Objectives To investigate the efficacy and safety of CS1002, an over-the-counter cough treatment containing diphenhydramine, ammonium chloride and levomenthol in a cocoa-based demulcent.

Design A multicentre, randomised, parallel group, controlled, single-blinded study in participants with acute upper respiratory tract infection-associated cough.

Setting 4 general practitioner (GP) surgeries and 14 pharmacies in the UK.

Participants Participants aged ≥18 years who self-referred to a GP or pharmacist with acute cough of <7 days' duration. Participant inclusion criterion was cough severity ≥60 mm on a 0–100 mm visual analogue scale (VAS). Exclusion criteria included current smokers or history of smoking within the past 12 months (including e-cigarettes). 163 participants were randomised to the study (mean participant age 38 years, 57% females).

Interventions Participants were randomised to CS1002 (Unicough) or simple linctus (SL), a widely used cough treatment, and treatment duration was 7 days or until resolution of cough.

Main outcome measures The primary analysis was intention-to-treat (157 participants) and comprised cough severity assessed using a VAS after 3 days' treatment (prespecified primary end point at day 4). Cough frequency, sleep disruption, health status (Leicester Cough Questionnaire (LCQ-acute)) and cough resolution were also assessed.

Results At day 4 (primary end point), the adjusted mean difference (95% CI) in cough severity VAS between CS1002 and SL was −5.9 mm (−14.4 to 2.7), p=0.18. At the end of the study (day 7) the mean difference in cough severity VAS was −4.2 mm (−12.2 to 3.9), p=0.31. CS1002 was associated with a greater reduction in cough sleep disruption (mean difference −11.6 mm (−20.6 to 2.7), p=0.01) and cough frequency (mean difference −8.1 mm (−16.2 to 0.1), p=0.05) compared with SL. There was greater improvement in LCQ-acute quality of life scores with CS1002 compared with SL: mean difference (95% CI) 1.2 (0.05 to 2.36), p=0.04 after 5 days' treatment. More participants prematurely stopped treatment due to cough improvement in the CS1002 group (24.4%) compared with SL (10.7%; p=0.02). Adverse events (AEs) were comparable between CS1002 (20.5%) and SL (27.6%) and largely related to the study indication. 6 participants (7%) in the CS1002 group reduced the dose of medication due to drowsiness/tiredness, which subsequently resolved. These events were not reported by participants as AEs.

Conclusions Although the primary end point was not achieved, CS1002 was associated with greater reductions in cough frequency, sleep disruption and improved health status compared with SL.

Trial registration number EudraCT number 2014-004255-31.