AIDS and Behavior

, Volume 13, Issue 1, pp 23–32

Correlates of Adherence to Antiretroviral Therapy in HIV-Infected Children in Lomé, Togo, West Africa

Authors

  • Julie Polisset
    • Institut de Santé Publique, d’Epidémiologie et de Développement (ISPED)Université Victor Segalen Bordeaux 2
  • Francine Ametonou
    • Institut de Santé Publique, d’Epidémiologie et de Développement (ISPED)Université Victor Segalen Bordeaux 2
  • Elise Arrive
    • Institut de Santé Publique, d’Epidémiologie et de Développement (ISPED)Université Victor Segalen Bordeaux 2
  • Anthony Aho
    • Association, Action contre le SIDA
    • Institut de Santé Publique, d’Epidémiologie et de Développement (ISPED)Université Victor Segalen Bordeaux 2
Original Paper

DOI: 10.1007/s10461-008-9437-6

Cite this article as:
Polisset, J., Ametonou, F., Arrive, E. et al. AIDS Behav (2009) 13: 23. doi:10.1007/s10461-008-9437-6

Abstract

We assessed pediatric adherence to antiretroviral therapy (ART) and examined associated factors among children in Togo, West Africa. Structured interviews of caregivers of consecutively enrolled HIV-infected children receiving ART in three HIV/AIDS care centers in Lome, Togo were conducted. Child perfect adherence reflected caregivers’ report of no antiretroviral drug doses missed neither in the past 4 days nor in the month before the interview. A total of 74 ART-treated children were included (median age 6 years). Of these, 42% of caregivers declared perfect adherence. In univariate analyses, the major factors relating to child non-adherence were: being female, living in an individual setting (vs. compound with enlarged family), receiving other ART than an NNRT-based regimen, drug regimens with six pills/spoons or more per day, caregiver other than biological parent, caregiver not declaring HIV-status, not participating to support groups and having perceived difficulty of antiretroviral (ARV) administration. In multivariate analysis, female gender, living in an individual setting, receiving other than NNRTI-based regimen and caregivers’ perceived difficulty of ARV administration remained independently associated with the reported child’s non-adherence. These data show low rates of perfect adherence to ART in children in West Africa, influenced by child and caregiver characteristics and suggest a need for counseling and education interventions as well as continuous psychological and social support.

Keywords

AdherenceAntiretroviral therapyCorrelatesChildren Africa

Introduction

Major steps have been taken in recent years in the fight against the HIV/AIDS pandemic, especially in the expansion of treatment in developing countries. Nonetheless, recent data from the World Health Organization showed that of the 2.3 million children aged 0–14 years living with HIV in 2006, about 780,000 were estimated to be in need of antiretroviral therapy (ART). More than 90% of these children lived in resource-limited countries, especially in sub-Sahara Africa (WHO 2007). In the absence of treatment, most infants and children younger than 5 years with perinatally acquired HIV infection experience rapid progression to severe symptomatic disease and death (Marston et al. 2005; Newell et al. 2004; UNAIDS/WHO 2006). Access to ART for HIV-infected children in sub-Saharan Africa has been limited so far. Currently, children represent only 6% of the overall population receiving ART when this age group represents 14% of the total population in need of ART (Mills et al. 2006a; WHO 2007). Lack of affordable and simple HIV-diagnostic testing technology for children less than 18 months of age, limited availability of pediatric drug formulations as well as an unmet need of trained practitioners in the appropriate use of ART are among the obstacles to scaling up ART for children in developing countries (Boni et al. 2000; Calmy et al. 2004; Ginsburg et al. 2006; Kline 2006).

ART, defined as the combination of three potent antiretroviral drugs, improves health and quality of life of HIV-infected children (Fassinou et al. 2004; George et al. 2007). However, it is a serious challenge for care and patient management as it requires multiple medications and frequent drug dosing (Gill et al. 2005). Adherence failure may lead to selective resistant viral strains and ultimately to treatment failure (Mills et al. 2006a). Few studies have been carried out in sub-Saharan Africa regarding adherence to ART among children (Arrivé et  al. 2005; Bikaako-Kajura et al. 2006). In Togo, West Africa, 110 000 people were living with HIV/AIDS in 2005; among them 9,700 were children between 0 and 14 years-old (UNAIDS/WHO 2006). In this country, like others where the HIV/AIDS pandemic is a priority, non-government organizations are among the main actors in the fight against HIV/AIDS through prevention and care initiatives. The objective of this survey was to assess correlates of pediatric ART adherence among several NGOs operating within limited-resource settings in West Africa.

Methods

Study Design and Population

We conducted a cross-sectional survey to evaluate treatment adherence and investigate its barriers among children receiving ART in the three main HIV/AID treatment centres in Lome: “Action Contre le SIDA” (ACS), “Espoir Vie Togo” (EVT) and “Aides Médicales et Charité” (AMC), all non-governmental organizations located in peri-urban areas of the city, during a 4-week period in May 2006. To be enrolled in the study, HIV-infected children met the following selection criteria: 15 years of age or less, receiving ART for more than 1 month, being followed-up at one of the three HIV/AIDS care centers and being accompanied by a caregiver (biological parent, adoptive parent, foster parent or other relative who looks after a child including being responsible for administration of the antiretroviral regimen). The clinical monitoring database of each of the centers was used to screen for children eligible for this study.

Data Collection

Caregivers of enrolled HIV-infected children were consecutively interviewed (face-to- face for an average of 30 min) by two of the authors (JP, FA) with the help from a translator when necessary. Recruitment was either health-centre based when the child arrived for consultation or home-based as part of routine home visits by health staff. Caregivers were administrated a pre-designed, structured and anonymous questionnaire and checklist. Biological correlates of treatment success (viral load and CD4 count) were not available in this setting.

Variables of Interest

Socio-demographic characteristics of the child and the caregiver were collected as well as child’s antiretroviral treatment characteristics (drug class and regimen) and psycho-social variables such as: caregiver’s relationship with the child through a ten-item checklist, basic knowledge of HIV/AIDS issues, caregiver’s perception of the disease, opinions regarding disclosure and medication-related issues, opinion of possible difficulties of antiretroviral (ARV) administration and potential useful interventions to improve adherence. In addition, among children 6 years of age and above, knowledge of his/her HIV status was also collected. Adherence was scored as perfect when caregivers reported zero missed doses over the last 4 days and over the last month, before the interview, as proposed in the literature (Van Dyke et al. 2002).

Data Analysis

The statistical association between the variables of interest and reported adherence was assessed by χ2 or Fisher exact test for qualitative data or Kruskal-Wallis test for quantitative data. Variables with P-value ≤.10 in univariable analysis were considered for the multivariable logistic regression analysis to study the relative effect of each characteristic expressed as odds ratio (OR) and selected using a backward procedure (Steyerberg et al. 2000). Associations between the multi-variable covariate set were evaluated for potential multi-coliniarity and, in cases where the association between a pair of covariates was high, only the variable of primary interest was included in the multivariate covariate set.

Results

Characteristics of Study Participants

A total of 83 children were receiving ART at the time of the study in the three HIV/AIDS care centers. Nine children were not enrolled as no formal caregiver was identified; the remaining 74 (89%) children were included in the survey. Thirty-seven children (50%) were from the ACS site, 28 from EVT (38%) and 9 from AMC (12%) The median age of these children at the time of the survey was 6 years (interquartile range [IQR] = 4–9), 86% being less than 10 years and 74% attending school (Table 1). Seventy-seven percent of the children were orphaned by a father, a mother or both. The mother was the caregiver for the child for 52% of the cases, while other caregivers included an uncle or aunt 23%, grandmother 15%, farther 7%, mother in law 3% and sister 1%. The median age of the caregivers was 37 years old (IQR = 32–45). The majority were woman (88%), married (40%), with a high education level (secondary school or above) (42%) and had, on average, three children to take care of at home. Forty-two caregivers interviewed (57%) declared being HIV-seropositive. Of these, 22 where receiving ART in the same centre than the child. All caregivers reporting they were HIV-seropositive were the mothers or the fathers of the child. Among all caregivers, 47% reported to have had disclosed the child’s HIV-infection to others. However, 46% were not aware of basic HIV/AIDS issues, 16% perceived difficulty administrating ARVs to the child and 85% reported not having disclosed the child his/her serological status. The percentage of caregivers not being aware of basic HIV/AIDS issues and the percentage of caregivers perceiving difficulties in administrating ARVs did not differ whether they had declared they were HIV-positive or not: 43% vs. 50%, χ2 (1, N = 74) = 0.54, P > .05; 14% vs. 19%, χ2 (1, N = 74) = 0.61, P > .05, respectively. Most of the children were living in community settings (65%), particularly if the caregiver was the parent (77% vs. 48%), χ2 (1, N = 74) = 6.35, P < .05. Boys were more likely than girls to have their parent as a caregiver (67% vs. 42%), χ2 (1, N = 74) = 4.77, P < .05.
Table 1

Characteristics of the HIV-infected children and their caregivers included in the study of correlates of adherencea to antiretroviral therapy in Lomé, Togo, May 2006

Variables

Adherent (n = 31)

Non adherent (n = 43)

Total (n = 74)

n (%)

n (%)

n (%)

Gender

    Female

9 (29.0)

23 (53.5)

32 (43.2)

Age

    0–5

14 (45.2)

22 (51.2)

36 (48.6)

    6–10

11 (35.5)

17 (39.5)

28 (37.8)

    11–15

6 (19.3)

4 (9.3)

10 (13.6)

Schooling

    Yes

25 (80.6)

30 (69.8)

55 (74.3)

Orphanb

    Yes

25 (80.6)

32 (74.4)

57 (77.0)

Living conditions

    Community settingc

25 (80.6)

23 (53.5)

48 (64.9)

    Individual house

6 (19.4)

20 (46.5)

26 (35.1)

Child’s knowledge of his/her HIV status

    Yesd

8 (25.8)

3 (7.0)

11 (14.9)

Child’s antiretroviral treatment

    NNRTI-based

25 (80.6)

21 (48.8)

46 (62.2)

    NRTI-based

4 (13.0)

20 (46.5)

24 (32.4)

    PI-based

2 (6.4)

2 (4.7)

4 (5.4)

Free antiretroviral treatment

    Yes

25 (80.6)

38 (88.4)

63 (85.1)

Six pills/syrup spoons or more per day

    Yes

11 (35.5)

26 (60.5)

37 (50.0)

Reporting adverse ART events

    Yes

4 (12.9)

5 (11.6)

9 (12.2)

Child received home-based care

    Yes

20 (64.5)

23 (53.5)

43 (58.1)

Caregiver

    Father or mother

23 (74.2)

20 (46.5)

43 (58.1)

    Othere

8 (25.8)

23 (53.5)

31 (41.9)

Caregiver’s high education level (secondary school or above)

    Yes

14 (45.2)

17 (39.5)

31 (41.9)

Caregiver living with partner

    Yes

9 (29.0)

22 (51.2)

31 (41.9)

Caregiver’s reported HIV infection

    Yes

22 (71.0)

20 (46.5)

42 (56.8)

Caregiver’s participation to group support discussions

    Yes

19 (61.3)

14 (32.6)

33 (44.6)

Caregiver’s reported knowledge of HIV

    Yes

20 (64.5)

20 (46.5)

40 (54.1)

Caregiver’s disclosure of the child’s HIV-infection to others

    Yes

18 (58.1)

17 (39.5)

35 (47.3)

Caregiver’s perceived difficulty of ARV treatment

    No

30 (96.8)

32 (74.4)

62 (83.8)

NNRTI: non-nucleosidic reverse transcriptase inhibitors; NRTI: nucleosidic reverse transcriptase inhibitors; PI: protease inhibitors; ARV: antiretroviral; ART: antiretroviral therapy

aEstimated by the caregiver’s declaration, a child is considered adherent when all doses were administered the past 4 days and the month before the interview

bOrphaned of father, mother or both

cliving in compound with enlarged family

dAll were ≥6 year-old

eOthers included, foster parents and other family members (aunt, grandmother, uncle)

Forty-six children (62%) were receiving a non-nucleoside reverse transcriptase inhibitors-based regimen, 24 (32%) a nucleoside reverse transcriptase inhibitors-only regimen and the remaining were receiving a protease inhibitors-based one. The median duration of treatment was 10.5 months (IQR = 6.7–22.7). Eleven caregivers (15%) declared paying for the child’s antiretroviral treatment. For half of the enrolled children, this treatment was based on six pills/syrup spoons of drugs per day or more, according to the caregivers. A high number of pills/syrup spoons per day was associated with the type of drugs taken: 19% of the children receiving NNRTI-based regimens versus 100% of the children receiving other regimens had six or more pills/syrup spoons per day, χ2 (1, N = 74) = 45.04, P < .0001 The main reported adverse drug effects were vomiting and abdominal pain in 12% of cases and the main treatment constraint as reported by the caregiver was the need to take the medication with meals. The percentage of caregivers perceiving difficulties of ARV administration was 9% when a NNRTI-based regimen was given and 29% when another regimen was given, (N = 74), P < .05. Answers by caregivers on which activities they would like to receive to help them manage child’s regimen was mainly related to financial constraints and how to announce the serological status to the child.

Adherence Rates and Associated Factors

Perfect adherence was reported for 42% of the children (95% confidence intervals [CI] = 30.5–53.9). The reasons enumerated by the caregivers to explain why the child did not take the medication were: out-of-stock of drugs due to the supplier (43%), forgetting (22%), vomiting (14%) and refusal of the child (11%). Eight caregivers (11%) reported perfect adherence on the 30-day measure, but also reported that the child missed one or more doses over the last 4 days. As shown in Tables 1 and 2, imperfect adherence reported by caregivers was more common when the child was a female, as well as those receiving six pills/syrup spoons or more per day, those receiving other ART than the NNRTI-based regimen, those living in an individual housing, those whose caregiver was other than the father or the mother, and for children aged 6 years or more, those not knowing their HIV status (N = 38, P < .05, Table 1). Children whose caregiver had declared he/she was HIV-positive were more likely to be adherent, as well as those whose caregiver had reported participating in support group discussions and not perceiving difficulty in ARV administration.
Table 2

Correlates of non-adherence to antiretroviral therapy in HIV-infected children in Lomé, Togo, May 2006 (n = 74)a

Variables

Univariable models

Multivariable models

 

Initial

Finalc

OR

[CI]

DF

OR

[CI]

DF

OR

[CI]

DF

Gender

  

1

  

6

  

4

    Male

1

  

1

  

1

  

    Female

2.8

[1.0–7.5]*

 

3.3

[0.9–11.4]

 

3.8

[1.1–12.7]*

 

Ageb

  

2

      

    0–5

1

        

    6–10

1.0

[0.3–2.8]

       

    11–15

0.4

[0.1–1.8]

       

Schooling

  

1

      

    No

1

        

    Yes

0.5

[0.1–1.7]

       

Orphanh

  

1

      

    No

1

        

    Yes

0.7

[0.2–2.2]

       

Living

  

1

      

    Community settingd

1

  

1

  

1

  

    Individual house

3.6

[1.2–2.2]*

 

4.3

[1.0–17.0]*

 

6.0

[1.6–21.8]**

 

Child’s antiretroviral treatment

  

1

      

    NNRTI-based

   

1

  

1

  

    Othere

 

[1.4–12.8]**

 

3.9

[1.1–14.1]*

 

3.9

[1.1–13.4]*

 

Free antiretroviral treatment

  

1

      

    No

         

    Yes

1.8

[0.2–6.7]

       

Six pills/syrup spoons or more per day

  

1

      

    No

1

        

    Yes

2.8

[1.0–7.3]*

       

Reporting adverse ART eventsb

  

1

      

    No

1

        

    Yes

0.9

[0.2–3.7]

       

Child received home-based care

  

1

      

    No

1

        

    Yes

0.6

[0.2–1.7]

       

Caregiver

  

1

      

    Father or mother

1

  

1

     

    Otherf

3.3

[1.2–9.1]*

 

2.0

[0.5–7.3]

    

Caregiver’s high education level (secondary school or above)

  

1

      

    No

1

        

    Yes

0.8

[0.3–2.1]

       

Caregiver living with partner

  

1

      

    No

1

  

1

     

    Yes

2.6

[0.9–6.9]g

 

1.3

[0.3–4.7]

    

Caregiver’s reported HIV infection

         

    No

1

        

    Yes

0.4

[0.1–1.0]*

       

Caregiver’s participation to group support discussions

  

1

      

    No

1

        

    Yes

0.3

[0.1–0.8]*

       

Caregiver’s reported knowledge of HIV infection

  

1

      

    No

1

        

    Yes

0.5

[0.1–1.3]

       

Caregiver’s reported disclosure of child’s HIV sero-status to others

  

1

      

    No

1

        

    Yes

0.5

[0.1–1.3]

       

Caregiver’s perceived difficulty of ARV administration

  

1

      

    Yes

1

  

1

  

1

  

    No

0.1

[0.01–0.8]*

 

0.1

[0.009–1.1]

 

0.09

[0.009–0.1]*

 

CI = 95% Confidence interval; aOR = adjusted Odd Ratio; DF: Degree of freedom; NNTI = non-nucleoside reverse transcriptase inhibitors, ART = antiretroviral therapy; ARV: antiretroviral

aEstimated by the caregiver’s declaration, a child is considered adherent when all doses were administered the past 4 days and the month before the interview

bIn univariate analysis, one cell has expected value less than 5

cIn multivariate analyses, 25% of the cells have expected values less than two but more than zero

dLiving in compound with enlarged family

eProtease or nucleoside reverse transcriptase inhibitor-based regimens

fOther included foster parents and other family members (aunt, grandmother, uncle)

gVariable with .1 ≥ P > .05 in univariable analysis, included in the multivariable analysis

hOrphaned of father, mother or both

P < .05, ** P < .01

We found no association of perfect adherence with several of child’s characteristics: age, attending school, being an orphan (of mother, father or both), having received free ART, reporting adverse ART events and having received home visits by a health worker. Regarding the caregiver, the following factors were also not associated with the child’s perfect adherence: educational level, living with a partner, disclosure of child’s HIV sero-positive status, and basic knowledge of HIV. To reduce multi-coliniarity in the covariate set, several covariates that were significant at the univariate level were not subsequently included in the multivariate analyses. The HIV status of the caregiver was associated with a high number of other variables (Table 3); this variable was not included in the multivariate analysis. Neither was the number of pills for the same reasons, this variable being associated with the type of regimen, and judged to be a consequence of the type of treatment. Similarly, the caregiver’s participation to group support discussions was significantly more common if the caregiver was the mother or the father of the child (65%) than somebody else (16%), so this variable was also not included. In multivariate analysis, female gender (adjusted odds ratio [aOR]: 3.8; CI: 1.3–12.6), living in an individual setting (aOR: 6.0; CI: 1.6–21.8), receiving other than NNRTI-based regimen (aOR: 3.9; CI: 1.1–13.4) and caregivers’ perceived difficulty of ARV administration (aOR: 10.6; CI: 1.0–111.1) remained independently associated with the child’s reported non-adherence (Table 2). This model accurately classified 66% of the observations.
Table 3

Association table for the covariables considered for the multivariable analysis on determinants of non-adherence to antiretroviral therapy in HIV-infected children in Lomé, Togo, May 2006, χ2 (1, n = 74)

Variables

1

2

3

4

5

6

7

8

9

1. Gender

0.37

0.84

0.88

4.77

0.57

8.76**

1.33

3.89*

2. Living conditions

 

0.34

0

6.36*

2.35

1.61

0.02

5.47*

3. Child’s antiretroviral treatment

  

45.04***

0.01

1.22

0.05

5.06*

0.19

4. Six pills/syrup spoons or more per day

   

0.50

0.05

0.05

1.59

0.22

5. Caregiver

    

5.73*

17.50***

0.39

70.02***

6. Caregiver living with partner

     

3.29

3.61

4.77*

7. Caregiver’s participation to group support discussion

      

0.17

15.24***

8. Caregiver’s perceived difficulty of ARV administration

       

0.27

9. Caregiver’s reported HIV infection

        

P < .05, ** P < .01; *** P < .0001; ARV: antiretroviral

Discussion

ART treatment has been shown to dramatically modify the course of HIV infection in children reducing mortality by fivefold or more and resulting in higher survival rates (George et al. 2007; Gibb et al. 2003). However, emergence of resistance to ARVs when drugs are not sustained makes adherence an essential component of successful life-long therapy. Till date, a limited number of studies have been carried out among HIV-infected children concerning adherence to ART (Arrivé et al. 2005; Katko et al. 2001; Reddington et al. 2000; Williams et al. 2006). To our knowledge, our study is among the few to identify non-adherence factors in children in resource-poor settings (Arrivé et al. 2005; Bikaako-Kajura et al. 2006). It has identified patient/caregiver psychosocial variables, treatment regimens and care factors as significant correlated factors influencing child adherence to ART. Perfect child’s adherence to ART in this study was 42%; female gender, living in an individual setting; receiving other than NNRTI-based regimen and perceived difficulty of ARV administration by the caregiver were identified as correlates to the child’s imperfect adherence. These results are in agreement with studies undertaken in developed countries (Mellins et al. 2004; Mills et al. 2006b).

It is difficult to estimate with precision ART adherence in children since the measurement methods are not standardized. Indeed, it is often recommended to associate the patient’s and or caregiver’s declaration with virological outcomes (Fairley et al. 2005; Williams et al. 2006). Using this method, non-adherence has been reported to range between 70% and 84% in recent studies (Giacomet et al. 2003; Williams et al. 2006). Comparatively, the adherence rate obtained here (42%), based on self-report measures only, is relatively low considering that this measure tends to overestimate adherence (Van Dyke et al. 2002). Collecting information on adherence from multiple sources of information has been recommended as self-reporting of medication taken during the previous 4 days may not capture periods of non-adherence that are intermittent or present over more extended periods or because of inconsistencies in two time-intervals due to measurement issues as in our study (Dolezal et al. 2003). Even though we have used two time-periods to assess adherence the restriction of self-reporting adherence measures due to financial constraints is a limitation of this study as reported also in previous reports (Wamalwa et al. 2007).

Based on univariate analysis results, we, as others in India (Kumarasamy et al. 2005), have observed that children were less adherent to ART when they are taken care of by foster parents or second-degree relatives. This is in contrast with results from other reports in the United States and Europe (Giacomet et al. 2003; Williams et al. 2006) that have shown that adherence was lower in children when medication was administered by biological parents. The dynamics of these relations and their impact on ART adherence appear to vary across cultures. Based on our results, it is necessary to consider targeting information and education interventions on the importance of adherence toward foster caregivers. Additionally, the social environment where HIV-positive children live has been reported to have a relation with adherence to ART (Steele et al. 2007). Adherence may be lower among children who live in a common court than those living without other tenants. The main reason suggested is that it would be easier for the caregivers to administer treatment to the child with absolute discretion. Our study showed the opposite which could be linked to the possibility that the existing community network in our setting favors social support (Kumarasamy et al. 2005; Nachega et al. 2006).

In this sense, we, can perhaps relate but not affirm that living in a family compound which helps create social networks can be influential to adherence (Altice et al. 2001; Reddington et al. 2000). Social community relationship built on support, trust and open communication can have positive results regarding adherence to ART in children.

Discrimination and stigma are among the concerns of caregivers that may contribute to non-disclosure of the child’s HIV status with family members or others (Abaia-Barrero and Castro 2006). Among the most important support activity needed and identified by the caregivers in this study, was how to announce the serological status to the child. Disclosure of HIV/AIDS diagnosis to infected children is indeed a complex process that presents a challenge to both families and health care providers (Myer et al. 2006; Wiener et al. 2007). Further research in this area commented on the difficulty in determining the role of disclosure considering the confounding effect of child age (Bikaako-Kajura et al. 2006; Mellins et al. 2004). In this study, disclosure of HIV status to the child above 6 years of age enhanced the child’s cooperation with treatment. However, this variable was not included in the multivariate analysis as its age-relation would have required stratification for which our sample size did not allow.

Prior research concerning the burden of ART medication regimen and its association with adherence has been inconsistent (Brackis-Cott et al. 2003; Gibb et al. 2003; Mills et al. 2006b). We found an association between non-adherence and the complexity of medication regimens. A lower pill burden may enhance patient adherence to therapy as so, developing fixed-dose combinations that are appropriately formulated for children is a priority (Boni et al. 2000).

Our data shows a positive association between the caregiver reporting being HIV-infected and the perfect dose adherence to the child’s ART regimen. Other findings have revealed that those patients who had accepted their HIV-seropositivity and disclosed their serostatus not only improved adherence for themselves (Stirratt et al. 2006) but also predicted better child adherence (Martin et al. 2007). These results strengthen advocacy for the implementation of counseling and education for caregivers regarding treatment benefits to help them accept their HIV seropostive status and as so, improve pediatric ART adherence.

Financial constraint as a barrier to ART adherence was highlighted by caregivers in this setting and has been also reported in recent studies (Brigido et al. 2001; Orrell 2005). Efforts should be undertaken to adopt a family-oriented approach that is free of charge or implement mechanisms which can identify indigents and be able to support minimal financial needs. Equally noteworthy is that ART availability in sites treating HIV/AIDS patients needs to be well secured to avoid stock-outs that will have an effect on the quality of the care and management of these patients (Mills et al. 2006a).

Our study confirms the crucial role the caregiver has in pediatric ART adherence taking in consideration the knowledge of the disease, the perception and belief in the efficacy of the treatment as well as the motivation and the capacity he/she has to facilitate the regimen prescribed (Katko et al. 2001; Mellins et al. 2004). An important number of caregivers interviewed considered it a duty to be involved in the treatment of the child. As such, psychological and social support for the child as well as the caregiver could be beneficial.

The health care system has an important role in determining adherence. Decentralization of pediatric HIV treatment from third-line health services to district level is a major challenge to improve coverage of pediatric ART in resource-poor settings. This requires monitoring medication adherence as well as providing strategies to enhance adherence. Trust and communication with health providers as well as using directly observed therapy or watching children taking therapy have been reported to improve adherence (Myung et al. 2007). These interventions require trained and committed health staff and time to be spent with patients and their caregivers. Further data are needed to evaluate the long-term sustainability of this approach.

Potential limitations to the survey need to be acknowledged. Among them, its cross-sectional design which only allows to evaluate the association between adherence and other factors measured simultaneously, rather than the variability and consistency of adherence over time and the use of self-reported adherence measures that may overestimate the true prevalence of adherence (Farley et al. 2003). The sample size was limited with the consequences of having reduced power to identify factors possibly associated with adherence, as well as having restricted the accuracy of our estimations. Despite a small sample size, the sample was likely representative of children and adolescents in care at the participating centers, as refusal rates of participation were extremely low.

In conclusion, the results presented here will set the basis to design specific interventions that will enhance adherence to ART amongst children, among them: (1) educating and motivating caregivers through the implementation of education programmes including the development of teaching tools, to support adherence for the HIV-positive child in need of ART, prior to starting the treatment (Nicholson et al. 2006); (2) simplifying treatment regimens when possible and tailoring ART regimen to the daily activities of the child and his/her family (Pontali et al. 2001); and (3) preparing the child as well as the caregivers for the management of side effects. Finally, the overall goal of this study was to explore the important factors that influence ART adherence for children in order to develop appropriate adherence interventions. In light of the current results, it must be recognized that adherence relies critically on a secure and reliable drug supply. Even the most dedicated, motivated, and skilled caregivers cannot provide the medications needed by HIV positive children in their care if they are out of stock or otherwise not available to them. Clearly efforts to address child ART adherence must take medication access into account while also providing for the educational, skill development, and social support needs of the caregivers.

Acknowledgments

The authors would like to acknowledge the French association “Initiative Développement” as well as the staff of the NGOs “Action contre le SIDA”, “Aides Médicales et Charité” and « Espoir Vie Togo”, based in Lome, Togo for their support for this study. We are grateful to the caregivers and children for their participation in the interviews. Comments during the preparation of the manuscript from Isabelle Gobatto, Marthe-Aline Jutand, Anne Tijou-Traoré and Francois Dabis are greatly valued. We also thank two AIDS and Behavior reviewers for their useful comments and suggestions.

Copyright information

© Springer Science+Business Media, LLC 2008