Effectiveness and cost-effectiveness of a nurse-delivered intervention to improve adherence to treatment for HIV: a pragmatic, multicentre, open-label, randomised clinical trial

Marijn de Bruin (Corresponding Author), Edwin J M Oberjé, Wolfgang Viechtbauer, Hans-Erik Nobel, Mickaël Hiligsmann, Cees van Nieuwkoop, Jan Veenstra, Frank J Pijnappel, Frank P Kroon, Laura van Zonneveld, Paul H P Groeneveld, Marjolein van Broekhuizen, Silvia M A A Evers, Jan M Prins

Research output: Contribution to journalArticlepeer-review

38 Citations (Scopus)
17 Downloads (Pure)


BACKGROUND: No high-quality trials have provided evidence of effectiveness and cost-effectiveness of HIV treatment adherence intervention strategies. We therefore examined the effectiveness and cost-effectiveness of the Adherence Improving self-Management Strategy (AIMS) compared with treatment as usual.

METHODS: We did a pragmatic, multicentre, open-label, randomised controlled trial in seven HIV clinics at academic and non-academic hospitals in the Netherlands. Eligible participants were patients with HIV who were either treatment experienced (ie, with ≥9 months on combination antiretroviral therapy [ART] and at risk of viral rebound) or treatment-naive patients initiating their first combination ART regimen. We randomly assigned participants (1:1) to either AIMS or treatment as usual (ie, containing a range of common adherence intervention strategies) using a computer-generated randomisation table. Randomisation was stratified by treatment experience (experienced vs naive) and included block randomisation at nurse level with randomly ordered blocks of size four, six, and eight. 21 HIV nurses from the participating clinics received three training sessions of 6 h each (18 h in total) on AIMS and a 1·5 h booster training session at the clinic (two to three nurses per session) after each nurse had seen two to three patients. AIMS was delivered by nurses during routine clinic visits. We did mixed-effects, intent-to-treat analyses to examine treatment effects on the primary outcome of log10 viral load collected at months 5, 10, and 15. The viral load results were exponentiated (with base 10) for easier interpretation. Using cohort data from 7347 Dutch patients with HIV to calculate the natural course of illness, we developed a lifetime Markov model to estimate the primary economic outcome of lifetime societal costs per quality-adjusted life-years (QALYs) gained. This trial is registered at ClinicalTrials.gov (number NCT01429142).

FINDINGS: We recruited participants between Sept 1, 2011, and April 2, 2013; the last patient completed the study on June 16, 2014. The intent-to-treat sample comprised 221 patients; 109 assigned to AIMS and 112 to treatment as usual. Across the three timepoints (months 5, 10, and 15), log viral load was 1·26 times higher (95% CI 1·04-1·52) in the treatment-as-usual group (estimated marginal mean 44·5 copies per mL [95% CI 35·5-55·9]) than in the AIMS group (estimated marginal mean 35·4 copies per mL [29·9-42·0]). Additionally, AIMS was cost-effective (ie, dominant: cheaper and more effective) since it reduced lifetime societal costs by €592 per patient and increased QALYs by 0·034 per patient.

INTERPRETATION: Findings from preparatory studies have shown that AIMS is acceptable, feasible to deliver in routine care, and has reproducible effects on medication adherence. In this study, AIMS reduced viral load, increased QALYs, and saved resources. Implementation of AIMS in routine clinical HIV care is therefore recommended.

FUNDING: Netherlands Organisation for Health Research and Development.

Original languageEnglish
Pages (from-to)595-604
Number of pages10
JournalLancet Infectious Diseases
Issue number6
Early online date3 Mar 2017
Publication statusPublished - Jun 2017


  • Journal Article


Dive into the research topics of 'Effectiveness and cost-effectiveness of a nurse-delivered intervention to improve adherence to treatment for HIV: a pragmatic, multicentre, open-label, randomised clinical trial'. Together they form a unique fingerprint.

Cite this