Linkage to HIV care and its determinants in the late HAART era: a systematic review and meta-analysis

Abstract

ABSTRACTPoor engagement into HIV care limits the effectiveness of highly active antiretroviral therapies (HAART) to improve survival and reduce transmission. The design of effective interventions to enhance linkage to care is dependent on evidence about rates of entry into HIV care. This is a systematic review and meta-analysis on linkage measurement and its determinants in the late era of HAART (post-2003), in high-income countries. We searched the PubMed and Web of Science databases, restricting our sample to the late HAART era (post-2003) until February 2016, and to high-income countries. We retained only studies that produced quantified outcomes. We rejected the studies with a high risk of bias, and followed a standard meta-analytic approach. Because there was a high heterogeneity (I2 > 90%), the aggregated findings were based on a random-effects model. A total of 43 studies were identified, all of them following a cohort of patients newly diagnosed until referred to specialized care. For a one-month period, the meta-proportion was 71.1% (IC95%: 61.0%–81.2). For a three-month duration, the meta-proportion of linkage to care was 77.0% (IC95%: 75.0%–79.0). For a one-year period, the meta-proportion was 76.3% (IC95%: 54.2%–98.4%). The proportions were lower when lab tests were used as referral indicator, with a pooled meta-proportion of 76.7% (IC95%: 73.0%–80.4), in comparison to a value of 80.8% (IC95%: 68.7%–92.9) for consultations. Being black or male were the most commonly observed determinants of delayed entry into care. Young people, injecting drug users, people with low socioeconomic status, or at a less advanced stage of disease also experienced lower proportions of timely linkage. Timely engagement into care is below 80% and specific sub-groups are particularly at risk of late entry. These findings confirm earlier evidence that linkage to care remains low, and that efforts should focus on vulnerable populations.