In the linked randomised controlled trial (doi:10.1136/bmj.b4248), Martins and colleagues assess the effectiveness of a food incentive to enhance completion of treatment for tuberculosis in Timor-Leste. The meal, or “feijuada,” was provided at health centres and comprised meat, beans, and vegetables with rice.1 It seems unlikely that anyone could describe the free lunch as anything but beneficial. Or could they? We need to know if the intervention is effective in achieving its desired purpose and what the opportunity costs may be before deciding if it is beneficial.

Although the success of treatment depends on treatment adherence, many patients do not follow their prescriptions and treatment recommendations.2 In the case of chronic communicable diseases such as tuberculosis and HIV infection, maximising adherence is crucial to optimising the outcomes of treatment not only for the individual but also for the community, through decreased transmission and risk of generation of drug resistant strains.

Experience across a variety of diseases shows that adherence is determined by a complex …