Brown 2011.

Methods	Setting: 2 hospitals in Malawi, outpatient STI clinics Enrolment: participants enrolled from 2 October 2008 to 2 September 2009 Follow up: 2 weeks after initial diagnosis follow‐up was scheduled but authors did not report number of index patients returning for follow‐up
Participants	240 newly diagnosed HIV‐positive men (n = 100) and women (n = 140) from 2 Malawian hospitals were randomised Inclusion criteria From Lilongwe HIV‐positive test result for first time 18 years or older Sexually active in the last 90 days Willing and able to provide locator information for sexual partners Agreed to be randomised to method of PN and eligible Exclusion criteria Previously diagnosed with HIV
Interventions	All index patients were provided with referral cards, all counselled on importance of safe sex behaviour, staged according to WHO, blood drawn for CD4 count Simple patient referral (n = 77) Index patients notify partners themselves Contract referral (n = 82) Index patients were given 7 days to notify their partners after which a healthcare provider contacted partners, who had not reported to the clinic, for counselling and testing Provider referral (n = 81) Notification of partners within 48 hours by community outreach workers who were trained HIV testing counsellors or nurses
Outcomes	Primary outcome Partner visit to the clinic during the 30 days after index enrolment (identified as partners if they presented a partner referral card or their name was on the log of named partners) Secondary outcomes Harms ‐ abandonment (reported by index patient (2 weeks after enrolment) and partners (at clinic visit)) Partners testing positive (clinic records)
Notes	Authors did not report the number of index patients who came for 2‐week follow‐up. Authors were contacted but data from Malawi on 2‐week follow‐up were not available Ethics approval from Institutional Review Board at the University of North Carolina, Chapel Hill and the National Health Sciences Research Committee in Malawi Power was set at 85% to detect an absolute difference of 25% between passive referral and the 2 active referral study arms ‐ therefore need 80 index patients in each arm ‐ respective arms had 77, 82, 81 therefore sufficient
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised using a permuted block design with randomly allocated block sizes of 6, 9 and 12 stratified by sex and study site
Allocation concealment (selection bias)	Low risk	Was concealed in a sealed envelope until the end of the enrolment visit (after all partner data and locator information had been collected)
Incomplete outcome data (attrition bias) All outcomes	Low risk	Clinic visit and test outcome data were available for each partner of every index patient who was randomised Harms ‐ number of index patients returning for 2‐week follow‐up was not given, therefore, loss to follow‐up cannot be calculated
Selective reporting (reporting bias)	Low risk	Outcome for each partner of every index patient who was randomised was available. Protocol not available in 3 trial registries
Other bias	Low risk	No other bias identified
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding of participants not possible
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Personnel not blinded to which group index patient or partner belonged ‐ partner identified if presented with a patient referral card or if their name was found on the log. Index patient returned 2 weeks after enrolment and were asked if partners were notified, how they were notified and what their behaviour was like (harms). HIV antibody‐negative or antibody‐indeterminate specimens were tested for the presence of HIV RNA using the ultrasensitive Roche Amplicor Monitor HIV RNA assay. Primary outcome low risk