150 likes | 303 Views
Randomisation ceremonies: from ping-pong to football matches. Richard Hayes. JAMA 1995; 273: 408-412. 2,838 citations . Aims of public randomisation ceremonies. To explain background to trial To explain need for random allocation and procedures to be used for this
E N D
Randomisation ceremonies: from ping-pong to football matches Richard Hayes
JAMA 1995; 273: 408-412 2,838 citations
Aims of public randomisation ceremonies • To explain background to trial • To explain need for random allocation and procedures to be used for this • To carry out randomisation using transparent procedure, involving key stakeholders Rationale is to ensure fairness, transparency, buy-in, engagement Stakeholders: Government representatives, community leaders, community representatives
Ghana bednet trial • Rural area of Northern Ghana • Widely scattered households • Divided into 96 clusters of geographically contiguous households - average of 1400 persons per cluster • 48 clusters: impregnated bednets • 48 clusters: control arm – bednets at end of trial • Mortality in children (6-59m) reduced by 17% • Randomisation carried out by lottery in 10 paramount chiefdoms in study area Binka et al, TMIH 1996, 1: 147-154
MEMA kwaVijana trial • 20 rural communities in Mwanza Region, Tanzania • Adolescent sexual health intervention • In-school sexual health education • Youth-friendly services • Youth condom distributors • Cohort of 9,645 young people (Av age 15.5y) followed up for 36 months • Further cross-sectional survey carried out 5-6y later • Effects on knowledge, attitudes and some reported sexual behaviours • No impact on HIV, pregnancy, other STIs Ross et al, AIDS 2007, 21:1943 Doyle et al, PLoSMed 2010, 7:e1000287
MkV randomisation scheme • 20 clusters divided into three strata based on initial survey of HIV prevalence • High-risk: 6 clusters • Medium-risk: 8 clusters • Low-risk: 6 clusters • Stratified randomisation Possible allocations = 6C4 x 8C3x 6C3 = 28,000
MkV restricted randomisation • Restricted randomisation • Used to ensure adequate balance across study arms • Restriction criteria • Mean HIV prevalence similar • Mean chlamydia prevalence similar • Two communities near mines evenly distributed • Even distribution over four districts • 953 of the 28,000 allocations satisfied criteria Hayes et al, Contemporary Clinical Trials 2005, 26:430
MkV randomisation ceremony • Held with senior stakeholders • List of allocations presented, numbered 001–953 • Stakeholders asked to draw ping-pong balls numbered 0-9 from bucket • Number 039 selected: divides 20 communities into two sets of 10 • Final draw of ping-pong ball used to decide which set of 10 receives intervention
Zamstar trial • Impact of 2 intensive TB control strategies evaluated in 2x2 factorial CRT • Arm 1: Enhanced case finding • Arm 2: Household intervention • Arm 3: ECF + household • Arm 4: Neither intervention • 24 communities in Zambia and S Africa • Household intervention decreased TB prevalence by 20% and reduced transmission to children • No effect of ECF Ayles et al, Trials 2008, 9:63
Zamstar randomisation scheme • 24 clusters divided into 4 strata based on initial TST survey in schoolchildren • Zambia/High TST: 8 clusters • Zambia/Low TST: 8 clusters • S Africa/High TST: 4 clusters • S Africa/Low TST: 4 clusters • Stratified randomisation Possible allocations = = 3,657,830,400
Zamstar restricted randomisation • Restriction criteria • Similar number with high estimated HIV prevalence in each arm • Similar mean TST prevalence • Similar number of “open” communities • Similar number of urban communities • Political constraints • Randomly selected allocations until 1,000 which satisfied restriction criteria Sismanidis et al, Clin Trials 2008, 5:316
Zamstar randomisation ceremony • Held on football pitch during tournament in Zambia • List of acceptable allocations numbered 000-999 • 24 volunteers representing clusters • Senior officials asked to draw footballs numbered 0-9 from sack • Number 773 selected: volunteers divided into four groups labelled A-D • More footballs drawn to determine study arm allocated to each group
Special thanks to • Fred Binka for the Ghana video • BabisSismanidis for the Zamstar video • The PIs and investigators of the Ghana bednet, MkV and Zamstar trials