Analyzing implementation of public health interventions: A need for rigor, and the challenges of stakeholder involvement.

OBJECTIVES: This article shows how conceptual models can help to develop and evaluate public health interventions. It also reports on the challenges of getting stakeholders involved. METHOD: The analysis is based on the reflexive approach applied by the authors during their participation in two public health intervention research (PHIR) projects, in France and in Burkina Faso. RESULTS: In Paris, PHIR aimed to enable sub-Saharan immigrants to appropriate the existing means of prevention and sexual health care and to strengthen their empowerment in view of preserving their health. Evaluation was carried out using mixed methods. The intervention process theory is based on Ninacs' conceptual model of individual empowerment. The Consolidated Framework For Implementation Research (CFIR) was mobilized a posteriori to analyze the process. PHIR stemmed from collaboration between a research team and two associations. The different stakeholders were involved in the evaluation process, as were, at certain times, persons in highly precarious situations. In Ouagadougou, a community-based dengue vector control intervention was deployed to address an essential but neglected need. As regards evaluation, we opted for a holistic, mixed method approach (effectiveness and process). The contents of the intervention were determined based on tacit knowledge, a community preference survey and solid evidence. The theoretical framework of the intervention consisted in an eco-biological model of vector control. The implementation analysis combined an internal assessment of implementation fidelity with an external CFIR process analysis. All stakeholders were involved in the evaluation process. DISCUSSION: Analysis confirmed not only the value of process evaluations in PHIR, but also the primordial importance of a rigorous approach. Stakeholder involvement is a major challenge to be addressed early in the planning of RISPs; with this in mind, effective and ethically sound assessment mechanisms need to be drawn up. Interdisciplinary evaluative approaches should be preferred, and the use of justified, relevant, and flexible frameworks is highly recommended. CONCLUSION: Lessons learned for those wishing to engage in the process evaluation of a public health intervention are hereby presented.

[Practices and Factors Associated with Recent HIV Testing in the General Population, Côte d'Ivoire: Results of the ANRS Study 12323 DOD-CI]. / Pratiques et facteurs associés au dépistage récent du VIH en population générale, Côte d'Ivoire. Résultats de l'étude ANRS 12323 DOD-CI.

With 2.9% of HIV prevalence in the general population, Côte d'Ivoire is one of the countries most affected by the HIV epidemic in West Africa. In this country, only 63% of people infected with HIV are aware of their status. A cross-sectional phone survey was conducted with a representative sample of 3,867 individuals to describe the practices and factors associated with a recent HIV testing (≤ 1 year) in Côte d'Ivoire. Data relative to the last done HIV test as well as the socio-demographic characteristics, sexual behavior, access to information, perceptions, capacities and autonomy as well as social and geographical environment of the participants were collected. Logistical regression models were used to identify the associated factors with a recent HIV test (≤ 1 year). Lack of information is one of the main barriers to HIV testing (only 60% of individuals know a place to get tested). For men, despite the fact that HIV testing is free of charge, poor economic conditions seem to be a barrier to testing. The social environment, including peer influence, also appears to have an effect on testing among men. For women, testing is associated with their perceptions of HIV exposure. There is a need to rethink the current HIV testing communication in Côte d'Ivoire and to identify economic or social incentives to remove access barriers to HIV testing.

Avec 2,9 % de sa population infectée par le VIH, la Côte d'Ivoire fait partie des pays d'Afrique de l'Ouest les plus touchés par l'épidémie à VIH. On estime que seules 63 % des personnes infectées par le VIH connaissent leur statut. Une enquête transversale, par téléphone, a été réalisée auprès d'un échantillon représentatif de 3 867 personnes afin de décrire les pratiques et les facteurs associés à la réalisation récente (≤ 1 an) d'un dépistage du VIH en Côte d'Ivoire. Les données collectées concernaient le dernier test réalisé ainsi que les caractéristiques sociodémographiques, comportements sexuels, accès à l'information, perceptions, capacités, autonomies ainsi que l'environnement social et géographique des participants. Des modèles de régression logistique ont été réalisés afin d'identifier les facteurs associés à la réalisation récente d'un test du VIH (≤ 1 an). Le manque d'information est l'un des principaux freins au dépistage (seuls 60 % des individus connaissent un lieu où réaliser un test). Chez les hommes, malgré la gratuité du dépistage, une faible condition économique semble être un frein à la réalisation d'un test. L'environnement social, notamment l'influence des pairs, semble aussi avoir un effet sur le recours au dépistage chez les hommes. Chez les femmes, le dépistage est associé à leurs perceptions d'exposition au VIH. L'offre de dépistage actuelle en Côte d'Ivoire nécessite de repenser la communication autour du test ainsi que d'identifier des incitatifs économiques ou sociaux permettant de lever les freins au dépistage.

Effect of organizational models of provider-initiated testing and counseling (PITC) in health facilities on adult HIV testing coverage in sub-Saharan Africa.

The purpose of this paper is to identify which Provider-Initiated HIV Testing and Counseling (PITC) organizational models are the most efficient to maximize testing coverage. We conducted a systematic literature review to identify published articles that evaluated routine PITC programs implemented in adult health facilities in Sub-Saharan Africa. We considered only articles measuring PITC offer, PITC acceptability and PITC coverage. Adjusted meta-regression models were performed to measure the association between PITC offer, acceptability and coverage with PITC organizational model. A total of 30 articles were included in the meta-analysis. Overall, 85.4% [95%CI: 77.2-93.5] of patients were offered a test, and 87.1% [82.4-91.7] accepted the test resulting in a PITC coverage of 74.3% [66-82.6]. Four types of PITC organizational models were identified: PITC initiated and performed during the consultation (model A), PITC initiated before consultation (model B), PITC referred on-site (model C) and PITC referred off-site (model D). Compared to model A, model B had a similar coverage (aOR: 1.02 [0.82-1.26]). However, coverage was lower for model C (aOR: 0.81 [0.68-0.97]) and model D (aOR: 0.58 [0.44-0.77]). Initiating the testing process before or during medical consultation is recommended for maximizing testing coverage among patients.

[Enhanced prenatal HIV couple oriented counselling session and couple communication about HIV (ANRS 12127 Prenahtest Trial)]. / Intervention de conseil VIH prénatal orienté vers le couple et communication conjugale autour du VIH (Essai ANRS 12127 Prenahtest).

BACKGROUND: The Prenahtest study investigated the efficacy of a couple-oriented HIV counselling session (COC) in encouraging couple HIV counselling and testing, and improving intra-couple communication about sexual and reproductive health. We report here on the effect of COC on intra-couple communication about HIV. METHODS: Within this 4-country trial (India, Georgia, Dominican Republic and Cameroon), 484 to 491 pregnant women per site were recruited and individually randomized to receive either the COC intervention, enhanced counselling with role playing, or standard post-test HIV counselling. Women were interviewed at recruitment, before HIV testing (T0), and 2 to 8 weeks after post-test HIV counselling (T1). Four dichotomous variables documented intra-couple communication about HIV at T1: 1) discussion about HIV, 2) discussion about condom use, 3) suggesting HIV testing and 4) suggesting couple HIV counselling to the partner. An intra-couple HIV communication index was created: low degree of communication ("yes" response to zero or one of the four variables), intermediate degree of communication ("yes" to two or three variables) or high degree of communication ("yes" to the four variables). To estimate the impact of COC on the intra-couple HIV communication index, multivariable logistic regressions were conducted. RESULTS: One thousand six hundred and seven women were included in the analysis of whom 54 (3.4%) were HIV-infected (49 in Cameroon). In the four countries, the counselling group was associated with intra-couple HIV communication (P≤0.03): women allocated to the COC group were significantly more likely to report high or intermediate degrees of intra-couple communication about HIV (versus low degree of communication) than women allocated to standard counselling. CONCLUSION: COC improved short-term communication about HIV within couples in different sociocultural contexts, a positive finding for a couple approach to HIV prevention.

[Contraceptive use and incidence of pregnancy among women after HIV testing in Abidjan, Ivory Coast]. / Pratiques contraceptives et incidence des grossesses chez des femmes après un dépistage VIH à Abidjan, Côte d'Ivoire.

BACKGROUND: Within the framework of programs for the prevention of mother-to-child HIV transmission, women who discover their HIV-infection during their pregnancy receive perinatal interventions in order to reduce the risk of HIV transmission to the child. They also receive family planning counselling and free contraceptives in order to avoid a new pregnancy. In this study, we compared contraceptive use and pregnancy incidence between HIV-positive and HIV-negative women who were offered HIV counselling and testing during a program of prevention of mother-to-child HIV transmission. METHODS: In the Ditrame Plus program in Abidjan, 546 HIV-positive and 393 HIV-negative women were HIV-tested prenatally and followed up 2 years after delivery. At each post-partum visit, proportions of contraceptive use were noted, by method. The pregnancy incidence was calculated as the number of pregnancies for 100 women-years at risk. Factors related to the arrival of a new pregnancy were analyzed by Cox model. RESULTS: Between 6 and 24 months post-partum, proportions of women using modern contraception varied from 52 to 65% among HIV-positive women, and from 65 to 75% among HIV-negative women. Pregnancy incidence for 100 women-years at risk was 5.70 (95%CI: 4.17-7.23) and 4.37 (95%CI : 2.83-5.91) (p=0.237) and unwanted pregnancy incidence was 1.07 (95%CI: 0.41-1.73) and 2.39 (95%CI: 1.25-3.53) (p=0.023), respectively among HIV-positive and HIV-negative women. The end of post-partum abstinence, the death of the index child and the end of breastfeeding were positively linked to the arrival of a new pregnancy in the post-partum period. CONCLUSION: Among these women prenatally HIV-tested, family planning counselling and regular follow-up was accompanied by a high rate of contraceptive use after delivery, and consecutively to a low pregnancy incidence irrespective of serostatus. In particular, HIV-positive women had fewer unwanted pregnancies than HIV-negative women. Integration of adequate family planning services in the post-partum follow-up in prevention programs plays an important role in reducing the risk of mother-to-child transmission, by reducing pregnancies among HIV-positive women.

Sexual prevention of HIV within the couple after prenatal HIV-testing in West Africa.

The resumption of sexual activity after delivery is a key moment in the management of the risk of sexual HIV transmission within the couple for women who have been prenatally tested for HIV. In this study, we have investigated consistent condom use during the resumption of sexual activity and its evolution over time among women tested for HIV infection during pregnancy. We tested for HIV during pregnancy 546 HIV-infected and 393 HIV-negative women within the Ditrame Plus ANRS project in Abidjan; these women were followed-up for two years after delivery. Most HIV-negative women (96.7%) disclosed their HIV-test result to their partners, whereas only 45.6% of HIV-infected women did so (p<0.001). Partners of HIV-infected women were more likely to be tested for HIV before resuming sexual activity than partners of HIV-negative women (11.7% vs. 7.4%, p=0.054). Less than one third of both HIV-infected and HIV-negative women reported having systematically used condoms during the resumption of sexual activity. The proportions of HIV-infected and HIV-negative women having consistently used condoms were respectively 26.2% and 19.8% (p=0.193) at 3 months post-partum, 12.1% and 15.9% (p=0.139) at 12 months post-partum, and 8.4% and 10.6% (p=0.302) at 18 months post-partum. In our study, although women had been prenatally tested for HIV and properly counselled on the sexual risk of HIV transmission, male partners were not tested for HIV before the resumption of sexual activity after delivery, very few couples were using condoms systematically and condom use was decreasing over time.

Prophylactic use of cotrimoxazole against opportunistic infections in HIV-positive patients: knowledge and practices of health care providers in Côte d'Ivoire.

We present here the results of a survey conducted in Côte d'Ivoire, Africa, among healthcare providers, on the knowledge of prophylactic use of cotrimoxazole to prevent opportunistic infections in HIV-infected persons. The survey was conducted in 15 health centres, involved or not in the 'initiative of access to treatment for HIV infected people'. Between December 1999 and March 2000, 145 physicians and 297 other health care providers were interviewed. In the analysis, the health centres were divided into three groups: health centres implicated in the initiative of access to treatment for HIV-infected people with a great deal of caring for HIV-infected people, health centres implicated in this initiative but caring for few HIV-infected people, and health centres not specifically involved in the care of HIV-infected people. Six per cent of physicians and 50% of other health care providers had never heard of cotrimoxazole prophylaxis. The level of information about this prophylaxis is related to the level of HIV-related activities in the health centre. Among health care providers informed, knowledge on the exact terms of prescription of the cotrimoxazole is poor. In conclusion, it appears that the recommendations for primary cotrimoxazole prophylaxis of HIV-infected people did not reach the whole health care provider population. Most physicians are informed but not other health workers, even if the latter are often the only contact of the patient with the healthcentre. The only medical staff correctly informed are the physicians already strongly engaged in the care of HIV-infected people.

[Impaired fertility in women infected with HIV-1. Implications for sentinel serosurveillance]. / Réduction de la fertilité chez les femmes infectées par le VIH-1. Implications pour la sérosurveillance sentinelle.

BACKGROUND: Given the relationship between HIV infection and fertility, antenatal clinic-based HIV prevalence may not provide a good estimate of the community HIV prevalence. The objective of this work was to evaluate the impact of HIV infection on fertility among women attending antenatal clinics in Bobo-Dioulasso (Burkina Faso), and to discuss possible implications on HIV sentinel surveillance. METHODS: In the context of a phase II/III clinical trial of a short course of Zidovudine during pregnancy (DITRAME - ANRS 049 trial) we consecutively proposed voluntary counselling and HIV testing (VCT) to 1349 women aged at least 18 years, carrying a pregnancy of 7 months or less and living in Bobo-Dioulasso. During pre-test counselling session, a standardised questionnaire was administered to collect detailed information regarding socio-demographic characteristics and obstetrical history. Blood samples were then taken and tested for HIV after written informed consent. RESULTS: Mean age (+/- standard deviation) at first sexual intercourse was similar among HIV-infected (HIV+) (16.7+/- 2; n=83) and HIV-negative (HIV-) women (16.9+/- 2; n=1336). However, HIV+ women aged 25 years and above had, on the average fewer pregnancies (3.8+/- 1.5; n=37) than HIV- women (5.0+/- 2.3; n=567), p<0.01. Similarly, these HIV+ women had, on average, less live births (2.8+/- 1.3; n=35) than HIV- ones (3.7+/- 2.1; n=555), p=0.02. Other sexual and obstetrical characteristics such as maternal age, proportion of primigravidae, stillbirths or spontaneous abortions were comparable between HIV+ and HIV- women. CONCLUSIONS: Our data suggest that the level of fertility of HIV+ women aged 25 years and above is significantly lower than for HIV- women. Therefore, HIV+ women in this age group are likely to be under-represented among antenatal clinic attendees. These findings suggest adjusting antenatal clinic-based HIV sentinel surveillance data for age and fertility in order to derive a good estimate of the community HIV prevalence.

Impaired fertility in HIV-1-infected pregnant women: a clinic-based survey in Abidjan, Côte d'Ivoire, 1997.

OBJECTIVE: Differences in fertility among HIV-1-positive and HIV-negative women tested in prenatal clinics were suspected by routine data collection in Abidjan, Côte d'Ivoire. This study was conducted on detailed fertility patterns among women at the same antenatal care centres, in order to assess these differences. METHOD: The survey was carried out on 1201 consecutive women who agreed to be tested for HIV. Data collected included a detailed account of pregnancies, the time interval between the last two pregnancies, and the health status at the time of the survey. Blood samples were tested for HIV and syphilis with informed consent. RESULTS: Despite an earlier exposure to pregnancy risk, HIV-1-infected women aged 25 years and above, had, on the average, fewer pregnancies than uninfected women. An analysis of the interval between the last two pregnancies among multigravidae showed that, all things being equal, being HIV-1 positive decreased the risk of being pregnant by 17% (Cox regression, hazard ratio = 0.83, 95% confidence interval (CI): 0.69-0.99). This shift in the occurrence of the last pregnancy was more profound among HIV-1-positive women already at the symptomatic or AIDS stage, than among asymptomatic women. CONCLUSION: These data confirm that women infected by HIV-1 would become pregnant less often than uninfected women, for an equal exposure to the risk of pregnancy. Therefore HIV-1-positive women could be under-represented at antenatal centres. Programmes involving such settings both for epidemiological surveillance and the reduction of mother-to-child transmission should take this observation into account.

Impact of chloroquine resistance on malaria mortality.

Over 12 years, from 1984 to 1995, we conducted a prospective study of overall and malaria specific mortality among three rural populations in the Sahel, savanna and forest areas of Senegal. The emergence of chloroquine resistance has been associated with a dramatic increase in malaria mortality in each of the studied populations. After the emergence of chloroquine resistance, the risk of malaria death among children 0-9 years old in the three populations was multiplied by 2.1, 2.5 and 5.5, respectively. This is the first study to document malaria mortality at the community level in Africa before and after the emergence of chloroquine resistance. Findings suggest that the spread of chloroquine resistance has had a dramatic impact on the level of malaria mortality in most epidemiological contexts in tropical Africa.

HIV-1 infection and reproductive history: a retrospective study among pregnant women, Abidjan, Côte d'Ivoire, 1995-1996. Ditrame Project.

The aim of this paper is to determine the differences of fertility between HIV-1 infected and uninfected women in Abidjan, Côte d'Ivoire, using data available in an observational study conducted in 1995 and 1996 in 2 antenatal care centres in the district of Yopougon, Abidjan, within an intervention programme to reduce mother-to-child HIV-1 transmission (DITRAME project, ANRS 049). Fertility indicators have been constructed from retrospective data on pregnancies and births, and univariate and multivariate analyses have been performed on these indicators and stratified by age groups to compare HIV-1 positive and HIV-negative populations. The main outcome measures were the number of pregnancies, the number of miscarriages, the risk of miscarriage and the proportion of primigravida. Four thousand, three hundred and ninety-six women agreed to HIV testing: 12.1% were found to be HIV-1 infected. HIV-1 positive women had significantly fewer pregnancies than HIV-negatives in age-groups 25-29 (P = 0.05) and 30-34 (P = 0.008). The risk of having had at least one abortion or stillbirth was significantly higher for HIV-1 infected women than for HIV-negatives (OR = 1.28, 95% CI: 1.02-1.60), when controlling for social and demographic factors. This study suggests that HIV-1 infection has deleterious consequences on female fertility, with lower fertility rates and more frequent adverse pregnancy outcomes. Family planning and antenatal care services should consider antenatal HIV counselling and testing in women in areas of high HIV prevalence.

Role of immunizations in the recent decline in childhood mortality and the changes in the female/male mortality ratio in rural Senegal.

In early 1987, immunizations were introduced to an isolated area of eastern Senegal where there had previously been no regular immunizations. Since immunizations were the only change introduced in the area during this period, this allowed the authors to study the impact of immunizations of mortality in different age groups and the possible interaction with sex differences in mortality. They compared mortality rates for the 6 years before and the 6 years after the introduction of immunization. Neonatal mortality declined 31% (95% confidence interval (CI) 17 to 43); between 1 and 8 months of age, the reduction was 20% (95% CI -2 to 37); and between 9 and 59 months of age, mortality declined 48% (95% CI 39 to 56). Excluding acute measles deaths, the reduction was 16% (95% CI -8 to 35) between 1 and 8 months of age and 32% (95% CI 20 to 43) between 9 and 59 months of age. The decline was stronger in villages that maintained high coverage after the initial national campaign, whereas mortality increased again in the villages where the coverage declined. Since the reduction in mortality was most marked after 9 months of age, measles immunization is likely to have been the most important vaccination. Both female and male mortality declined but not equally quickly. The reduction in mortality in the neonatal period was significantly greater in males than in females, resulting in an increase in the female/male mortality ratio from 0.64 (95% CI 0.50 to 0.83) to 0.96 (95% CI 0.71 to 1.30), p = 0.04. After 9 months of age, the reduction in mortality was somewhat greater in females than in males, resulting in a decrease in the female/male mortality ratio from 1.04 (95% CI 0.85 to 1.28) to 0.79 (95% CI 0.62 to 1.02), p = 0.10.

Lower mortality for female-female twins than male-male and male-female twins in rural Senegal.

Twins have been registered prospectively for 12-22 years in 42 small villages in the Bandafassi area of Eastern Senegal. We studied 98 pairs of twins to test whether twins in opposite-sex pairs have higher postneonatal mortality than same-sex twins. Neonatal mortality for twins was 41.3%; mortality for infants and for children under age 5 years was 53.0% and 66.8%, respectively. Neonatal mortality was identical for same-sex and opposite-sex twin pairs, but much higher for boys than girls [relative risk = 1.8; 95% confidence interval (CI) = 1.2-2.6]. There was clustering of double neonatal deaths for all types of twins. In the postneonatal period, female-female twins had lower mortality than other twin types. Twins had higher post-neonatal mortality as long as the co-twin was alive [mortality rate ratio (MR) = 2.6; 95% CI = 1.0-6.7]. Girls had excess mortality when the co-twin was of the opposite sex (MR = 4.3; 95% CI = 1.2-15.3), whereas there was no difference for boys. In conclusion, contact with a co-twin of the opposite sex increased child mortality for female twins. Our data are not sufficient to determine whether this difference is specific for girls or applies to boys as well.

Barriers to universal child immunization in rural Senegal 5 years after the accelerated Expanded Programme on Immunization.

Although the Expanded Programme on Immunization (EPI) has been a worldwide success, weak points remain, particularly in Africa. In Senegal, for example, immunization coverage was low in 1990 (60%), in part because of poor results in rural areas. In order to identify obstacles to EPI in such areas, we carried out an immunization survey in Bandafassi, a rural area of Senegal, where 6078 inhabitants lived in 23 small villages. Only 41% of children aged 1-10 years were completely vaccinated in February 1992, with considerable variations in coverage from one village to another, according to their geographical location: 71% of children were completely vaccinated in villages less than 10 km from the health centre, whereas in remote villages only 10% of children had been completely vaccinated. There was no variation according to ethnic group. From 1987 to 1992, the gap in immunization coverage between the remote villages and those located close to the health centre has steadily increased. There is a need to improve the performance of the mobile teams in the remote villages and to increase awareness about the importance of immunization.