We described cohorts of Vietnamese HIV-infected patients with late presentation to care. Most patients registered at OPCs already with advanced stages of HIV infection. These findings were similar to reports from other resource-limited settings, including sub-Saharan Africa 
, China 
, India 
, Thailand 
and Haiti 
. Even after several years of experience managing the national ART program, late presentation remains problematic in Vietnam as it does in many other countries 
. Although Vietnam’s ART program scaled up rapidly from 2005 to 2009, the median CD4 cell count at clinic registration and before ART remained under 100 cells/mm3
throughout this period. Significant stigma and discrimination towards HIV-infected persons, weak Information/Education/Communication programs and poor linkage between counseling and testing services and OPCs may be factors contributing to late access to care. Moreover, the scale-up rate in the early stage of the ART program (i.e., 2004–2006) was slow due to time and resources needed to set up OPCs, train health care workers, and to gain experience for more rapid ART initiation and wider coverage.
Although patients in this cohort presented with advanced infections, 12-month retention on treatment was high (>80%). Such retention is similar to those reported from a systematic review of cohorts in sub-Saharan Africa 
, as well as country specific reports from Rwanda 
and South Africa, Malawi and Côte d’Ivoire 
. The national ARV program in Thailand also achieved a high survival rate of 0.89 after 1 year and 0.78 after 5 years 
. Moreover, despite low CD4 cell counts at baseline, our evaluation showed a positive immunologic response to ART. The gain in CD4 cell count continued steadily through the course of ART, with a median gain of 94 cells/mm3
at six months and 142 cells/mm3
at 12 months. The trend continued at 24, 36, and 48 months. This promising response was similar to results found in sub-Saharan Africa 
, China 
, Thailand 
, and Haiti 
This programmatic evaluation also identified factors associated with attrition. Being male, having a history of IDU, low CD4 cell count, high WHO stage, and low BMI were risk factors for attrition. Some of those, such as being male, low CD4 cell count, high WHO stage, and low BMI were common risk factors reported in resource-limited settings 
. History of IDU was a strong risk factor for attrition, which could be explained by poor adherence, active drug use, or co-morbidities. Clinic factors (i.e., size, type, location) were not found to be associated with attrition, although factors specific to PWID (e.g., access to MMT) were not available for analysis because the MMT program was still being piloted at the time of the study.
The important difference between this study and other ART cohorts is the high proportion of PWID. IDU has been the major route of transmission in Vietnam 
. PWID in this evaluation presented with more advanced HIV infections and were more likely to have evidence of HCV co-infection. Cohorts of PWID in other countries have reported similar findings: PWID were more likely to have lower nadir CD4 cell count measurements, present with AIDS defining illnesses 
, have higher rates of active TB 
, have higher prevalence of hepatitis/HIV co-infections 
, and receive HIV diagnosis late 
. In Vietnam, major barriers for PWID to access care and treatment include poor socioeconomic status, discrimination (and in some cases, self-discrimination), lack of knowledge about HIV disease and benefits of early testing and care, concern about confidentiality, lack of transportation, limited availability of IDU-centered services such as MMT and outreach support (Family Health International 360, unpublished data).
Although retention of PWID at 12 months was also promising (81.5%), fewer PWID than non-PWID were retained over the course of ART: while non-PWID maintained retention rates of 88.2% at 3 years and 85.7% at 5 years, the retention rates of persons with a history of IDU were significantly lower, only 67.4% at three years and 63.3% at five years. Additionally, PWID were found to have less robust immune reconstitution than patients without a history of IDU (Table S2
). Other reports have shown that persons with a history of IDUs were less likely to achieve viral load suppression 
, more likely to have treatment interruption, 
and lower CD4 gain 
. In our evaluation, contributing factors to less favorable outcomes of PWID might include ongoing drug use, co-morbidities or poor adherence, although data were not available to identify any associations. Because adherence data were inconsistently recorded in the charts, we were not able to analyze adherence among PWID. Anecdotal reports from clinics participating in the evaluation showed that ongoing drug use accounted for a number of patients who had died of presumed overdose or had been confined to drug treatment centers without continuing access to comprehensive care and ART. Efforts have been made by VAAC to provide care to drug treatment center residents and maintain ART access. Expansion of MMT and other harm reduction programs targeting PWID with better linkages to OPCs is underway and should help increase access to care and improve adherence among PWID on ART.
The present evaluation had several limitations. First, data quality was suboptimal because the source was patient charts used by health care providers. Indeed, many data were missing, including baseline CD4 cell count, weight, WHO stage, and viral hepatitis sero-status. Testing was not always routinely performed and therefore results were not always recorded. CD4 measurement and viral hepatitis testing was not widely available at all clinics, especially early in the ART program (i.e., 2004–2006). Another important data quality issue was related to the adherence of healthcare workers to national guidelines. Although Vietnamese national guidelines recommend WHO stage, weight, clinical status, and adverse events to be recorded at every patient visit, we found that a considerable number of charts were missing this data (Table S1
). ART laboratory monitoring was also often not performed per guidelines, similar to another study 
. Critical tests such as CD4 cell count, liver transaminases, and hemoglobin sometimes were not done at 6-month intervals as recommended. Some patients did not have any CD4 measurements at follow up or had CD4 measurements done more than 1 year after ART initiation. These quality issues were reported back to VAAC to help establish a more rigorous quality control program with the goal of improving quality of care.
A second limitation was that we were not able to assess what happened to patients who were lost to follow-up or to assess the reasons for deaths as the charts did not routinely record these data. Other studies tracking loss to follow-up reported that one third 
to half 
of LTFU patients died and many of the patients were untraceable. A meta-analysis of studies from Africa, Asia and Latin America reported 40% of LTFU patients had died, and among patients for whom re-contact was successful, reasons for not returning to clinics included transfer to other programs, financial difficulties, improving or decreasing health status and stigma 
. In the context of Vietnam, we would expect similar circumstances. Many of the LTFU patients may have already died. Especially in this population of PWID, active drug use may contribute to loss to follow up, either because patients died or were admitted to government drug treatment centers.
The third limitation was the ascertainment of IDU history. Many patients who had used intravenous drugs may not have revealed their history, leading to a falsely low proportion of PWID. High HCV co-infection among reported non-PWID (18.1%) may serve as a proxy for IDU status. Also, we were also not able to obtain data on whether patients continued intravenous drug use while on ART or whether they had access to MMT.
Despite these limitations, the evaluation provided significant insights and useful lessons for the Vietnamese ART program. The evaluation served as an assessment of data quality at OPCs and based on its results recommendations were made to VAAC to 1) revise the current patient chart format to be more user-friendly and to capture additional important patient monitoring indicators; 2) provide training to OPC staff to increase data use for internal quality improvement; 3) strengthen patient monitoring; 4) emphasize the importance of routine quality control activities; and 5) consider an evaluation of LTFU patients to better understand their outcomes and reasons for LTFU. Although the evaluation did not provide data to assess specific factors associated with less desirable outcomes among PWID, such as ongoing drug use, lack of access to MMT, and adherence, we believe that strengthening IDU-specific interventions, such as linkage to MMT which is becoming rapidly more available, as well as counseling and treatment support, will help reduce LTFU and improve treatment outcomes. In an effort to increase early access to ART, VAAC revised the national guidelines in 2011 to raise the threshold of CD4 cell count for ART initiation to 350 cells/mm3
, and it recommended ART to all patients with active TB regardless of the counts. The Vietnam MOH also committed to piloting the WHO/UNAIDS “Treatment 2.0″ initiative 
. Up to now, the MMT program has been expanded to 41 clinics with stronger linkages to OPCs. As a result of the combination of these efforts, PWID may expect to experience more favorable outcomes.
Data from this routine program evaluation of ART scale-up in Vietnam demonstrated a high retention rate and good immune reconstitution in response to treatment, indicating a high level of program quality in this predominately PWID population. However, efforts to promote earlier identification of HIV-infected persons and earlier initiation of ART remain a priority, as well as efforts to improve the quality of clinic services and data to further reduce LTFU among PWID.