Inadequate Monitoring in Advanced Stages of Disease with Lack of Supportive Counseling Increases Attrition among Patients on Antiretroviral Treatment at a Large Urban Clinic in Uganda

Abstract

Background:

The purpose of this case–control study was to identify risk factors for loss to follow-up (LTFU).

Methods:

Cases and controls were selected from HIV-positive patients, aged 18 years and older, on antiretroviral therapy (ART) at the Infectious Diseases Clinic (IDC) in January 2008. As cases, we selected 209 patients who in 2008 did not return to the clinic within 90 days of their scheduled appointment date. As controls, we randomly selected 626 patients from the 5872 patients who were following up at the end of December 2008.

Results:

In multivariable logistic regression analysis, urban or semiurban residence, World Health Organization disease stage III or IV at ART initiation, a median CD4 count at last visit <200 cells/mm³, tuberculosis (TB) in the 6 months before the last visit, absence of counseling before ART initiation, and no disclosure of HIV status were associated with LTFU.

Conclusion:

This study demonstrates the importance of close patient monitoring in advanced stages of disease, supportive counseling for patients initiating ART, extra psychosocial support for patients with TB and HIV coinfection, assisting patients with disclosure, and setting up a good referral system to retain patients on ART.

Keywords

counseling inadequate monitoring HIV-positive patients advanced disease Uganda patient attrition

Introduction

Although the success of antiretroviral therapy (ART) scale-up has been widely acknowledged in resource-constrained settings, poor retention in care is a major challenge faced by ART programs.^1

–4 Most patient attrition occurs within the first year on ART, and patient retention across low- and middle-income countries in 2009 was estimated at 82% after 12 months on ART.¹

A meta-analysis in 2008 showed a retention rate of 86% at 6 months and 76% at the end of 2 years.⁵ Brinkhof et al reported that over 40% of HIV-positive African adults who did not return for their clinic appointment were dead.⁶ This high risk of death is associated with late initiation of ART in patients with advanced disease.^7,8

Long distance to the clinic is one of the major causes for loss to follow-up (LTFU).^9,10 Some patients may self-transfer to clinics closer to their homes or workplaces, while others may move to a different location in search of jobs and business opportunities.¹⁰ Others may become LTFU because of family relocation, marriage, or war. Other factors influencing LTFU include stigma, low education level, religious beliefs, poor quality of services,^9

–12 formal and informal costs, poverty, adverse effects of drugs, nondisclosure, pregnancy, long waiting times, alcohol abuse, and use of traditional medicines.^13
–15

Causes of attrition may vary in different clinics and among different patient populations.⁴ The purpose of our study was to identify factors associated with increased LTFU at the Infectious Diseases Clinic (IDC), a large HIV clinic in Kampala, Uganda.

Methods

Study Setting

The IDC is a clinic owned by the Infectious Disease Institute (IDI), Makerere University College of Health Sciences. The IDC is located at the Mulago hospital campus that is the main national referral hospital in Uganda. Over 400 patients receive care at the IDC daily.¹⁶ At the IDC, task shifting is implemented in which stable patients on ART are seen by nurses for their routine appointments or managed through pharmacy-only refill program (PRP). These patients are reviewed by doctors once every 6 months.¹⁷

At the beginning of 2007, the IDC initiated a patient tracking system that was using patients’ or their relatives’ phone numbers or home visits if the phone number was unavailable to track the HIV-positive patients on ART. However, more than 30% of the targeted patients were untraceable. Reasons included wrong and missing telephone contacts and home addresses.¹⁸

In the last 3 months of 2007, additional information was noted in the patients’ medical records including the patient’s physical address, their consent to a home visit if they could not be reached by phone, the ownership of the phone contact, the patient’s disclosure status if the phone contact belonged to someone else, and their consent to phone tracking. Using the updated patient information, from January to December 2008, 2 home visitors tracked HIV-positive patients on ART after the patients missed their appointment date by 90 days. Patients were classified as self-transferred, dead, stopped care, and untraceable based on the outcome of the tracking intervention.

Study Design

We conducted a case–control study, whereby cases and controls were selected from HIV-positive patients, aged 18 years and older, on ART at the IDC in January 2008. Cases were all patients who in 2008 did not return to the clinic within 90 days of their scheduled appointment date. Controls were randomly selected from the patients who were following up until December 31, 2008.

Sampling Procedure

A case–control design selects study subjects on the basis of an outcome, in this case LTFU. The analysis therefore will show the prevalence of factors differing between the outcome groupings, which may be predictive of the outcome or be potentially useful as targets for intervention to decrease LTFU.

A total of 346 cases were identified from the IDC database, and 137 were excluded from the study for the following reasons: 43 were officially transferred out, 18 had missing return appointment dates, 20 had died, 49 had participated in other research studies that had a tracking component, and 7 had been found to have resumed care before 90 days after further investigation. Therefore, 209 cases were included in the study (Figure 1).

Figure 1.

Flowchart showing the selection of patients lost to follow-up (cases) from the patients on antiretroviral therapy in 2008 at the Infectious Diseases Clinic (IDC).

Three controls for each case were selected at random from a list of all patients who were still in follow-up. Using the information from the IDC database, 5872 active patients were identified, of which 1127 patients were excluded, as they were participating in research studies with a tracking component. The identification numbers of 4745 patients were arranged in ascending order. Using a sampling interval of seven, 677 controls were sampled, and only 626 were included in the study since 31 had formally been transferred to other centers and 10 were excluded as their files were missing and could not be found during the 12 months of the study (Figure 2). We never matched on any factor, as we wanted to explore all possible relationships with LTFU.

Figure 2.

Flowchart showing the selection of the controls.

Data Collection

Data were abstracted from 3 sources: the IDC database, medical notes, and home visitor reports. A team of trained research assistants working with nurses, counselors, and medical officers verified the data with the patient’s medical notes. Patients who reported living in Kampala or in a large neighboring town were categorized as “urban or semiurban” patients and the rest as “rural” patients.

Information on patient outcome, including dead, transferred, or untraceable, was obtained from home visitors’ reports. Data were abstracted using data abstraction forms. Forms were checked for completeness and correctness. Data were double entered into EPIDATA statistical software (Version 3.1, EpiData Association, Odense, Denmark), cleaned, edited, coded, and exported to STATA statistical software (Version 11, Stata Corp, College Station, Texas) for analysis. Information was analyzed anonymously.

Statistical Analysis

We analyzed the differences between patients who were still in follow-up and those LTFU. Analysis of variance (ANOVA) test was used to compare mean ages, Kruskal Wallis test for median CD4 counts, and Pearson chi-square test or Fisher exact test for cell frequencies <5 were used for proportions. We used 2-tailed tests and an α-level of .05 for all our analyses. Risk factors were first explored individually using bivariate logistic regression analysis in order to select the variables for the model. Then in a multivariable model, hypothesized risk factors and confounders were included based on bivariate P value ≤.2.

We further explored the characteristics of the LTFU patients based on the tracking outcome, including self-transferred, dead, and untraceable. We analyzed the differences between patients who were self-transferred, dead, and untraceable. Bivariate comparisons were done using the following tests: ANOVA test to comapre mean ages and duration on ART, Pearson chi-square test or Fisher exact test for proportions, and Kruskal-Wallis test for median CD4 count, respectively. All statistical analyses were performed with STATA software, version 10.0, College Station, Texas.

Ethics

The study was approved by the IDI Scientific Review Committee and the institutional review boards for School of Medicine, Makerere University College of Health Sciences and the Uganda National Council for Science and Technology.

Results

In bivariate analysis, factors associated with LTFU included female sex, older age >30 years, World Health Organization (WHO) disease stage III or IV at ART initiation, urban or semiurban residence, a median CD4 count of <200 cells/mm³ at last visit, tuberculosis (TB) in 6 months before the last visit, ART regimen at last visit, absence of counseling, and nondisclosure of HIV status (see Table 1).

Table 1.

Characteristics of Patients Lost to Follow-Up (LTFU) and Patients Still in Follow-Up: Bivariate Analysis.

Variable	All Patients (835, 100%)	LTFU (n₁ = 209)	Not LTFU (n₂ = 626)	P Value
Sex: female	546 (65.3)	124 (59.3)	422 (67.4)	.033
Age at last visit, y, mean (CI 95%)	38 (38-39)	37 (35-38)	39 (38-39)	.012
Marital status: married	394 (47.1)	96 (45.9)	298 (47.5)	.689
Employment: employed	464 (55.5)	105 (50.2)	359 (57.2)	.077
Residence: urban or semiurban	554 (66.3)	160 (76.5)	394 (62.9)	.000
Distance: <20-km radius	603 (72.1)	148 (70.8)	455 (72.5)	.624
WHO III or IV	516 (63.6)	152 (73.1)	364 (60.3)	.001
ART experienced at IDI clinic registration	105 (12.5)	18 (8.6)	87 (13.9)	.045
CD4 at ART initiation, median (IQR)	125 (41-193)	119 (37-194)	128 (44-193)	.446
CD4 at last visit, median (IQR)	278 (176-407)	183 (72-278)	306 (206-427)	.000
CD4 at last visit categorized, <200 cells	257 (31.1)	145 (23.3)	112 (54.6)	.000
TB in 6 months before the last visit	53 (6.3)	28 (13.4)	25 (4.0)	.000
First regimen received at clinic: D4T30_3TC_NVP/EFV	421 (50.4)	86 (41.1)	335 (53.6)
ZDV_3TC_NVP/EFV	395 (47.3)	119 (56.9)	276 (44.1)	.003
Regimen containing LPV/R	18 (2.1)	4 (1.9)	14 (2.2)
Regimen received at last visit: D4T30_3TC_NVP/EFV	91 (10.9)	65 (31.1)	26 (4.1)
ZDV_3TC_NVP/EFV	694 (83.2)	138 (66.0)	556 (88.9)	.000
Regimen containing LPV/R	49 (5.8)	6 (2.8)	43 (6.8)
Health worker seen at last visit: doctor	795 (96.0)	207 (99.0)	588 (94.9)	.004
No counseling before ART initiation	168 (20.1)	62 (29.6)	106 (16.9)	.000
Never disclosed HIV status	419 (50.1)	203 (97.1)	216 (34.4)	.000
No treatment supporter at ART initiation	340 (40.6)	71 (33.9)	269 (42.9)	.023

Abbreviations: IQR, interquartile range; TB, tuberculosis; d4T, stavudine; ZDV, zidovudine; 3TC, lamivudine; NVP, nevirapine; EFV, efavirenz; LPV/r, lopinavir/ritonavir; CI, confidence interval; WHO, World Health Organization; ART, antiretroviral therapy; IDI, Infectious Disease Institute.

The variables employment, ART status at registration, ART regimen at ART initiation, and presence of treatment supporter had P values which were less than .2 at bivariate analysis. However, they were not included in the multivariable model.

Employment at bivariate analysis had a P value of .078; it is possible that some of the study patients involved in agricultural farming were misclassified as not being formally employed. This is because there was no additional information to separate the subsistence farmers from the commercial farmers. The ART experience at the IDC registration had a low significance at bivariate analysis (.053); the majority (87.5%) of the patients were ART naive when they registered at the IDC, and this variable was thought to have limited influence at multivariable analysis. At bivariate analysis, the variable ARV regimen at ART initiation had a significant P value. We, however, never included it in the model. In 2007, a year before the study period, following a change in Uganda ministry of health guidelines, there was a massive switch of 321 (75.5%) patients from stavudine (d4T)-based regimen to zidovudine (ZDV)-based regimen. Inclusion of this variable therefore might result in confusing results. The variable presence of treatment supporter is necessary in assessing adherence. In this study, the LTFU patients were tracked, and the home visitors used their findings to update the patients’ file. During our data abstraction, information about treatment supporters was more complete for cases than for controls. Measurement of this variable was unreliable in this study, as there was disagreement between the abstracted data and the self-reported disclosure by the patients. Including treatment supporter as a variable in multivariable model could lead to systematic bias.

In multivariable analysis, urban or semiurban residence, WHO disease stage III or IV at ART initiation, a median CD4 count of <200 cells/mm³ at last visit, ARV regimen at last visit, TB in 6 months before the last visit, and absence of counseling and nondisclosure of HIV status were associated with LTFU (see Table 2).

Table 2.

Risk Factors for Loss to Follow-Up, Multivariable Analysis.

Variable	Crude Odds Ratio (95%)	P Value for Crude Odds Ratio	Adjusted Odds Ratio (95%)	P Value for Adjusted Odds Ratio
Sex: female	0.070 (0.51-0.97)	.033	0.67 (0.41-1.12)	.130
Age at last visit: <30 y	Reference		Reference
30 y and above	1.72 (1.15-2.58)	.008	1.56 (0.82-3.00)	.175
Employment: not employed	1.32 (0.97- 1.82)	.078^a
Residence: urban or semiurban	2.29 (1.59-3.31)	.000	3.17 (1.75-5.73)	.000
ART experienced at IDI clinic registration	0.59 (0.34-1.01)	.053^a
WHO III and IV	1.74 (1.23-2.47)	.002	1.58 (1.36-3.07)	.007
CD4 at last visit: ≥200 cells/mm³	Reference
CD4 at last visit: <200 cells/mm³	4.07 (2.92-5.67)	.000	4.71 (2.85-7.79)	.000
Regimen received at last visit: D4T30_3TC_NVP/EFV	Reference		Reference
ZDV_3TC_NVP/EFV	0.10 (0.06-0.17)	.000	0.06 (0.03-0.14)	.000
Regimen containing LPV/R	0.06 (0.02-0.16)	.000	0.2 (0.01-0.11)	.000
TB in 6 months before the last visit	3.72 (2.12-6.55)	.000	2.33 (1.08-5.57)	.048
Health worker seen at last visit: doctor	0.20 (0.05-0.83)	.027^a
No counseling before ART initiation	2.05 (1.42-2.94)	.000	7.06 (4.19-12.83)	.000
Never disclosed HIV status	3.12 (1.85-5.18)	.000	4.29 (1.82-9.36)	.000
No treatment supporter at ART initiation	0.66 (0.48-0.93)	.015^a

Abbreviations: d4T, stavudine; ZDV, zidovudine; 3TC, lamivudine; NVP, nevirapine; EFV, efavirenz; LPV/r, lopinavir/ritonavir; WHO, World Health Organization; ART, antiretroviral therapy; IDI, Infectious Disease Institute.

^a These variables were included for the purposes of discussion but were not included in the model.

Patients’ characteristics based on tracking outcome are shown in Table 3. The findings indicated that none of the patients found alive reported to have stopped medication. In all, 72 (34.4%) patients had self-transferred to other treatment centers, 76 (36.3%) were dead, and 61 (29.1%) were untraceable. Among those who were dead, 58 (76.3%) were in WHO disease stage III or IV at ART initiation. The majority (160, 76.5%) of the patients that were LTFU were residing in an urban or semiurban area. Only 61 (29.2%) LTFU patients had to travel a distance of more than 20 km to access care from the IDC. All 76 patients who were reported dead were seen by a medical doctor during their last visit.

Table 3.

Characteristics of Patients Lost to Follow-Up (LTFU) Based on the Tracking Outcome.

Variable	Total (209, 100%)	Self-Transferred (72, 34.4%)	Dead (76, 36.3%)	Untraceable (61, 29.1%)	P Value
Sex: female	124 (59.3)	44 (61.1)	42 (55.2)	38 (62.3)	.658^a
Age, y, at last visit, mean (95% CI)	37 (36-39)	36 (35-38)	37 (34-39)	37 (34-40)	.943^b
Marital status: married	96 (45.9)	37 (51.4)	36 (47.3)	23 (37.7)	.274^a
Employment: employed	105 (50.2)	39 (54.1)	36 (47.3)	30 (49.1)	.697^a
Residence: urban or semiurban	160 (76.5)	58 (80.5)	55 (72.4)	47 (77.0)	.460^a
Distance from IDC: <20 km	148 (70.8)	48 (66.7)	56 (73.6)	44 (72.1)	.621^a
ART status at IDC registration: ART experienced	18 (8.61)	8 (11.1)	7 (9.2)	3 (4.9)	.455^c
WHO III or IV at ART initiation	152 (73.1)	50 (70.4)	58 (76.3)	44 (72.1)	.709^a
CD4 at ART initiation, median (IQR)	119 (37-194)	138 (61-208)	104 (10-196)	110 (54-173)	.085^d
CD4 at last visit, median (IQR)	183 (72-278)	226 (118-335)	122 (18-260)	168 (88-267)	.013^d
CD4 at last visit: <200 cells	112 (54.6)	31 (44.2)	45 (60.8)	36 (59.0)	.098^a
On ART for <6 months	108 (51.6)	28 (38.8)	49 (64.5)	31 (50.8)	.008^a
Duration on ART in months, mean (95% CI)	9 (8-11)	13 (10-17)	5 (4-7)	10 (7-13)	.001^b
ART regimen first received at IDC
D4T30_3TC_NVP/EFV	86 (41.1)	25 (34.7)	37 (48.6)	24 (39.3)
ZDV3TCNVP/EFV¹²	119 (56.9)	44 (61.1)	39 (51.3)	36 (59.2)	.205^c
Regimen containing LPV/R¹²	4 (1.9)	3 (4.17)	0	1 (1.6)
ART regimen at last visit IDC
D4T30_3TC_NVP/EFV	65 (31.1)	16 (22.2)	33 (43.4)	16 (26.2)
ZDV_3TC_NVP/EFV¹²	138 (66.3)	52 (72.2)	42 (55.2)	44 (72.1)	.028^c
Regimen containing LPV/R¹²	6 (2.8)	4 (5.5)	1 (1.3)	1 (1.6)
TB in 6 months before last visit	28 (13.4)	8 (11.1)	10 (13.1)	10 (16.6)	.645^a
Health work last seen: doctor	207 (99.0)	70 (97.2)	76 (100)	61 (100)	.202^c

Abbreviations: ANOVA, analysis of variance; IQR, interquartile range; TB, tuberculosis; d4T, stavudine; ZDV, zidovudine; 3TC, lamivudine; NVP, nevirapine; EFV, efavirenz; LPV/r, lopinavir/ritonavir; CI, confidence interval; WHO, World Health Organization; ART, antiretroviral therapy; IDC, Infectious Diseases Clinic.

^a Pearson chi-square test.

^b ANOVA test.

^c Fisher exact test.

^d Kruskal-Wallis test.

In bivariate analysis, factors that were different among the categories of LTFU were median CD4 counts at last visit, duration on ART of <6 months, mean duration on ART, and ARV regimen at last visit.

At the end of the tracking exercise, 29.1% of the LTFU patients were untraceable, implying that it was impossible to know whether such patients were dead or had self-transferred to other ART centers. The untraceable were similar to the dead in terms of sex, age, marital status, CD4 count at ART initiation, and CD4 count at last visit. However, the untraceable had been on ART longer than the dead (median duration of 8 and 5 months, respectively; P = .032). The untraceable were similar to the self-transferred in terms of sex, age, marital status, and CD4 count at ART initiation. However, at last visit the untraceable had a lower median CD4 count than that of the self-transferred (median 177 and 237 cells/mm³, respectively; P = .048); their duration on ART was shorter (median 8 and 15 months, respectively; P = .017).

Discussion

In our study, most factors that were significant at bivariate analysis remained significant at multivariable analysis, implying that LTFU patients were truly different from patients who were still following up. Our study confirms the results of other studies that patients in more advanced stages of the disease are at greater risk for LTFU.^{12

–15,18

–23} Patients whose CD4 count did not increase to more than 200 cells/mm³ during ART were more likely to be LTFU. Such patients remain at higher risk of developing life-threatening opportunistic infections/malignancies and therefore need to be closely monitored.²⁴

Many patients became LTFU shortly after initiating ART. It is possible that the majority of those patients were referrals from Mulago hospital. Indeed, many patients admitted at Mulago hospital become aware of their HIV status through the hospital routine HIV counseling and testing program. On discharge, patients are referred to the IDC to start HIV care without being informed about other care options. Such patients may drop out later because they prefer to receive care in a health facility closer to their home or workplace.¹⁰

Patients who had TB within 6 months before their last visit were 2 times more likely to become LTFU than those without TB. Patients with TB were less likely to have disclosed their HIV status (data not shown). The IDC operates a specialized separate clinic for HIV-positive patients coinfected with TB. This separation of patients with TB from the other patients could stigmatize them leading to dropping out of care.²⁵ These results indicate that care programs should develop strategies that provide extra psychosocial support to all HIV-positive patients coinfected with TB.

The proportion of the dead in our study was small compared to those reported by other studies; this could be a result of contacting patients after a shorter time of missing their clinic appointments of 90 days when compared to 6 months, allowing for timely interventions for those who may be ill.^6
–8

It is possible that some of the patients who died within 6 months of ART initiation could have developed an immune reconstitution inflammatory syndrome (IRIS) since many initiated ART at a very low CD4 count.²⁶ Unfortunately, information about IRIS was not available in the patient records. The IRIS however is a rare cause of mortality.²⁷

All the patients who were dead had been seen by a doctor during their last visit. The doctors may have missed signs of serious disease conditions during their last visit.²⁶ It is important that high-risk patients such as those on ART with a CD4 count of <200 cells/mm³ within 6 months of treatment should be screened for opportunistic infections and malignancies. Moreover, very sick patients should be given shorter return dates and those who miss their visits should be tracked within a week.

Following the results of this study, the IDC has instituted measures to reduce LTFU as well as mortality among patients LTFU. Patients are tracked 1 week after missing a scheduled appointment if they have a CD4 count of <200 cells/mm³, are on ART for <6 months, and have been diagnosed or are being investigated for TB.

Patients that self-transferred were also very sick at ART initiation, like the dead and untraceable. The former had a median CD4 count of <200 cells/mm³ at ART initiation, but by their last visit, their median CD4 count was >200 cells/mm³, unlike the dead or untraceable. The majority of the patients that self-transferred had been on ART for more than 12 months, unlike the dead with a mean duration on ART of <6 months and untraceable with a mean duration on ART of 10 months. This could imply that the patients that self-transferred to other treatment centers were generally stable and feeling well by their last visit. The ART programs with high case loads should devise counseling and sensitization messages targeting patients who have improved after 1 year on ART to be transferred to treatment centers of their convenience for continuity of care. This has the benefit of decongesting the large ART centers that can then concentrate on the provision of quality care for the very sick.¹⁰

The untraceable patients from this study seemed to have been a heterogeneous group with some patients dead and others possibly self-transferred. Indeed, their median CD4 count and duration on ART were in between those of the dead and those who self-transferred.

Urban and semiurban residence was associated with more LTFU. It is possible that some of the very sick patients, who reported to reside in an urban or semiurban area at the start of ART, were staying with a friend or a relative in order to access good medical care. Such patients may have returned to their official residences and self-transferred once their health condition improved during ART. On the other hand, it is possible that patients, who reported residing in a rural area when initiating ART, were highly motivated to be treated at the IDC and were determined to continue receiving care from the same facility. Another reason could be that people in urban or semiurban areas are more mobile because they are in search for work, business, or accommodation. Some also may self-transfer, because of stigma, to a health care institution outside the area where they live.¹⁰

Our study illustrates the importance of close monitoring of patients in advanced stages of the disease as well as patient counseling before ART initiation to reduce attrition from care. Patients who have just initiated ART need close monitoring to reduce incidences of mortality among LTFU. Stable patients who have been on ART for more than 1 year need to be targeted for referral to a center of their convenience for routine monitoring as these might self-transfer if such a system is not available at the facility. The large proportion of the patients could not be traced due to missing information of treatment supporter or next of kin. Patients who had disclosed their HIV serostatus were less likely to be LTFU. To have a treatment supporter, patients need to disclose their HIV status to someone else. Such patients may get better socioeconomic support. Benefits may include having someone to send to the clinic for pick up of ARV drugs in case of sickness or a person to be contacted during tracking.

A limitation of our study is that we used information that was routinely collected for clinical care and not for research purposes; therefore, it may not have been very complete. We never controlled for factors like pregnancy status and religion, which could influence LTFU because these data were not available. The proportion of untraceable patients was high because of wrong addresses, incomplete data about a treatment supporter or next of kin, and a limited number of home visitors who had no time to visit patients’ homes several times.

Conclusion

This study demonstrates the importance of close monitoring of patients in advanced stages of the disease, supportive counseling for patients as they initiate ART, extra psychosocial support for HIV-positive patients coinfected with TB, assisting patients with disclosure, and setting up a good referral system to retain patients on ART.

Footnotes

Acknowledgments

We thank all the patients and staff for all their contributions toward the data used for this evaluation. We thank the Belgian Technical Cooperation (BTC), Gilead, University of California, San Francisco, and Infectious Diseases Institute for financial support. Special thanks to Agnes Kiragga, Joseph Sempa, and Philippa Easterbrook for their contribution in the streamlining of the tracking process at the IDC.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the authorship and/or publication of this article. The Belgian Technical Cooperation (BTC), Gilead, University of California, San Francisco, and Infectious Diseases Institute provided financial support.

References

WHO, UNAIDS, UNICEF 2010. Towards universal access: scaling up priority HIV/AIDS interventions in the health sector. http://www.who.int/hiv/pub/2010progressreport/en/. Accessed March 25, 2013.

Chi

Yiannoutsos

Westfall

. Universal definition of loss to follow-up in HIV treatment programs: a statistical analysis of 111 facilities in Africa, Asia, and Latin America. PloSMed. 2011;8(10):e1001111.

Hogg

Heath

Bangsberg

. Intermittent use of triple-combination therapy is predictive of mortality at baseline and after 1 year of follow-up. AIDS. 2002;16(7):1051–1058.

Rosen

Fox

Gill

. Patient retention in antiretroviral therapy programs in sub-Saharan Africa: a systematic review. PLoS Med. 2007;4(10):e298.

Brinkhof

Dabis

Myer

. Early loss of HIV-1-infected patients on potent antiretroviral therapy programmes in lower-income countries. Bull World Health Organ. 2008;86(7):559–567.

Brinkhof

MWG

Pujades-Rodriguez

Egger

. Mortality of patients lost to follow-up in antiretroviral treatment programmes in resource- limited settings: systematic review and meta-analysis. PLoS ONE. 2009;4(6):e5790.

Lawn

Myer

Harling

Orrell

Bekker

Wood

. Determinants of mortality and non-death losses from an antiretroviral treatment service in South Africa: implications for program evaluation. Clin Infect Dis. 2006;43(6):770–776.

Stringer

Zulu

Levy

. Rapid scale-up of antiretroviral therapy at primary care sites in Zambia: feasibility and early outcomes. JAMA. 2006;296(7):782–793.

Wallman

Baker

. Which resources pay for treatment? a model for estimating the informal economy of health. Soc Sci Med. 1996;42(5):671–679.

10.

Alamo

Colebunders

Ouma

. Return to normal life after AIDS as a reason for lost to follow up in a community-based antiretroviral treatment program. J Acquir Immune Defic Syndr.2012;60(2):e36–e45.

11.

Coetzee

Hildebrand

Boulle

. Outcomes after two years of providing antiretroviral treatment in Khayelitsha, South Africa. AIDS. 2004;18(6):887–895.

12.

Amanyire

Wanyenze

Alamo

. Client and provider perspectives of the efficiency and quality of care in the context of rapid scale-up of antiretroviral therapy. AIDS Patient Care STDS. 2010;24(11):719–727.

13.

Dahab

Charalambous

Hamilton

. That is why I stopped the ART0: patients’ & providers’ perspectives on barriers to and enablers of HIV treatment adherence in a South African workplace programme. BMC Public Health. 2008;8:63.

14.

Hardon

Akurut

Comoro

. Hunger, waiting time and transport costs: time to confront challenges to ART adherence in Africa. AIDS Care. 2007;19(5):658–665.

15.

Unge

Johansson

Zachariah

Some

Van Engelgem

Ekstrom

. Reasons for unsatisfactory acceptance of antiretroviral treatment in the urban Kibera slum, Kenya. AIDS Care. 2008;20(2):146–149.

16.

Kamya

Mayanja-Kizza

Kambugu

. Predictors of long-term viral failure among Ugandan children and adults treated with antiretroviral therapy. J Acquir Immune Defic Syndr. 2007;46(2):187–193.

17.

Infectious Diseases Institute College of Health Sciences Makerere University Annual report 2010. http://www.idi-makerere.com/docs/report010Cover.pdf7. Accessed April 6, 2013.

18.

Muwanga

Easterbrook

Schaefer

. Losses to Follow-up in a Large ART Program in

Uganda. Abstract number 840, the 15th Conference on Retroviruses and Opportunistic Infections 3 rd – 6th

February 2008, Boston, USA.

19.

Dalal

MacPhail

Mqhayi

. Characteristics and outcomes of adult patients lost to follow-up at an antiretroviral treatment clinic in Johannesburg, South Africa. J Acquir Immune Defic Syndr. 2008;47(1):101–107.

20.

Braitstein

Brinkhof

Dabis

. Mortality of HIV-1-infected patients in the first year of antiretroviral therapy: comparison between low income and high-income countries. Lancet. 2006;367(9513):817–824.

21.

Brown

Thorne

Foster

. Factors affecting attrition in a longitudinal study of patients with AIDS. AIDS Care. 2006;18(7):821–829.

22.

Dudley

Jin

Hoover

Metz

Thackeray

Chmiel

. The multicenter AIDS cohort study: retention after 9 1/2 years. Am J Epidemiol. 1995;142(3):323–330.

23.

Hessol

Schneider

Greenblatt

. Retention of women enrolled in a prospective study of human immunodeficiency virus infection: impact of race, unstable housing and use of human immunodeficiency virus therapy. Am J Epidemiol. 2001;154(6):563–573.

24.

Hogg

Yip

Chan

. Rates of disease progression by baseline CD4 cell count and viral load after initiating triple drug therapy. JAMA. 2001;286(20):2568–2577.

25.

Abebe

Deribew

Apers

. Knowledge, health seeking behavior and perceived stigma towards tuberculosis among tuberculosis suspects in a rural community in Southwest Ethiopia. PLoS ONE. 2010;5(10):e13339.

26.

Worodria

Massinga-Loembe

Mazakpwe

. Incidence and predictors of mortality and the effect of tuberculosis immune reconstitution inflammatory syndrome in a cohort of TB/HIV patients commencing antiretroviral therapy. J Acquir Immune Defic Syndr. 2011;58(1):32–37.

27.

Worodria

Menten

Massinga-Loembe

. Clinical spectrum, risk factors and outcome of immune reconstitution inflammatory syndrome in patients with tuberculosis-HIV coinfection. Antivir Ther. 2012;17(5):841–848.