Introduction

The pediatric HIV epidemic continues to be of global health significance, particularly in resource-limited countries. Sub-Saharan Africa is disproportionally affected; it is home to over 90 % of pediatric HIV cases worldwide [1]. Although there has been increased access to antiretroviral therapy (ART) in resource-limited settings, pediatric HIV ART coverage still lags behind that of adults. At the end of December 2013, only 23 % of HIV-infected children were receiving ART [2]. With the recent World Health Organization (WHO) recommendation to start ART in children regardless of WHO clinical stage or CD4 cell count [3], pediatric ART coverage will increase worldwide. As ART coverage increases, two perennial challenges with treatment—HIV status disclosure and adherence to treatment—are bound to escalate [4, 5]. Treatment success depends on an unwavering cooperation of the patient and optimum adherence to treatment doses and schedules. One can speculate that when a patient does not understand the reasons for taking a prescribed medication, the cooperation needed could be compromised. This is akin to giving medication to HIV-infected children without disclosing their HIV status to them. Moreover, with HIV treatment, an optimum adherence of ≥95 % is required to achieve sustained virologic suppression and to avoid evolution of drug-resistant HIV variants [6, 7].

Since 1999, the American Academy of Pediatrics has recommended age-appropriate disclosure of HIV status to children, with full disclosure occurring by adolescence in order to achieve better disease outcomes and assist with the child’s psychological adjustment to knowing his or her status [8]. In 2011, the WHO published a similar set of guidelines which recommended beginning the disclosure process at 6 years old, with full disclosure accomplished by the age of twelve [4]. However, several studies from resource-limited settings have reported unacceptably low levels of disclosure of HIV status to HIV-infected children [5, 9, 10]. Reasons that caregivers do not disclose their child’s HIV-positive status to the child include fear of social rejection and isolation, parental sense of guilt, worry that the child will not keep their diagnoses to themselves, and concerns on how disclosure will affect the psychological health of the child [9, 11, 12]. There are emerging reports on the benefits of disclosure of HIV status, including psychological health, improved adherence and better clinical outcomes [13–15]. However, the prevalence of pediatric HIV disclosure is poor, ranging from 0 to 69 % in one systematic review [12].

Despite the fact that adherence is the sine qua non for successful HIV treatment and prevention of disease progression, it continues to be low in pediatric HIV, with one study in Ethiopia finding only 34.8 % of children to have an adherence rate of ≥95 % using unannounced home-based pill count [16]. The low prevalence of pediatric adherence is due to complex and inter-related factors, which are often beyond the control of the child [17–20]. Examples of factors affecting adherence in HIV-infected children are: (1) total dependence on the caregiver for procurement and administration of medication [20]; (2) lack of appropriate pediatric drug formulation and the taste of available formulations [19, 21]; and (3) socioeconomic status of the caregiver and access to service delivery [18, 20]. Moreover, in resource-limited settings where access to more expensive second-line regimens are limited, adherence to first-line regimens is critical. Thus children and their caregivers are constantly negotiating on ways to ensure optimum adherence to therapy. One factor that might help with the success of this negotiation is disclosure of HIV status to the child. Disclosure might help the child understand why he or she is taking daily medications and the benefits thereof.

There have been multiple studies in resource-poor settings seeking to find associations between disclosure status and levels of ART adherence. A number of studies have reported both psychological and medical benefits of disclosure of their HIV positive status to children and adolescents [9, 12]. However, many of these studies have been qualitative in nature and have not shown quantifiable associations of status disclosure with medication adherence, instead using caregiver and child interviews as the basis for establishing the connection between the two [21–24]. However, as disclosure becomes a more prominent focus in pediatric HIV treatment, there is a need for more longitudinal and quantitative data in order to establish the associations and effects of disclosure on adherence. In this way, interventions may be designed to utilize the disclosure process as a chance to develop better adherence habits in children and adolescents on antiretroviral treatment. This review sought to synthesize published quantitative data on the relationship between disclosure and adherence in order to better understand the impact the disclosure process is having on HIV-infected children in resource-limited countries.

Methods

Primary Search Strategies, Information Sources, and Inclusion Criteria

We conducted a systematic search of electronic medical databases, including PubMed (inception-October 2015), MEDLINE (January 1998–October 2015) and the Cochrane Database of Systematic Reviews (January 1998–October 2015). Search strategy involved the following phrases and Boolean operators: pediatric* AND* medication* AND* adherence, pediatric* AND* hiv* AND* disclosure, and caregiver* AND* hiv* AND* disclosure. Searches were conducted by JSN, and the final search was completed on October 13, 2015. The authors also examined the references and bibliography lists of two identified systematic reviews to maximize identification of relevant articles for inclusion [9, 12]. Research protocols and study inclusion criteria were determined a priori by the authors, but the protocol was not previously published in a systematic review database.

Publications were eligible for selection if they included quantitative data on the association between HIV status disclosure and adherence to ART in pediatric patients in resource-limited settings. The age range considered was 0–19 years with both children (0–9 years) and adolescents (10–19 years) included. Additionally, the study had to have been conducted in a low- or middle-income nation as determined by 2015 World Bank classification [25]. Only studies published since 1999, the year of the American Academy of Pediatrics disclosure recommendation publication [8], were considered. Disclosure in the included studies was defined as the child knowing that he or she has the HIV infection, as determined by either caregiver or child report. Furthermore, the study had to have a defined measure of adherence, such as patient report, caregiver report, pill count or electronic methods. Each study included a threshold for what was considered “adherent” versus “non-adherent” based on the method of adherence determination that the authors had used. Cross-sectional and observational study designs were considered for inclusion.

Article Review and Data Extraction

Studies included in the review were obtained through a two-step screening process. Initial search results were scanned for possible inclusion. Studies that clearly did not relate to HIV, studies conducted only in an adult population, studies published prior to 1999, studies conducted in resource rich nations, and studies with only qualitative data were excluded on initial examination (Fig. 1). Publications that passed the initial screening were then more thoroughly examined for predetermined inclusion criteria by two independent researchers (JSN and ARS). Disagreements over inclusion were settled by consensus. After the second review of the articles, records were excluded [10, 13, 14, 22–24, 26–39] for lack of quantifiable associations of adherence with disclosure or conduction of the study in a “high income” setting as classified by the World Bank system [25]. Data regarding study population and setting, medication adherence measurement tools, medication adherence levels, disclosure levels and quantified associations between disclosure and adherence were extracted for review and analysis. Risk of inherent bias regarding medication adherence determination used in the studies was noted during review of the articles and is discussed further below. Quality of the studies was noted during review by evaluating study design, sample sizes, inherent study bias and publication bias. Descriptive statistics regarding sample sizes of the studies included were calculated using Microsoft Excel (2013). No statistical meta-analysis of data was performed due to different methods of measuring adherence, different age ranges in the study populations (which would skew the rates of disclosure), and, ultimately, different methods used to calculate associations between adherence and disclosure.

Fig. 1
figure 1

Search Algorithm

Full size image

Results

Studies Included in the Review

Figure 1 represents search results using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart paradigm [40]. Initial keyword searches in the three databases yielded a total of 2228 results (PubMed = 1392; MEDLINE = 789; Cochrane = 47), while record retrieval from two systematic reviews [9, 12] yielded 63 results, for a total of 2291 records. After excluding repeated titles, 1348 records remained. From the initial screening of titles and abstracts, 1314 of these results were excluded due to non-HIV related topics (N = 878), publication date prior to 1999 (N = 69), conduction of the study in a resource-rich setting (N = 81), exclusively adult study population (N = 98), lack of data on disclosure rates or adherence rates (N = 110), qualitative study design (N = 76), and case report study design (N = 2). Thirty-four articles were then saved for full-text review by JSN and ARS. After the second review of the articles, studies were excluded [10, 13, 14, 22–24, 26–39] for poorly defined adherence (N = 2) [23, 38], lack of quantifiable associations of adherence with disclosure (N = 15) [10, 13, 14, 22, 24, 26–28, 30–32, 34, 35, 37, 39] or conduction of the study in a “high income” setting (N = 3) [29, 33, 36] as classified by the World Bank system [25]. Fourteen studies met all of the selection criteria (Table 1).

Table 1 Characteristics of study populations
Full size table

Study Designs and Sample Characteristics

The study designs were predominately cross-sectional (12 of 14), with only 2 studies being prospective observational cohorts (Table 1). There was a wide variation in the sample sizes of the studies included with a mean sample size of 293.93 (SD, 219.74) and a range of 57–792. The total number of children and adolescent included in the studies was 4115. The ages ranged from 2 months to 19 years old. With regard to region of study, 85.7 % were from Africa (East Africa, 7; West Africa, 2; and South Africa, 3) and 14.3 % from Asia (India, 1 and Thailand, 1). Of the studies from Africa, 4 of 12 were conducted in Ethiopia. Mother-to-child transmission and blood transfusions were the modes of HIV transmission considered in 13 of the studies, while 1 study also included horizontally infected adolescents in their sample population [41].

Adherence Measures and Prevalence

Levels of adherence and methods for adherence measurement varied widely in the various studies examined (Table 2). Three of the studies used multiple methods in order to determine level of adherence [15, 42, 43]. In total, 10 studies reported adherence values based on various forms of caregiver recall or report [15, 42–50]; 3 used pharmacy or clinic-based pill count [15, 43, 51]; 2 utilized child report [41, 42]; 2 performed unannounced home-based pill count [16, 43]; 2 utilized Medication Event Monitoring Systems (MEMS, Aardex, Switzerland) [15, 32]; and 1 study used Visual Analogue Scale (VAS) [15]. Studies indicated that they considered “adherence” to involve either a patient taking ≥95 % of medication doses over a given period of time [15, 16, 32, 41, 43–45, 48, 51] or no missed doses on recall in a specified time window [42, 46, 47, 49, 50], although the length of time patients were followed differed between the studies. Most authors chose to report their data as “percentage of patients obtaining adherence” (Table 2), and data for these studies ranged from 34.8 % [16] (using announced home-based pill count) to 94.1 % [43] (using clinic-based pill count). Five of the studies reported the percentage of patients adherent separately in disclosed versus undisclosed children [16, 42, 43, 50, 51]. Three studies chose to examine adherence as a continuous variable with each patient having a percentage adherence calculated. Two of these studies reported median adherence [15, 51] and one reported mean adherence [32] with a range of 81–99.2 % (Table 2).

Table 2 Associations of medication adherence and status disclosure
Full size table

Prevalence of Disclosure

Disclosure in the studies was defined as the child knowing that he or she is infected specifically with HIV, determined by either child or caregiver report. In these studies, disclosure was generally low, ranging from 8.4 to 79 % (Table 2). Haberer et al. did not report percentages of disclosure, but instead gave an absolute number of disclosed children within the specific age range of 9–15 years old [15]. While nine studies contained children below the age of six, the WHO threshold age for the beginning of the disclosure process [4], in their study population [15, 16, 32, 43–48], only three of these studies [15, 43, 47] explicitly stated that they corrected their statistics on disclosure to reflect only patients in their study at least 6 years of age or older.

Association of Disclosure with Adherence

The association of disclosure with adherence was found to vary widely between the studies (Table 2). Five studies showed no statistically significant relationship between disclosure and adherence [32, 43, 49–51]. Two studies found an association of disclosure with non-adherence [42, 45], while two others showed non-disclosure to be associated with adherence [16, 46]. The remaining five studies reported either a positive association of disclosure with adherence [15, 41, 44, 48] or an association of non-disclosure with non-adherence [47]. Thus, disclosure was shown to be associated with better adherence in five studies, but shown to be associated with worse adherence in four studies. Seven studies included disclosure and adherence in a multivariate logistic regression model to determine the relationship between the two factors independent of other demographic factors that were found to have statistically significant impacts on adherence [15, 16, 41, 42, 44–46]. In these analyses, Vreeman et al. (OR = 1.31 [0.86–1.98]) showed no association [42]; Biressaw et al. (aOR = 2.35, [1.09, 5.06]) [16], Dachew et al. (aOR = 0.27, [0.24, 0.32]) [45], and Biadgilign et al. (aOR = 2.53, [1.24, 5.19]) [46] showed disclosure and adherence to be negatively associated; while Arage et al. (aOR = 3.47, [2.10, 6.81]) [44], Haberer et al. (IRR = 0.62, [0.46–0.81]) [15], and Cluver et al. (OR = 2.18, [1.47–3.24]) [41] showed positive associations between disclosure and adherence.

Discussion

Adherence to ART and disclosure of HIV status to HIV-infected children continue to pose a challenge in the management of pediatric HIV in resource-limited countries. It is critical to understand the psychological and treatment effects that disclosure can have on these children and adolescents to inform pediatric HIV management and ensure optimal outcomes. While psychological, medical and social positive impacts of disclosure have been reported mainly in qualitative studies [12, 22, 35, 39], this review was undertaken to synthesize quantitative data showing the effect that disclosure is having on medication adherence.

While we expected to find a clear benefit of disclosure on adherence outcomes in these patient populations, the outcomes reported by the studies reviewed show mixed results in their association between disclosure and adherence. Five studies reported no association [32, 43, 49–51], five showed a positive effect of disclosure on adherence [15, 41, 44, 47, 48], and four reported decreased adherence in children who knew their HIV status [16, 42, 45, 46]. The studies reporting a negative effect provided several explanations for this, including children’s fear of social stigma and, therefore, the tendency to hide taking medications from others [42], as well as denial of their status, leading to increased refusal to take medications [16]. Another surmised factor was that disclosure might increase depressive symptoms and lead to decreased desire to develop strong adherence habits [42]. Several reasons were given for the positive association between adherence and disclosure, such as disclosure empowering the child to partner with their caregiver in maintaining adherence, greater willingness to take medications due to knowing why the medications were being taken, and more open discussions about adherence occurring between the caregivers, the healthcare providers, and the children [15, 32, 41].

The quality of the studies evaluated was generally low and may limit the generalizability of results, while reinforcing that longitudinal studies are needed to make definitive conclusions about the associations of disclosure with adherence. Twelve of the fourteen studies were cross-sectional, thus limiting the evaluation of causation between status disclosure and development of adherence habits. The methods of determining adherence were variable, with eight of the studies choosing to exclusively use caregiver or child recall to determine adherence (Table 2). These recall measures are subject to social desirability bias and may thus overestimate adherence [43], further decreasing the quality of evidence. Sample sizes were small and sub-analyses were not conducted, limiting the conclusions that can be drawn from the data. Additionally, it is worth noting that only three of the records reviewed examined the association of disclosure with adherence as their primary endpoint [41, 42, 51]. Of these three, Vreeman, et al. showed no association by caregiver report and negative association by child report [42]; Sirikum et al. showed no association by pharmacy-based pill count [51]; and Cluver, et al. showed a moderate association between the two based on patient report, especially when disclosure occurred before the age of twelve (OR = 2.65, 95 % CI [1.34–5.22]) [41]. The latter study, undertaken with patients 10–19 years old in Eastern Cape, South Africa, was also the most broad-reaching study, with all pediatric HIV patients in the Eastern Cape contacted for possible enrollment [41]. Only the Sirikum et al. study, conducted in Thailand, measured adherence pre- and post-disclosure in a prospective manner. This study showed no statistically significant change in adherence at 6 or 12 months post-disclosure; however, the authors noted that this lack of association may have resulted from unusually high levels of adherence reported at baseline [51]. The lack of prospective data on adherence levels after disclosure and the low quality of evidence in the studies found highlight the need for further longitudinal research to be done.

Due to lack of information on adherence values, there were several studies ultimately omitted from our review that nevertheless showed some degree of quantitative association between disclosure and disease outcomes. One study in Romania reported a decrease in disease progression and an increase in survival for children and teens who knew their HIV-positive status [13]. A study in South Africa reported an increase in viral suppression for those patients who had gone through the process of disclosure [32]. A study conducted in Uganda demonstrated in their sample population (N = 42) some patients who were “non-disclosed” or “partially disclosed” reported “frequently” missing medication doses, while none of the patients who were “fully disclosed” reported missing frequent doses [23]. Moreover, a multinational study conducted in Cote d’Ivoire, Senegal and Mali published a statistically significant association between disclosure and retention in care at pediatric ART clinics [14]. These studies add additional evidence to the benefit of disclosure on health outcomes, though no conclusive link between the two can be drawn from these studies and those included in the review.

Several limitations to this review warrant consideration. Only 14 studies met inclusion criteria, and thus there was a limited amount of evidence from which to draw conclusions. There was lack of geographical diversity in the studies that met the inclusion criteria. Eleven of the studies were conducted in sub-Saharan Africa, which may limit generalizability beyond this setting. Another weakness is the use of caregiver or patient report to determine adherence in many of these publications, which has been shown to overestimate adherence and have a weaker correlation with viral load suppression than pill count or MEMS methods of measuring adherence [16, 43]. A publication bias may be present, as studies that found no association or a negative association of disclosure with adherence may not have published their results. While we sought to be as rigorous as possible in our search, it is possible that there were studies that were missed using our search criteria. Taken together, the review has several strengths; one of the strengths of the study is the utilization of quantitative data in order to examine the relationship of disclosure and adherence independent of other demographic and psychosocial factors that may also influence adherence. In addition, the review reveals an important need for more high quality, longitudinal data to conclude any form of causation between the disclosure process and adherence development.

Recommendation for Practice and Research

A major finding that this study revealed is the paucity of data measuring effects of disclosure on treatment outcomes. Most of the associations reported between disclosure and adherence have come from cross-sectional data, and very few longitudinal studies have quantitatively examined the effects of disclosure on adherence habits and behavioral changes over time. Future research should investigate ways of integrating disclosure counseling with medication adherence counseling, while finding ideal measures for adherence to reduce the incidence of social desirability bias inherent in some of the caregiver and patient recall measures in use.

Pediatric HIV programs should not consider disclosure as a one-time event, but a process where both caregivers and children are equipped with the knowledge and skills to maximize the treatment benefits of disclosure. Disclosure and adherence interventions should be culturally appropriate, be executed in tandem, and personalized, taking into consideration the cognitive and developmental stage of the child. Several such support interventions are being piloted in some resource-limited settings [30, 35, 52]. A trusting and collaborative relationship between healthcare systems, caregivers, HIV-infected children, and healthcare providers should be established in order to increase adherence through the process of disclosure, thereby achieving better health outcomes for children and adolescents living with HIV.