covariates such as couple relationship length, if necessary. A little adjustment for confounding should be necessary due to our randomization. We will employ generalized estimating equations (GEE) with robust standard errors to obtain correct inferences because inference will be valid if the chosen correlation structure is slightly mis-specified [64]. Statistical significance will be for p < 0.017 for the three pairwise comparisons of the three arms to account for multiple comparisons. Between-arm differences for the other outcomes in aim 1 (mean numbers of new HIV+ diagnoses of male partners and new serodiscordant couples) will be modeled with generalized estimating equations (GEE). Each of the secondary outcomes, including proportions of HIV prevention behaviors (PrEP and/or condom use), facility delivery, postnatal healthcare utilization, maternal VL suppression, and HIV-free infant survival up to 18 months, is binary (yes/no). Between-arm comparisons for the probabilities of these outcomes are facilitated with the same GEE models described for the primary outcome. Our assessment for potential mediation and moderation will follow the approach described by Valeri and VanderWeele [65]. We will refer to the treatment effect estimates from these models as estimates A where we will then fit a second model, which takes our original GEE model and incorporates possible mediating variables, such as couple relationship dynamics and social consequences. We will refer to the parameter estimates for these three covariates as estimates B. We will then determine the direct and indirect effects of the treatment on each of the primary and secondary outcomes, with corresponding standard error estimates determined using bootstrap methods [66]. These models can also be adjusted for any potential confounders that are discovered, although we expect the randomization to account for a majority of any potential confounding. Analyses with intact dyads enable the investigation of couple-based research questions of how relationship dynamics affect behavior change in partnerships. For example, we might investigate whether one’s own relationship satisfaction or one’s partner’s relationship satisfaction is more associated with couple testing uptake. To that end, we will extend the analyses described above to include actor and partner effects for covariates and mediators. In order to fit an actor-partner interdependence model (APIM) [67] to our data, we will change our GEE model to a random-effects model so that we can include a random effect for each couple, which will allow us to divide the variation in outcomes into within- and between-couple effects. We will assess the cost-effectiveness of the home-based couple intervention compared to two less resource-intensive strategies of HIV self-testing kits and standard of care. We hypothesize that this intervention will prove to be a cost-effective strategy compared to alternative strategies. However, cost-effectiveness might be sensitive to the intensity of services provided, levels of compensation, the extent of training, levels of adherence to ART treatment, and other attributes. We will develop a decision analysis model using the TreeAge Pro 2020 software (Williamstown, MA, USA). We will calculate the direct costs of each strategy utilizing established guidelines for costing HIV interventions [68, 69] from a program perspective using micro-costing techniques. In addition to costs, we will use data on the changes in HIV status, acquiring opportunistic infections, and mortality which are measured in disability-adjusted life year (DALY, representing a year of healthy life lost due to death or disability) to provide inputs into a Markov model. Markov models are among the most frequently used modeling techniques in clinical decision analysis and health economic evaluation and are particularly helpful when a decision analysis involves the analysis of risk over long periods of time [70, 71]. The proposed state transition models will combine Markov’s health state transitions with the probability that individuals will experience transient events that lead to either a different health state (e.g., HIV transmission) or that can carry significant costs or mortality risk, such as hospitalization for an opportunistic infection. State-transition models have been utilized in many different populations and diseases, including diabetes, cardiovascular disease, HIV, and malaria [72]. To determine the cost-effectiveness of the home-based couple intervention, we will calculate the incremental cost-effectiveness ratios (ICERs) for the intervention vs each of the comparison groups (standard care; HIV self-testing kits). The numerator of the ratio is the difference in costs expressed in US dollars (purchasing parity adjusted); the denominator is the difference in effectiveness measured in DALY. A discount rate of 3% will be applied to both costs and effectiveness. The calculated ICERs will be referenced against the WHO-recommended thresholds to determine whether the home-based couple testing intervention is very cost-effective compared to its comparators for an ICER less than gross domestic product per head, cost-effective for an ICER less than three times gross domestic product per head, or not cost-effective otherwise [73]. We will implement one-way and multiple-way sensitivity analyses to assess the robustness of the cost-effectiveness of the intervention under various uncertainties. This trial is supported by a Data Safety and Monitoring Board (DSMB) independent of the funder, investigators, and competing interests. The overall responsibility of the DSMB is to protect the ethical and safety interests of participants recruited into the Jamii Bora Study while protecting as far as possible the integrity of the study and the scientific validity of the data. The DSMB will review the safety data to identify potential harm from participation in the intervention and other issues designated as necessary for their input. The DSMB will meet at 6-month intervals with the initial meeting taking place just prior to the study recruitment initiation. At all of these meetings, the DSMB will review all accrued data up to that point to assess whether the study aims are being met and to ensure that the benefits of the intervention outweigh any harm. The DSMB charter is available upon request from the trial sponsor. Adverse events may include (1) extreme discomfort from HIV/blood testing, (2) relationship break-ups due to the nature of the intervention, (3) episodes of intimate partner violence, and (4) consequences of learning preliminary positive results from HIV tests. Severe adverse events include adverse pregnancy outcomes, death of a participant, and extended hospitalization. These are specified within the DSMB charter, reported to the IRBs, and discussed at regular DSMB meetings.

• kwena, zachary
• kimbo, liza
• darbes, lynae a.
• hatcher, abigail m.
• helova, anna
• owino, george
• thirumurthy, harsha
• bukusi, elizabeth a.
• braun, thomas
• kilgore, meredith
• pisu, maria
• tamhane, ashutosh
• nghiem, van t.
• agot, kawango
• neilands, torsten b.
• turan, janet m.