This study evaluated the proportion of individuals linking to HIV care in a public sector service in Cape Town, South Africa. Only 63% of patients attended for a CD4 count measurement within 6 months of diagnosis. Although a substantial proportion of patients had CD4 counts ≤200 cells/µl (34%) and were therefore eligible for ART according to South African guidelines 
, only 67% of these started ART within 6 months. Among those who did have a timely CD4 count but were not yet eligible for ART, only 46% returned for a repeat CD4 count after a median time of 8 months. Individuals testing through ANC services had better linkage to HIV and ART care and higher CD4 counts at time of HIV diagnosis compared to individuals accessing the other services.
HIV is a chronic disease and comprehensive HIV care needs to be provided within a continuum of care 
. ART is just one of the components of HIV care and care of individuals not yet requiring ART is equally important 
. The continuum of HIV care starts when an individual is diagnosed with HIV. ART eligibility should be assessed when individuals are newly diagnosed and in regular (6 monthly) intervals thereafter. Individuals not yet eligible for ART should receive comprehensive HIV care including cotrimoxazole, isoniazid preventive therapy, screening for TB and cervical cancer, contraceptive advice, counseling and social support until they become eligible for ART. Following initiation of ART individuals needs to be supported within the same framework to ensure good adherence and retention in care.
We identified a number of important issues in our study. First, people who tested on their own initiative were least likely to have a timely CD4 count measurement done, underscoring the need to ensure that scale up of VCT programmes will be accompanied by clear plans to ensure that those who test positive go on to receive appropriate care. Second, almost a third (28%) of eligible patients with TB did not receive ART despite recommendations in favour of concomitant treatment since 2003 
, and ART being associated with a 64–95% reduction in mortality in such patients 
. This underscores the importance of integrating HIV and TB services 
This study shows that men and younger adults fail to access health services efficiently. Only 30% of clients tested for HIV were men. This is consistent with studies showing that HIV-infected men are less likely to access treatment 
, present with more advanced stages of HIV disease 
and have a higher mortality risk during ART 
. Repeated CD4 counts were less likely in individuals under 30 years of age as also reported elsewhere
It is important to note that less than half of patients whose first CD4 count was above the ART eligibility threshold came back for a repeat test. One way of improving ART uptake, and thus reduce mortality among patients who are otherwise lost to care, might be to change the CD4 threshold to 350 cell/µl in line with the latest World Health Organization recommendations 
Our overall finding that 33% of patients eligible for ART were lost to care is consistent with several reports from elsewhere in southern Africa. In a programme report from South Africa, only 55% of patients had a CD4 count measurement within 8 weeks of HIV diagnosis and 81% of eligible patients were on ART at 3 months follow-up 
. Out of 2483 patients eligible for ART in Uganda 637 (26%) did not start ART; a third of these patients died before ART initiation and another quarter were alive but not taking ART 
. In Mozambique only 57% of patients testing HIV positive entered HIV care and 31% of patients eligible for ART started ART within 3 months 
In our study only 63% of patients testing positive for HIV attended for a CD4 count measurement within 6 months. These outcomes are worse than those recently reported by a public-sector clinic in Johannesburg where 84.6% of patients who tested positive for HIV had a CD4 count measurement. The majority of these patients did not return for their CD4 result within 12 weeks 
. Data from the same clinic in Johannesburg showed that among patients not yet eligible for ART only 26% returned for a scheduled pre-ART medical visit within one year compared to 43% of our patients not yet eligible for ART returning for a repeat CD4 count 
Substantial improvement in linkage to HIV care for TB and VCT patients was observed in more recent years in this study and yet this was not accompanied by improvements in linkage to ART. Failure of linkage to HIV and ART services translates into incomplete ART coverage at population level, seriously undermining the potential for reductions in mortality, morbidity, TB incidence and HIV transmission.
The study has several strengths and limitations. Strengths include that the study was conducted in a routine clinical program where CD4 count testing and ART were provided free. Thus, the results should be generalisable to similar settings. The study was conducted over a prolonged period with increasing ART availability. Among the limitations is the fact that patients might have been misclassified as failing to link to care if they accessed care with a service provider other than the primary health care clinic or hospital. Thus, linkage to care might be underestimated. However the nearest other ART site is more than 10 km away, and residents of this poor community are unlikely to have sought care in such a distant ART site unless they had moved away. Second, we did not assess if patients who had a CD4 count measurement actually returned to receive the result. Thirdly, we did not investigate reasons for not linking to care. Studies that have ascertained outcomes among patients lost to care have reported that up to a third of patients who failed to initiate ART had died 
. Time cut-offs for linkage to care for both timely CD4 count and ART initiation are somewhat arbitrary. When no time cut-offs were used 75.3% (95% CI 70.3–80.3) of eligible individuals who had a CD4 count at some point during the study period eventually initiated ART.
In conclusion, while considerable attention has been paid to loss to follow-up and mortality among patients receiving ART 
, data on losses at earlier stages of the care pathway are scarce. As our study shows, a focus only on outcomes of those patients fortunate enough to initiate treatment fails to account for a substantial number of patients who are eligible for ART but do not receive it or not yet eligible but fail to reappear. Pre-ART defaulting should be encouraged in programme reporting. Programme adaptation to ensure retention in care between testing and ART should consider point of care CD4 count testing at time of HIV diagnosis as well as provision of integrated TB and HIV.