Africa has the world's highest rate of armed conflict, with nearly 1 quarter of African countries experiencing conflict in 2007.1 Individuals and families in regions of armed conflict face both acute and long-term risks to health, including injury, separation, food insecurity, and decreased access to health care. Many are displaced from their homes by fear or violence and must live in makeshift homes or camps for refugees/internally displaced people, where they face increased risks, such as infection, malnutrition, and food insecurity.2–4 Children are particularly vulnerable in crises and are at increased risk of abandonment, abduction, malnutrition, and communicable diseases.4
Ninety percent of the world's 2.5 million HIV-infected children live in sub-Saharan Africa. Twenty-six percent of these children receive antiretroviral therapy (ART).5–7 Numerous studies in Africa have shown success with ART in adults and children, with excellent adherence and associated clinical and immunological benefits.8–12 Further studies in Africa have assessed risk factors, including stigma and distance to care, and protective factors, including HIV-positive caregivers and early disclosure, involved with ART in children in nonconflict settings.11,13,14 Methods of ART delivery have also been assessed, including family centered care, in which a multidisciplinary approach to treating HIV-positive caregivers along with their HIV-positive children has led to improved outcomes in children with HIV.11,15–17 However, despite protective factors or the method of ART delivery, children with HIV remain vulnerable in situations of armed conflict due to increased risk of opportunistic infections, and, for those treated with ART, due to the potential risk of interrupted treatment and resultant viral resistance.18
In 2006, the World Health Organization released a consensus statement asserting that “ART delivery should be included as part of comprehensive HIV services in emergency settings”.19 There is only fledgling research on how best to implement these services and what the immediate and long-term effects of conflict on children treated with ART may be.20 A study on the immediate impact of the political and humanitarian crisis after the contested 2007 Kenyan presidential election showed small but statistically significant disruptions in clinical care and medication adherence among children on ART despite a comprehensive health care response.21 A study in a Kenyan slum during the same conflict found that the severity of the violence and subsequent decreased access to health care resulted in more significant LTFU of patients and treatment interruptions.22 In contrast, a study in northern Uganda assessed pediatric ART adherence in a setting of chronic conflict that was punctuated by periods of more extreme violence and found high adherence and low mortality, comparable with pediatric ART outcomes in politically stable sub-Saharan African countries.23 Studies among adults in conflict and postconflict settings including northern Uganda and the Democratic Republic of Congo have shown that with creativity and emergency preparedness planning, adherence, mortality, and LTFU data are similar to stable resource-limited settings.20,24–26 Together, these studies highlight the variability in the nature and severity of conflicts, in regional and organizational response capacities, and in the availability of resources—all of which can affect the provision of continued HIV treatment in conflict settings. There are multiple gaps in our understanding of the role of this variability and the impact of conflict. Some studies have assessed long-term adherence in chronic conflict settings, but have not compared that to adherence immediately after conflicts. One study differentiated the characteristic of orphan status, but no study has assessed how other social and physical contexts, such as tribe and location, mediate the effects of a conflict.
THE 2007 KENYAN POSTELECTION CRISIS
Kenya has a history of politics fueled by ethnic tensions.27 Allegations of election malpractice on December 27, 2007, led to widespread violence between tribes.28 According to government figures, more than 1300 people were killed in the postelection violence and more than 500,000 were displaced to various locations, many of these were not equipped for the influx with little security or food and poor sanitation. On February 28, 2008, incumbent government and opposition leaders formed a coalition government, ending most of the violence, and a resettlement program began in May 2008.29 In addition to the postelection crisis, there had been separate ongoing violence at the time of the election in Mount Elgon, a region in western Kenya.30,31
With such significant violence and displacement affecting the people in western Kenya, organizations such as the United States Agency for International Development (USAID)-Academic Model Providing Access to Healthcare (AMPATH) Partnership, a health care system providing services that include comprehensive HIV care to more than 130,000 HIV-infected adults and children, staged an immediate and multifaceted response. This response was conducted without significant preparation before the election or an established emergency protocol, although anecdotal evidence suggests that some clinics may have provided additional medication to patients due to concern about potential upheaval after the election.21 The AMPATH response included emergency provision of medicines; publicized instructions for HIV-infected patients through radio, newspaper, local television announcements, and a national hotline; sending medical teams to camps for internally displaced persons, satellite clinics, and patient homes; provision of food and social support services; and reopening most of the AMPATH clinics within a week of the onset of violence. The response is further detailed in previous publications.20,21
In this study, using routine electronic medical records, we assessed outcomes of the postelection crisis on clinic follow-up and medication adherence for HIV-infected children in western Kenya from the time of the crisis until 1 year after the crisis. We also studied whether children living in different social and physical environments demonstrated differences in adherence to either ART or clinic attendance during this period.
We conducted a retrospective study of a cohort of HIV-infected children treated in the AMPATH clinical care system before, immediately after, and in the subsequent year after the postelection crisis in Kenya. For all children in this cohort, we assessed loss-to-follow-up (LTFU) and clinic adherence. Among children on ART, we also assessed overall medication adherence and changes in medication adherence between periods. The 3 periods assessed were as follows: period 1/pre-election (October 26, 2007, to December 25, 2007); period 2/immediate postelection (December 26, 2007, to April 15, 2008); and period 3/long-term postelection (April 16, 2008, to December 31, 2008).
Moi University School of Medicine in Eldoret, Kenya, and Indiana University School of Medicine have collaborated since 1990.32 In 2001, the USAID-AMPATH was created as a joint initiative between these 2 medical schools and Moi Teaching and Referral Hospital, in partnership with USAID, to provide an HIV care system for patients in western Kenya.33–35 AMPATH serves a catchment area of more than 13 million people. Since 2001, more than 135,000 pediatric and adult patients have been treated within AMPATH, with 24,591 children under the age of 14 years now receiving care, and 7570 children currently on ART (as of June 2011). Comprehensive HIV care services, including free ART for all patients in need, are provided at Moi Teaching and Referral Hospital and at 26 urban and rural clinics, and at 23 satellite clinics that typically operate out of tents or more make-shift spaces at other facilities.33,36 A computerized medical record system supports clinical care and research,37 and the outcomes and adherence of adult and pediatric patients have previously been reported.38–40 Clinicians use standard paper encounter forms at all AMPATH clinic visits (http://amrs.iukenya.org/download/forms), recording information from patient interviews and exams. Dedicated data entry clerks then enter these data into the electronic AMPATH Medical Record System, with data entry validated by a random review of 10% of all data entered.
Eligible patients for the cohort included confirmed HIV-infected children younger than 14 years who were seen at any of the AMPATH clinics in period 1. Orphans were defined as children having a mother or a father dead. Children were defined as belonging to a minority tribe if they belonged to an ethnic group that constituted less than 10% of the population at the clinic they attended. Nonminority tribes were ethnic groups that constituted more than 10% of the clinic's total population. The tribes Luhya, Kalenjin, Luo, and Kikuyu are the 4 largest tribe groups seen within the AMPATH clinical care system.
Data Collection and Measures
The standard encounter forms used at all AMPATH clinic visits include demographics, medical history, social history, physical exam, laboratory data, symptom history, and medication adherence. Dedicated data entry clerks enter this information into the AMPATH Medical Record System (AMRS), an electronic data repository. All data for the study was collected using AMRS.
Two outcome variables of LTFU were assessed: initial loss-to-follow-up (late FU) and complete loss-to-follow-up (complete LTFU). Late FU was defined as patients who did not come for an appointment during period 2 but did return for an appointment during period 3. Complete LTFU was defined as patients who did not return to clinic in either period 2 or 3.
Poor Clinic Adherence
Children on ART are typically seen on a monthly basis in AMPATH, and HIV-infected children not on ART are seen every 2 to 3 months.21 Therefore, poor clinic adherence for patients on ART was defined as having less than 3 visits in period 2 (3.5 months in length) and less than 8 visits in period 3 (8.5 months in length). For children not on ART, poor clinic adherence was defined as less than 2 visits in period 2 and less than 3 visits in period 3. We calculated the percentage of children with poor clinic adherence (children with poor clinic adherence/total children) for each clinic adherence measure.
Children within the AMPATH system typically receive a 1-month supply of ART during clinic visits. Caregivers or children are asked about adherence at their monthly visits as part of their routine clinical encounter. Using the AMPATH clinical data, ART adherence was assessed for those children prescribed ART based on caregiver or child responses to the question, “During the last 7 days, how many doses of his/her antiretroviral medicines did the patient take?” The response options are: “none,” “few,” “half,” “most,” and “all.” ART adherence was defined as either “perfect” or “imperfect”. A child was defined as having perfect adherence if s/he reported “all” doses taken, and imperfect adherence if s/he had 1 or more visits where adherence was reported as “none,” “few,” “half,” or “most” of the doses. There is no validated measure to assess pediatric ART adherence in resource-limited settings, but this measure has been used in previous studies.40 The outcome variable of change in ART adherence was defined as having less perfect medication adherence in period 2 as compared with period 1 or having less perfect adherence in period 3 as compared with period 1. We calculated the percentage of children with less perfect ART adherence (children with less perfect ART adherence/total children on ART).
We performed descriptive statistics to illustrate the percentage of children who were lost-to-follow-up; had poor clinic adherence; and had a change in ART adherence by sociodemographic factors including age, gender, ART status, orphan status, and tribe; using Stata/SE 10 (Stata Corp, College Station, TX). Means between 2 groups were compared for statistically significant differences using t tests. Fisher exact tests were used to determine statistically significant differences among unordered categorical variables for small numbers (eg, tribe).
Two thousand five hundred forty-nine HIV-infected children younger than 14 years were seen at the AMPATH clinics in the period before the election. (Table 1) Thirty percent were orphaned at the time they were enrolled in care in AMPATH. The majority (63%) of children were on ART and were members of a nonminority tribe of their region (91%).
Children on ART had less initial LTFU (3.0% compared with 5.1%) and less complete LTFU (2.6% compared with 6.8%) than children not on ART. (Table 2) Among children not on ART, orphans had less LTFU than nonorphaned children, and those who only had a mother living had greater LTFU than those who only had a father living. Among children on ART, orphans tended to have greater LTFU in period 2, but less LTFU in Period 3. These differences, however, were only statistically significant for LTFU in period 2 as compared with no LTFU among children with neither parent living or only mother living. Members of the minority Kikuyu tribe had greater LTFU in both periods regardless of ART status.
Children on ART, with a more strict measure of clinic adherence, had higher percentages of low clinic adherence in period 2 and period 3 than children not on ART. For children not on ART, there were no significant differences based on orphan status between period 3 and period 1. Between period 2 and period 1, there was an increase in low clinic adherence for children with both parents living, and a decrease in low clinic adherence among those with only a father living. There was a significant difference when assessing tribal status between period 2 and period 1, but no significant difference when assessing tribal status between period 3 and period 1. For children on ART, there were no significant differences based on orphan status between period 2 and period 1. Between period 3 and period 1, there was an increase in low clinic adherence for those with both parents alive and a decrease in low clinic adherence for those with neither parent living or only mother living. There was a significant difference when assessing tribal status between period 2 and period 1 and between period 3 and period 1.
In period 2, 8.2% of children on ART had imperfect medication adherence. In period 3, 9.0% of children on ART had imperfect medication adherence (Table 4). Orphans had less perfect ART adherence in period 3 compared with period 1 and had less perfect adherence in period 3 than in period 2. Differences among orphan groups were not present, however, when period 2 was compared with period 1. Likewise, there were no significant differences in tribal status among children with less perfect versus improved or unchanged medication adherence in either periods 2 or 3, as compared with period 1.
Examining the longitudinal impact of a period of widespread violence and displacement of people reveals which HIV-infected children may be most at risk for disrupted clinical care and nonadherence to medications and when those children may be the most vulnerable.
Few data exist to characterize pediatric LTFU from HIV care systems in resource-limited settings, although there have been several studies on LTFU in adults, with various definitions of LTFU.41 Our analyses characterize both short-term and long-term LTFU for HIV-infected children over the course of a humanitarian crisis and then through its aftermath. Children on ART had less LTFU in both the immediate and long-term postconflict periods, which may reflect the increased education given to caregivers of children on ART about the importance of medication adherence and/or caregivers' perceptions of the increased health risks these children face. Although caregiver HIV status was unknown in this study, increased knowledge of symptoms and importance of medication adherence in caregivers who are HIV positive may have contributed to less LTFU among children on ART during this conflict and warrants future study.11
Somewhat surprisingly, orphans had less LTFU in the immediate-term and long-term postconflict periods. These findings support previous work done in our care system that suggested that orphans were not at greater risk of clinic nonadherence even though they were at higher risk of ART nonadherence.9,39 We have, however, previously seen higher rates of LTFU and less retention in the care system among orphans followed over longer periods of time.42 In this study, those who lost a father were more vulnerable to LTFU in the immediate period after the conflict. It is possible that the loss of a father makes a family more vulnerable in times of security risks and economic hardship, such as those seen immediately postconflict. This may be expected in a patriarchal culture such as that of western Kenya, where the father or male head-of-household has predominant control over family resources. Increased vulnerability in a household without a father could translate into difficulty coming to clinic. This finding also underscores the importance of carefully defining orphanhood to accurately examine the effects of losing either parent or both parents.
Between tribes, children in the Kikuyu tribe had greater LTFU in both period 2 and period 3, regardless of ART status. The majority of USAID-AMPATH clinics are in western Kenya, where traditional nonminority tribes include Luo, Luhya, and Kalenjin. Kikuyu are traditionally a minority tribe in western Kenya and may have experienced disproportionate displacement during the crisis as a result.43 This further underscores the need to understand the political and ethnic composition of a region to identify the at-risk populations that may need more specialized anticipatory care.
Children not on ART had improved clinic adherence in period 3 compared with period 2, likely reflecting the increased political, humanitarian, and economic stability. Children on ART had worse clinic adherence over time. This is likely related to the more strict measure of clinic adherence for children on ART. Assessing these data in combination with LTFU data shows that although children on ART had decreased clinic adherence over time, follow-up was still maintained. A greater percentage of those with both parents living had lower clinic adherence. Factors affecting clinic adherence may include additional resources the clinic offers such as nutritional support, school fees, and social work assessments that nonparental caregivers or single parents of orphans may be seeking.
Medication adherence data during the conflict for those on ART was comparable with data previously reported in this population years before conflict.9 Despite the stability of their follow-up and clinic adherence, orphans did have decreased ART adherence over the long term. Although decreased medication adherence is expected over time, this was most pronounced among orphaned children. This finding may be due to decreased parental supervision in nonparental caregiver or single parent homes. Alternatively, orphan caregivers may be more likely to report missed medication doses or less influenced by social desirability biases when divulging nonadherence. Whether economic risk factors such as increased food insecurity might impact medication adherence, but not clinic adherence, for orphans needs to be explored.
Although these data do show disruption in clinic and medication adherence, the disruption was overall quite low, and, in the case of medication adherence, comparable with previous data during a nonconflict period.9 Furthermore, follow-up, in particular for those on ART, was largely maintained over the long-term despite the severity of the conflict and the lack of emergency planning before the conflict. The immediate program response within a well-established infrastructure may have mitigated some of the disruption. Key components of the response included opening clinics as soon as possible to provide access to medications and communication between clinics and patients via radio, newspaper, television, and a hotline to provide instructions and information on clinic openings. To meet the needs of displaced patients and to assist in overcoming their security concerns with travel, medical teams were sent to internally displaced people camps, patient homes, and satellite clinics.20,21 These results provide further evidence of the feasibility of HIV programs and the provision of ART in resource-limited settings prone to conflict and the importance of assessing risk factors for patients within these settings.
There are several potential limitations of our study. First, it is difficult to generalize to other countries from an acute conflict situation within the political and ethnic context of one country or even between regions of that particular country. This conflict took place in a particular location, time, and place, and its effects were studied within a unique HIV care system that is considered an international model. Never-the-less, the vulnerabilities of one pediatric population in a resource-limited setting during a time of conflict, particularly when considered against the dearth of data describing the effects of conflict on children, does merit careful examination. Furthermore, the barriers to clinic follow-up and adherence are consistent with those identified in research from other conflict settings.3 Second, we used a fairly blunt measure of ART adherence that may have missed early episodes of nonadherence before the patient's return to clinic. Nonetheless, these analyses use a consistent measure over time and do capture changes in reported medication adherence. This measure of adherence was complemented by the LTFU and clinic adherence measures to more completely assess clinical outcomes among HIV-infected children. Third, these data were limited by the data available in the pediatric electronic medical records and did not include data not collected on the routine clinical encounter forms, such as displacement from homes, caregiver HIV status, CD4 counts, viral load, WHO stages, or rates of switch to second-line and third-line medications, all of which warrant further study. Despite these limitations, the robust nature of the AMPATH clinical care program and medical records system allows for both retrospective analysis and long-term follow-up of the clinical characteristics of the patient population. These data from Kenya provide an important addition to understanding how HIV care systems and humanitarian aid organizations can meet the needs of HIV-infected children in future crises.
HIV-infected children are at risk for disrupted care and decreased medication adherence in both the immediate and long-term periods after a humanitarian crisis and are differentially affected based on their social and environmental characteristics. Advance planning, with special attention to identified at-risk groups, may help to minimize disruption of HIV care during humanitarian crises. Understanding the context of the ethnic and social make-up of an area will help to identify these at-risk children.
The authors give special thanks to the families and to the health care providers of AMPATH, including the nurses, clinicians, and pharmacy staff, all of whom work tirelessly to ensure that the children of western Kenya receive the medications they need. In particular, we would like to thank Dr. Esther Nabakwe, Dr. Constance Tenge, Dr. Irene Marete, and the other members of the AMPATH Pediatric Research Working Group.
1. Harbom L, Wallensteen P. Armed conflict, 1989–2006. J Peace Res. 2007;44:623–634. Available at: http://www.thefreelibrary.com/The+2008+Armed+Conflicts+Report–preview.-a0181622343
. Accessed March 2009.
2. Seaman J, Maguire S. ABC of conflict and disaster. The special needs of children and women. BMJ. 2005;331:34–36.
3. Olupot-Olupot P, Katawera A, Cooper C, et al.. Adherence to antiretroviral therapy among a conflict-affected population in Northeastern Uganda: a qualitative study. AIDS. 2008;22:1882–1884.
4. Toole MG, Waldman RJ. The public health aspects of complex emergencies and refugee situations. Annu Rev Public Health. 1997;18:283–312.
5. De Baets AJ, Ramet J, Msellati P, et al.. The unique features of pediatric HIV-1 in sub-Saharan Africa. Curr HIV Res. 2008;6:351–362.
6. UNICEF. Children and Aids: third stock-taking report. Available at: http://www.unicef.org/publications/files/CATSR_EN_11202008.pdf
. Accessed March 2009.
7. UNAIDS. Report on the global AIDS epidemic. Available at: http://www.unaids.org/globalreport/global_report.htm
. Accessed August 2011.
8. Wools-Kaloustain K, Kimaiyo S, Diero L, et al.. Viability and effectiveness of large-scale HIV treatment initiatives in sub-Saharan Africa: experience from western Kenya. AIDS. 2006;20:41–48.
9. Nyandiko WM, Ayaya S, Nabakwe E, et al.. Outcomes of HIV-infected orphaned and non-orphaned children on antiretroviral therapy in western Kenya. J Acquir Immune Defic Syndr. 2006;43:418–425.
10. Reddi A, Leeper S. Antiretroviral therapy adherence in children: outcomes from Africa. AIDS. 2008;22:906–907.
11. Reddi A, Leeper SC, Grobler AC, et al.. Preliminary outcomes of a paediatric highly active antiretroviral therapy cohort from KwaZulu-Natal, South Africa. BMC Pediatrics. 2007;7:13.
12. Sutcliffe CG, Van Dijk JH, Bolton C, et al.. Effectiveness of antiretroviral therapy among HIV-infected children in sub-Saharan Africa. Lancet Infect Dis. 2008;8:477–489.
13. Braitstein P, Songok J, Vreeman RC, et al.. “Wamepotea” (they have become lost): outcomes of HIV-positive and HIV-exposed children lost to follow-up from a large HIV treatment program in western Kenya. J Acquir Immune Defic Syndr. 2011;57:E40–E46.
14. Vreeman RC, Nyandiko WM, Ayaya SO, et al.. Factors sustaining pediatric adherence to antiretroviral therapy in western Kenya. Qual Health Res. 2009;19:1716–1729.
15. Leeper SC, Montague BT, Friedman JF, et al.. Lessons learned from family-centered models of treatment for children living with HIV: current approaches and future directions. J Int AIDS Soc. 2010;13(suppl 2):S3.
16. Reddi A, Leeper SC, Sunpath H. Pediatric highly active antiretroviral therapy (HAART) in Africa: potential benefits of a “family-centered” model. J Infect Dis. 2008;198:938–939.
17. Rochat TJ, Bland R, Coovadia H, et al.. Towards a family-centered approach to HIV treatment and care for HIV-exposed children, their mothers and their families in poorly resourced settings. Future Virol. 2011;6:687–696.
18. Prendergast A, Tudor-Williams G, Jeena P, et al.. International perspectives, progress, and future challenges of paediatric HIV infection. Lancet. 2007;370:68–80.
19. WHO, UNHCR, UNAIDS, MSF, and UNICEF. Consensus statement: delivering antiretroviral drugs in emergencies: neglected but feasible. September 20, 2006. Available at: http://www.who.int/hac/techguidance/pht/HIV_AIDS_101106_arvemergencies.pdf
. Accessed August 2011.
20. Mills EJ, Ford N, Singh S, et al.. Providing antiretroviral care in conflict settings. Curr HIV/AIDS Rep. 2009;6:201–209.
21. Vreeman R, Nyandiko W, Sang E, et al.. Impact of the Kenya post-election crisis on clinic attendance and medication adherence for HIV-Infected children in western Kenya. Confl Health. 2009;3:5.
22. Unge C, Sodergard B, Thorson A, et al.. HIV treatment in times of civil strife: serious threats to antiretroviral drug access in the Kibera slum following the Kenyan elections. AIDS. 2008;22:1693–1694.
23. Kiboneka A, Nyatia RJ, Nabiryo C, et al.. Pediatric HIV therapy in armed conflict. AIDS. 2008;22:1097–1098.
24. Kiboneka A, Jones R, Nabiryo C, et al.. Combination antiretroviral therapy in population affected by conflict: outcomes from a large cohort in northern Uganda. BMJ. 2009;338:B201.
25. Culbert H, Tu D, O'Brien DP, et al.. HIV treatment in a conflict setting: outcomes and experiences from Bukavu, Democratic Republic of the Congo. PLoS Med. 2007;4:e129.
26. O'Brien DP, Venis S, Greig J, et al.. Provision of antiretroviral treatment in conflict settings: the experience of Medecins Sans Frontieres. Confl Health. 2010;4:12.
27. Haugerud A. The Culture of Politics in Modern Kenya. Cambridge, United Kingdom: Cambridge University Press; 1995.
29. Waki P. Waki commission of inquiry into post election violence. Available at: http://www.dialoguekenya.org/docs/PEV%20Report.pdf
. Accessed March 2009.
30. Mount Elgon: Does Anybody Care? Medicines Sans Frontieres; May 2008; http://www.msf.org/source/countries/africa/kenya/2008/MSF_MtElgon_May2008.pdf
. Accessed March 2009.
32. Mamlin J, Kimaiyo S, Nyandiko W. Academic institutions linking access to treatment and prevention: Case Study. Geneva, Switzerland: World Health Organization; 2004.
33. Einterz RM, Kimaiyo S, Mengech HN, et al.. Responding to the HIV pandemic: the power of an academic medical partnership. Acad Med. 2007;82:812–818.
34. Voelker R. Conquering HIV and stigma in Kenya. JAMA. 2004;292:157–159.
35. Inui TS, Nyandiko WM, Kimaiyo SN, et al.. AMPATH: living proof that no one has to die from HIV. J Gen Intern Med. 2007;22:1745–1750.
36. Cohen J, Kimaiyo S, Nyandiko W, et al.. Addressing the educational void during the antiretroviral therapy rollout. AIDS. 2004;18:2105–2106.
37. Siika AM, Rotich JK, Simiyu CJ, et al.. An electronic medical record system for ambulatory care of HIV-infected patients in Kenya. Int J Med Inform. 2005;74:345–355.
38. Wools-Kaloustian K, Kimaiyo S, Diero L, et al.. Viability and effectiveness of large-scale HIV treatment initiatives in sub-Saharan Africa: experience from western Kenya. AIDS. 2006;20:41–48.
39. Vreeman RC, Wiehe SE, Ayaya SO, et al.. Association of antiretroviral and clinic adherence with orphan status among HIV-infected children in western Kenya. J Acquir Immune Defic Syndr. 2008;49:163–170.
40. Vreeman RC, Wiehe SE, Pearce EC, et al.. A systematic review of pediatric adherence to antiretroviral therapy in low- and middle-income countries. Pediatr Infect Dis J. 2008;27:686–691.
41. Brinkhof MW, Pujades-Rodriguez M, Egger M. Mortality of patients lost to follow-up in antiretroviral treatment programmes in resource-limited settings: systematic review and meta-analysis. PLoS ONE. 2009;4:e5790. doi:10.1371/journal.pone.0005790.
42. Braitstein P, Katschke A, Shen C, et al.. Retention of HIV-infected and exposed children in a comprehensive HIV clinical care program in western Kenya. Trop Med Int Health. 2010;15:833–841.
43. Mynott A. Ethnic tensions dividing Kenya. BBC News. January 5, 2008. Available at: http://news.bbc.co.uk/2/hi/programmes/from_our_own_correspondent/7172038.stm
. Accessed October 16, 2010.