Combination antiretroviral therapy (cART) effectively suppresses HIV replication, reduces mortality, and improves the lives of children and adults with HIV.1–5 It is now recognized that nonadherence to clinic appointments is an independent risk factor for virologic failure in patients receiving cART.6,7 If clinical nonadherence is coupled with poor adherence to medication adherence, viral resistance to drugs8 and opportunistic infections can develop.9 With 2.3 million children younger than 15 years currently living with HIV,10 measuring and supporting long-term pediatric adherence to care and cART is a priority.
A study of 2619 veterans in the United States starting cART after 1997 found that the frequency of visits (defined as using the number of 3 monthly intervals the patients honored an appointment with a visit during the first year of care) affected the patients’ subsequent survival. Those who made at least 1 visit in all 4 quarters had a rate of death almost half that of patients who had only a single visit during the first year.11 In resource-limited settings (RLS), emerging studies among HIV-positive adults suggest a similar relationship between retention in care and mortality. In studies in China and South Africa, a direct relationship between missed visits and subsequent mortality was noted. From data from the China National Treatment Program, a direct relationship between missed visits during the first 6 months after engagement and subsequent mortality was noted. Adjusted for clinical and demographic factors including baseline CD4 level, missing 1–2 visits conferred a 1.27-fold rise in the hazard of death [95% confidence interval (CI): 1.08 to 1.48] and missing 3–5 visits conferred a 1.72-fold rise in the hazard of death (95% CI: 1.36 to 2.18).12 In South Africa, the risk of death after 9 months was higher among patients who missed 1 [hazard ratio (HR) = 1.2; 95% CI: 1.0 to 1.6] or more than 1 (HR = 2.1; 95% CI: 1.3 to 3.3) medical visit within the first 6 months on ART compared with missing none.13
Among adult patients initiating HIV-related care, missed clinic visits and loss-to-follow-up (LTFU) during the first year after enrollment are associated with delays in initiation to cART.14
For pediatric populations within RLS, estimates of pediatric retention in care are poorly described. There is paucity of pediatric studies describing the relationship between missed visits and mortality. In a previous study from our setting in western Kenya conducted on more than 1500 children on cART between June 2004 and March 2007, over half of HIV-infected children on cART missed at least 1 scheduled monthly clinic visit during their follow-up period, with 40% missing 10% of their scheduled clinic visits.15 Children with HIV in RLS are likely to receive their cART in a very different context, one that includes much higher rate of orphanhood, malnutrition, and significant poverty.10,16–24
Even as they face unique challenges, HIV care systems in RLS are rapidly growing in capacity. As these care systems scale-up, the opportunity exist to modify them to maximize patient outcomes. The primary objective of this study was to investigate the outcomes of variable clinic attendance. Within a growing pediatric HIV care system in western Kenya that has enrolled more than 33,000 HIV-exposed children since inception, including more than 7000 HIV-infected children, we sought to describe the effect of cumulative clinic adherence (CCA) to HIV clinic appointments on mortality and LTFU from care among HIV-infected children.
PATIENTS AND METHODS
This retrospective study used prospectively collected deidentified data from the computerized medical records of HIV-infected pediatric patients treated in the United States Agency for International Development–Academic Model Providing Access to Healthcare (USAID-AMPATH) clinical care system in western Kenya. All records for patients treated in the program are entered into a centralized Ambulatory Medical Record System. The study was approved by the Institutional Research and Ethics Committee of the Moi University School of Medicine and Moi Teaching and Referral Hospital (Eldoret, Kenya) and by the Institutional Review Board of the Indiana University School of Medicine (Indianapolis, IN).
AMPATH began in 2001 as a joint partnership between Moi University School of Medicine (Eldoret, Kenya), the Indiana University School of Medicine (Indiana, United States), and the Moi Teaching and Referral Hospital (Eldoret, Kenya) to provide an HIV care system for patients in western Kenya.25–28 In 2004, USAID became a major partner and the USAID-AMPATH Partnership was inaugurated. Since 2001, the AMPATH program has enrolled more than 150,000 patients including HIV-exposed and HIV-infected children in 65 Ministry of Health rural and urban facilities in western Kenya.25 A computerized medical record system supports clinical care and research.29 All HIV care and treatment are provided free at the point of care for patients, with care provided by physicians, clinical officers, and nurses trained and mentored within AMPATH.30
Eligible patients for this analysis included any child who was enrolled in any of the AMPATH clinics between November 2001 and March 2009, confirmed to be HIV infected, younger than 14 years at enrollment, had at least 1 follow-up visit recorded in the electronic medical record database, and had a CD4 assessment done within 3 months of enrollment.
Throughout the period of the study, clinicians followed detailed locally developed protocols consistent with World Health Organization guidelines. HIV infection was documented by 1 or more DNA polymerase chain reaction tests (Amplicor, Roche, Basel, Switzerland) for children <18 months and by 2 parallel HIV rapid enzyme-linked immunosorbent assay tests using Determine and Unigold for children >18 months. Before February 2009, cART was initiated for any children <6 years of age with a CD4 cell percentage of <15%, any child >6 years of age with a CD4 count <200 cells per cubic millimeter, and for all children with WHO clinical stages 3 or 4 or CDC stages B or C. After February 2009, cART was initiated for all HIV-infected children younger than 18 months, any children <6 years of age with a CD4 cell percentage of <25%, any child >6 years of age with a CD4 percent of <20% or CD4 count <350 cells per cubic millimeter, and for all children with WHO clinical stages 3 or 4 or CDC stage C. Before February 2009, the standard initial cART regimens used were zidovudine + lamivudine + nevirapine for those weighing <10 kg or stavudine + lamivudine + nevirapine or stavudine + lamivudine + efavirenz for those weighing >10 kg. After February 2009, the standard initial cART regimens used were abacavir + lamivudine + nevirapine or abacavir + lamivudine + efavirenz. Adherence counseling and education addressing both cART and clinic attendance was provided during the clinic session in which cART was initiated. The adherence counseling is done by clinicians (physicians or clinical officers) or designated trained adherence nurses or pharmacy staff. Although there is some variability in the personnel providing the counseling, training in standardized adherence counseling and the use of standardized forms for assessment and counseling provides consistency among the network of AMPATH clinics. The average duration of waiting time when visiting the clinics was around 3 hours during this period of follow-up. (ref Koskei/Kimaiyo article) Of note is that there were no medication stock-outs during the period describing the cohort.
Children started on cART were scheduled to be seen 2 weeks after initiation of therapy, and then every month thereafter. However, the very sick patients either clinically or in those with severe ISS, weekly visits were scheduled. During these visits, patients underwent both clinical and adherence assessments through caregiver report or self-report and received a monthly supply of antiretroviral medications. Children not on cART were seen 1 month after enrollment, then 2–3 monthly thereafter unless there was a clinical reason for more frequent review or unless there was clinical or immunologic deterioration requiring initiation of cART. Patients were given an appointment card at the time of their enrollment on which the clinician wrote the dates of their next AMPATH appointments, and this was updated at each visit.
All children and the caregivers saw a peer outreach worker at each visit to maintain updated locator information, including telephone number and a map of where they lived. More than 50% of the patients had phones or a contact with a phone where they could be reached (outreach records at enrollment). When a patient missed a scheduled clinic visit, outreach workers were informed, and attempts to contact the patient (or the pediatric caregiver) through phone and/or a home visit within 24 hours of a missed visit were initiated. Every effort was made to get to the patient.
Data Collection and Measures
Clinicians completed standardized encounter forms at all AMPATH clinic visits (http://amrs.iukenya.org/download/forms). The initial encounter form included demographics, medical history, medication history, dietary intake, social history, physical exam, and laboratory data. On return visit encounter forms, the clinician collected follow-up data, including interval symptoms, medication adherence, new diagnoses, laboratory data, and modifications in drug regimens. Clinicians recorded the next scheduled visit date on the encounter form, and dedicated data entry clerks entered this information into the AMPATH Medical Record System, with data entry validated by random review of 10% of the data entered.
Adherence to routine clinic visits was evaluated using appointment data from the electronic medical record system. The dates of actual clinic visits were compared with the dates of scheduled visits to determine number of missed visits during the time period during which the child was followed up. Earlier than scheduled visits were noted and taken as visits in this analysis. For those on cART, a missed visit was defined as going for more than 7 days beyond the expected appointment date without a clinic visit. For the children not on cART, a missed visit was defined as going for more than 14 days beyond the expected appointment date without a clinic visit. Cumulative adherence was defined at each clinic visit (except baseline) as the proportion of number of days adherent to clinic visits (adherent visits being made 7 days within scheduled visit day if on cART or 14 days if not on cART), divided by the sum of the days after enrollment up to the current visit. That is, cumulative adherence = (number of days adherent to clinic visits)/(number of days in program).
Thus, the cumulative adherence index is a continuous measure ranging from 0 to 1 and indicates how well the patients adhered to scheduled clinic visits. Figure 1 illustrates the definition of cumulative adherence with a hypothetical example. For children whose encounters were missing the next scheduled visit dates, we imputed them as 28 days after the current visit if the patient was on cART or 90 days if the patient was not.
LTFU was defined as having 3 months without clinic visits if on cART or 6 months if not on cART, with no information about the vital status of the patient before database closure. If multiple episodes of LTFU occurred for any given child, the last episode was used in the analysis. If a child came back after being initially defined as LTFU, they were not considered as LTFU.
Time-stationary covariates included gender, baseline age, weight-for-age z score, weight-for-height z score, and CD4 percent nearest available to baseline within 3 months. Time-varying covariates included cART (on vs. otherwise); CDC clinical stages (N, A, B, C); change of weight-for-age z score from baseline; change of weight-for-height z score from baseline; food pickup (yes vs. no), orphan (both parents dead vs. other); clinic type (referral hospital; district, and subdistrict hospital; rural hospital), and CCA.
The primary outcome variables of interest for this study were time to death and time to LTFU. Patients’ demographic and clinical characteristics at enrollment (or data closest to enrollment within 3 months if not available at enrollment) were compared using Fisher exact tests for categorical data or Kruskal–Wallis tests for continuous data. Age-specific CD4% was used to determine severe ISS for the study population. Children younger than 18 months, 18–60 months old, or older than 60 months were defined as severely ISS if the CD4% was less than 25%, 20%, and 15%, respectively. Clinical stages were classified as CDC classes N, A, B, and C equivalent to WHO stages 1, 2, 3, and 4, respectively.
To assess the short-term and long-term associations between adherence to clinic appointments and mortality and LTFU in the study population, Cox proportional hazards models with time-varying coefficients on the CCA were adopted. Potential confounders and clinically important characteristics were adjusted for in the Cox models, which included gender, baseline age, weight-for-age z score, weight-for-height z score, and CD4% as time-fixed covariates; and cART status, CDC staging, changes in z scores from baseline, nutrition supplementation received at each visit, orphan status, and clinic type attended as time-varying covariates. The coefficients of CCA were allowed to vary over time by including an interaction between CCA and logarithm-transformed time to detect time-varying effects of clinic adherence on mortality and LTFU, and time-varying hazard ratios of CCA were reported. We fitted similar Cox models stratifying on the ISS status to evaluate the effect of clinic adherence on the primary outcomes depending on the immunologic status of patients at enrollment. Children missing CD4 data within 3 months of enrollment were excluded from the multivariable analyses. All analyses were done with R version 22.214.171.124
There were 3255 children who met the inclusion criteria. Of these, 51.2% were male, with a median age of 5.2 years (IQR: 3.5–7.3), a median duration of follow-up of 586 days (IQR: 252–985), with 37.3% being in care in rural clinics. (Table 1) Among them, 1125 (34.5%) had severe ISS at baseline, 1268 (39.0%) did not have severe ISS, and 862 (26.5%) did not have CD4% assessments within 3 months of initial visit. Children who had CD4% information within 3 months postenrollment were compared with those who did not: they were older, less likely to be on cART at baseline, and a higher proportion of them attended the central referral hospital (Table 1).
During the follow-up period, 88 children were known to have died. Those with low weight-for-age and weight-for-height z scores at baseline; low weight-for-age z score change; low baseline CD4% and CDC stage C had higher greater risk for death. Male gender was protective against death (Table 2A) On fitting Cox proportional hazards models to assess the time-varying effects of adherence to clinic appointments for the 2393 patients with their CD4% ascertained within 3 months postenrollment, we found that the association between clinic adherence and mortality significantly changed over time. Figure 2A shows the adjusted hazard ratios (AHRs) of mortality associated with a 10% increase in CCA at different time points. In the short term, better adherence to clinic appointments in HIV-infected children was related to higher risk of mortality (with AHR > 1). For instance, at 3 months postenrollment, the AHR for mortality was 2.98 [with 95% confidence interval (CI): 2.06 to 4.31] with a 10% increase in CCA; at 6 months postenrollment, the AHR dropped to 1.84 (95% CI: 1.45 to 2.33). However, a “protective” effect of adherence to clinic visits became pronounced as the patients continued to receive care in the program. The AHRs of mortality at 12, 24, and 36 months were 1.15 (95% CI: 0.99 to 1.33), 0.71 (95% CI: 0.60 to 0.85), and 0.54 (95% CI: 0.43 to 0.68), respectively. We observed similar patterns of time-varying associations between the risk of mortality and clinic adherence among both severe ISS and nonsevere ISS patients. The children with severe ISS had an earlier protective effect of adherence to clinic on death than those with nonsevere ISS (Fig. 2A)
Loss to Follow-Up
During the follow-up period, 567 children became LTFU. Those with low weight-for-height z scores at baseline, low weight-for-age z score change, and attending rural clinics had higher greater risk for LTFU, whereas receiving food was protective against LTFU (Table 2A). The association between clinic adherence and LTFU was also time varying (Fig. 2B). At 3 months postenrollment, the AHR of CCA was 1.44 (95% CI: 1.33 to 1.55) with 10% increase of CCA, which declined to 1.13 (95% CI: 1.07 to 1.19) at 6 months. After 12 months in the program, better adherence to clinic was associated with lower risk of LTFU, with AHR at 12, 24, and 36 months, at 0.89 (95% CI: 0.86 to 0.92), 0.70 (95% CI: 0.67 to 0.74), and 0.61 (95% CI: 0.57 to 0.65), respectively. When stratifying on the ISS status, similar effects of clinic adherence and other covariates were observed in both groups of patients. The children with severe ISS had an earlier protective effect of adherence to clinic on LTFU than those with nonsevere ISS as shown in Figure 2B.
In this retrospective observational study, we sought to determine the relationship between clinic adherence and LTFU and death. The findings highlight that, early on in the follow-up period, there is a higher risk for LTFU and death associated with better clinic adherence. However, these risks of LTFU or death became lower for those with better clinic adherence after 12 and 24 months of follow-up, respectively. This indicates that better clinic adherence has a protective effect over time.
We also found that it took the severely ISS group a shorter period for better CCA to be protective compared with the nonsevere group to both death and LTFU. This may be explained by the fact that the severely ISS patients have greater clinical benefit from more vigilant follow-up and the treatment that they are initiated on compared with those with better immunity who are at lesser risk for HIV-related complications.
The early higher risk associated with clinic adherence and both LTFU and death can be traced to the fact that the more sickly patients with the worst immune suppression are generally given more frequent clinic appointments (in the express care model within the AMPATH program), which families attempt to attend because their children have more severe illness.32 Those patients presenting for care late in the disease stage have a higher risk for death, irrespective of the adherence to care. In this cohort, this was confirmed by a significantly higher risk for mortality in patients with severe immune suppression and severe clinical disease compared with the nonsevere. In addition, this has also been shown in an adult cohort in the same program.33 Other studies have shown that patients who present initially with lower CD4% rarely reach normal CD4%, even after more than 12 months of cART.34 This phenomenon is explained by the fact that in severely ISS patients, it takes longer to stimulate CD4 production arising from the fact that there is significant lymphoid structure destruction and bone marrow depression that impairs production of CD4 cells. These children are therefore still at risk for opportunistic infections despite close monitoring and adherence to clinic visits. Their early clinic adherence does not, therefore, reduce their risk of death from the lowered immunity. We know that it takes more than 3–6 months to start seeing a reduction in patients’ viral loads and an even longer lag time before there are appreciable changes in CD4%.1,3 Thus, the impact of cART and continued management for opportunistic infections are likely felt after about a year on treatment and follow-up.1,3 The findings from this large treatment program in Kenya confirm that the benefits of proper care and cART are accrued later, especially for those with severe clinical and immunological disease.
The increased risk for death for those with better clinic adherence persists until 24 months of follow-up, at which point better adherence to clinic becomes protective against death. This is in contrast to LTFU where the increased risk persists for the first 12 months before better adherence to clinic becomes protective against LTFU in this cohort. Children who are LTFU may have various reasons contributing to getting lost. These include death, transfer of services to other health care provider in desperation, HIV-related stigma, religious beliefs of being healed, and stopping to seek care due to inability to afford clinic attendance.35 Healthcare providers may not be aware of these reasons for withdrawing from treatment. The AMPATH program has realized the competing issues in this severely ill population and has put in place outreach and social support programs to help new severely ill patients to both attend clinics as required and to lessen the burden of frequent visits.27,36 The use of mobile technology, including SMS, has come in handy in follow-up of patients who are LTFU, and our program and others have published findings.37 We think the continued use of SMS to remind patients of the scheduled clinic visit date and prompt clinic visit for those who miss the clinic is potentially an area of intervention that the program is starting to use and evaluate.
The study, however, raises concerns that some of the patients who are lost to follow-up early on may be dying without the clinics being aware. In a pilot study among children in the same program who were described as LTFU, 16% of the HIV-infected children and 4% of HIV-exposed children categorized as LTFU were found to actually have died.34 The sample included HIV-exposed children some of whom turned out to be uninfected. Therefore, in western Kenya, we may still have a population of children who may be LTFU but are actually deceased.
It is worth noting that in the period after 12 months of follow-up, the “high-risk” patients may already have died or become LTFU so the subset of patients who make it past 12 months are already predisposed to survive and or be retained. This represents a survivor bias and may explain the better protective effect for LTFU after 12 months and for mortality after 24 months of follow-up.
The strengths of the study include the fact that it is based on both rural and urban populations of patients and uses data that is collected in routine care and therefore captures the issues at the care level rather than pure investigational level. Its findings are therefore likely to resonate with the majority of care systems in the country and region. One limitation of this analysis is the fact that we have quite a large population of children with no known immune status at recruitment who were therefore not used in the final analysis. By including everyone and not adjusting for CD4 show similar shape of the time-varying effect of CCA on both LTFU and mortality. The time at which CCA changed from a risk factor to protective is a bit longer compared with only including the n = 2393 with CD4 as in the current analysis. However, the implication of this analysis, bearing out the benefits of good adherence to clinic appointments in the long run, outweighs significant programmatic limitations. We hope and expect that with a larger cohort, the benefits would be more apparent.
CONCLUSIONS AND RECOMMENDATIONS
Children adherence to clinic visits during the first 6 months of HIV care was strongly associated with a higher risk of death and LTFU, but by 12 months, children with better clinic adherence had a reduced risk of LTFU and by 24 months had reduced risk of mortality.
These findings highlight the need to be more vigilant especially with severely immune-suppressed children initiating care and increase the social and medical armament that includes aggressive structured diagnosis and or blanket treatment of the HIV-related conditions known to cause highest mortality in our population. There is also need to ascertain and document the vital status of all patients reported to be LTFU by doing aggressive outreach.
The authors give special thanks to the families and to the health care providers of AMPATH and Moi Teaching and referral hospital, including the nurses, clinicians, nutritionists, social workers, outreach workers, pharmacy staff, and records and data assistants all of who work tirelessly to ensure that the children of Western Kenya receive the medical care they deserve. In particular, the authors would like to thank Dr Tenge, Dr Nabakwe, Dr Marete, Dr Apondi, Dr Gisore, Dr Songok, Dr Chumba, Dr Jakait, R. Too, J. Yaran, V. Cheboi, A. Koech, J. Chemwon, J. Aluoch, IreneTigoi, Mabonga, Lillian Boit, J. Sawe, M. Rugut, N. Warui, D. Wabuti, Nyambane, and the other current and past members of the AMPATH Pediatric Working Group. The authors also wish to thank the Institutional Review and Ethics Committee, the Director of MTRH, the Principal, College of Health Sciences, and the Dean, Moi University School of Medicine, for allowing us to collect data and conduct research on the patients they manage.
1. Doerholt K, Duong T, Tookey P, et al.. Outcomes for human immunodeficiency virus-1-infected infants in the United kingdom and Republic of Ireland in the era of effective antiretroviral therapy. Pediatr Infect Dis J. 2006;25:420–426.
2. Gibb DM, Goodall RL, Giacomet V, et al.. Adherence to prescribed antiretroviral therapy in human immunodeficiency virus-infected children
in the PENTA 5 trial. Pediatr Infect Dis J. 2003;22:56–62.
3. Hogg RS, Heath KV, Yip B, et al.. Improved survival among HIV-infected individuals following initiation of antiretroviral therapy. JAMA. 1998;279:450–454.
4. Needham DM, Hogg RS, Yip B, et al.. The impact of antiretroviral therapy on AIDS survival observed in a province-wide drug treatment programme. Int J STD AIDS. 1998;9:370–372.
5. World Health Organization. Progress on Global Access to HIV Antiretroviral Therapy: A Report on “3 by 5” and Beyond. Geneva, Switzerland: World Health Organization; 2006.
6. Lucas GM, Chaisson RE, Moore RD. Highly active antiretroviral therapy in a large urban clinic: risk factors for virologic failure and adverse drug reactions. Ann Intern Med. 1999;131:81–87.
7. Rastegar DA, Fingerhood MI, Jasinski DR. Highly active antiretroviral therapy outcomes in a primary care clinic. AIDS Care. 2003;15:231–237.
8. Mullen J, Leech S, O'Shea S, et al.. Antiretroviral drug resistance among HIV-1 infected children
failing treatment. J Med Virol. 2002;68:299–304.
9. San-Andres FJ, Rubio R, Castilla J, et al.. Incidence of acquired immunodeficiency syndrome-associated opportunistic diseases and the effect of treatment on a cohort of 1115 patients infected with human immunodeficiency virus, 1989–1997. Clin Infect Dis. 2003;36:1177–1185.
10. , et al.. 2006 Report on the Global AIDS Epidemic. Geneva, Switzerland: UNAIDS; 2006.
11. Giordano TP, Gifford AL, White AC Jr, et al.. Retention in care: a challenge to survival with HIV infection. Clin Infect Dis. 2007;44:1493–1499.
12. Zhang Y, Dou Z, Sun K, et al.. Association between missed early visits and survival among patients of China national free ART cohort. J Acquir Immune Defic Syndr. 2012;60:59–67.
13. Brennan A, Maskew M, Sanne I, et al.. Importance of clinic attendance in the first 6 months on ART: missing medical visits increases mortality. J Int AIDS Soc. 2010;13:49.
14. Lee VJ, Earnest A, Chen MI, et al.. Predictors of failed attendances in a multi-specialty outpatient centre using electronic databases. BMC Health Serv Res. 2005;5:51.
15. 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.
16. UNICEF. The State of the World's Children
2007. New York, NY: UNICEF; 2007.
17. Watts H, Gregson S, Saito S, et al.. Poorer health and nutritional outcomes in orphans and vulnerable young children
not explained by greater exposure to extreme poverty in Zimbabwe. Trop Med Int Health. 2007;12:584–593.
18. Lindblade KA, Odhiambo F, Rosen DH, et al.. Health and nutritional status of orphans <6 years old cared for by relatives in western Kenya. Trop Med Int Health. 2003;8:67–72.
19. Ayaya SO, Esamai FO, Rotich J, et al.. Socio-economic factors predisposing under five-year-old children
to severe protein energy malnutrition at the Moi Teaching and Referral Hospital, Eldoret, Kenya. East Afr Med J. 2004;81:415–421.
20. Sarker M, Neckermann C, Muller O, Assessing the health status of young AIDS and other orphans in Kampala, Uganda. Trop Med Int Health. 2005;10:210–215.
21. Monasch R, Boerma JT. Orphanhood and childcare patterns in sub-Saharan Africa: an analysis of national surveys from 40 countries. AIDS. 2004;18(suppl 2):S55–S65.
22. Bicego G, Rutstein S, Johnson K. Dimensions of the emerging orphan crisis in sub-Saharan Africa. Soc Sci Med. 2003;56:1235–1247.
23. Crampin AC, Floyd S, Glynn JR, et al.. The long-term impact of HIV and orphanhood on the mortality and physical well-being of children
in rural Malawi. AIDS. 2003;17:389–397.
24. Ryder RW, Kamenga M, Nkusu M, et al.. AIDS orphans in Kinshasa, Zaire: incidence and socioeconomic consequences. AIDS. 1994;8:673–679.
25. Einterz RM, Kelley CR, Mamlin JJ, et al.. Partnerships in international health. The Indiana University-Moi University experience. Infect Dis Clin North Am. 1995;9:453–455.
26. Voelker R. Conquering HIV and stigma in Kenya. JAMA. 2004;292:157–159.
27. Mamlin J, Kimaiyo S, Nyandiko W, et al.. Academic Institutions Linking Access to Treatment and Prevention: Case Study. Geneva, Switzerland: World Health Organization; 2004.
28. 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.
29. 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.
30. Cohen J, Kimaiyo S, Nyandiko W, et al.. Addressing the educational void during the antiretroviral therapy rollout. AIDS. 2004;18:2105–2106.
31. R Development Core Team. A Language and Environment for Statistical Computing. Vienna, Austria: The R Foundation for Statistical Computing; 2011.
32. Braitstein P, Siika A, Hogan J, et al.. A clinician-nurse model to reduce early mortality and increase clinic retention among high-risk HIV-infected patients initiating combination antiretroviral treatment. J Int AIDS Soc. 2012;15:7.
33. Siika AM, Ayuo PO, Sidle MJ, et al.. Admission characteristics, diagnoses and outcomes of HIV-infected patients registered in an ambulatory HIV-care programme in western Kenya. East Afr Med J. 2008;85:523–528.
34. Patel K, Hernán MA, Williams PL, et al.. Long-term effectiveness of highly active antiretroviral therapy on CD4+ cell evolution among children
and adolescents infected with HIV: 5 years and counting. Clin Infect Dis. 2008;46:1751–1760.
35. 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–e60.
36. 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.
37. Pop-Eleches C, Thirumurthy H, Habyarimana JP, et al.. Mobile phone technologies improve adherence to antiretroviral treatment in a resource-limited setting: a randomized controlled trial of text message reminders. AIDS. 2011;25:825–834.