MATERIALS AND METHODS

Settings
The trial was approved by the RMIT University Human Research Ethics Committee. Application (2003/70) was filed with the Clinical Trials Branch under its Clinical Trial Notification Scheme, Therapeutic Goods Administration (TGA), Department for Health and Ageing, Canberra, Australia. This clinical trial also has been registered with Australian and New Zealand Clinical Trial Registry (ACTRN12612000209897).
The study was conducted from September to December 2003 during the pollen season in Melbourne, Australia. Daily pollen counts were provided by School of Botany, University of Melbourne. There were two clinical sites for this trial, at the City Campus and Bundoora Campus of RMIT University.

Subject recruitment
During August and September 2003, subjects were recruited through newspapers and radio. Inclusion criteria were: (a) age between 18 to 65 years with two or more symptoms for at least two years with a total nasal symptom score of at least six; (b) a positive pollen skin prick test (SPT).
Subjects with any of the following conditions were excluded from the study: (a) initiated immunotherapy since last SAR season; (b) HIV positive; (c) other active respiratory disease within 30 days of study; (d) receiving oral corticosteroid treatment; (e) pregnancy or lactation; (f) nasal polyposis; and (g) unable to read or understand English.
All subjects provided informed consent prior to the trial and were free to withdraw at any time. Subjects were advised to continue with pre-existing treatment for rhinitis or concomitant illnesses providing details of medication usage were recorded in the medication record form.

Outcome measures
Questionnaires were used to obtain subjects' demographic data. The primary outcome measure for assessing severity of SAR symptoms was a Five-Point Scale (0 = absent, 1 = mild, 2 = moderate, 3 = severe, 4 = very severe) [11]. Nasal symptoms included sneezing, nasal discharge, nasal itch, and nasal obstruction while non-nasal symptoms included eye itch, watery eye and red eye, itch of palate and or throat were recorded daily. Rhinoconjunctivitis Quality of Life Questionnaire (RQLQ) was used as a secondary outcome measure (QOL Technologies, UK). RQLQ has 28 items in seven domains, namely, activity limitations, sleep impairments, non-nose/eye symptoms, practical problems, nasal symptoms, eye symptoms, and emotional functions [12]. This self-administered instrument assessed the impact of SAR symptoms on the subjects' quality of life. The self-assessment was based on a seven-point scale ranging from not troubled (0) to extremely troubled (6). To monitor medication intake, subjects recorded details of daily Western medication intake during the trial. The relief medication score was calculated using the following scoring system: Nasal spray or eye drop (Over the Counter, OTC): each daily dose = 1 point; OTC antihistamines: each daily dose = 2 points; Oral Corticosteroid or nasal sprays: each daily dose = 3 points [6].

Study design
This randomised, double-blind, placebo-controlled trial was conducted over a period of 10 weeks (two-week baseline and an eight-week treatment).
During screening, inclusion and exclusion criteria were checked and subjects fully briefed. Informed consent was obtained prior to SPT followed by examination by an ear, nose and throat (ENT) specialist. A SPT for seven-grass mix (Kentucky blue, Meadow Fescue, Orchid, Perennial Rye, Red top, Sweet Vernal & Timothy, Bermuda), Birch, Bermuda, Plantain, Cypress, Ragweed, Negative control (Saline) and positive control (Histamine: 10 mg/mL) (Richards Thomson Pty Ltd., Australia) were completed. A wheal reaction measuring 3 mm or larger than diluent control was considered positive [6]. For safety parameters, blood samples were taken for liver, kidney function and full blood examination (FBE) tests at the beginning and end of trial. Random allocation of subjects to treatment groups occurred using statistical software by the School of Mathematical and Geospatial Sciences, RMIT. The randomisation codes were broken only after data analysis by the blinded statistician [6]. Subjects attended one of the two clinics fortnightly for collection of capsules and assessment. They evaluated the effectiveness of the blinding process on study completion by answering the question "What treatment did you receive through the study?" All trial medication bottles were collected and left-over capsules counted fortnightly for compliance monitoring.

Treatment protocol
Subjects were required to take two capsules each time, three times daily (with 4 h interval) for a period of eight weeks. Each capsule contained 500 mg of a mixture of granules of either herbal medicine (Koda Internationals Pty Ltd., Australia) or placebo ingredient (rice starch; Sigma Pharmaceuticals Pty Ltd., Australia). The capsules were prepared by a TGA approved Good Manufacturing Practice (GMP) certified manufacturer (GMP Pharmaceuticals Pty Ltd., Australia). The two types of capsules were identical in weight and appearance.
All ingredients of RCM-102 (Table 1) were TGA approved listed substances for human consumption. To comply with Australian Guidelines for Good Clinical Practice (1991), the capsule contents and known side effects of each individual ingredient were disclosed to all subjects in the trial information booklet as part of the informed consent process.

Safety parameters
All unexpected events were recorded during the treatment period, including details of signs and severity of symptoms, duration and any actions taken to resolve them. FBE was done by the RMIT School of Medical Sciences. The liver and kidney function tests were done by Dorevitch Pathology Pty Ltd. in Melbourne, Australia prior to and after the trial.

Statistical analyses
Based on findings of a previous study [6] , the nasal symptom score for the treatment group was statistically significantly different compared to the placebo group (F = 6.118, p = 0.017). This yielded an effect size estimate of 0.66. For 80% power using a two-sided significance level of 5%, the required sample size is 38 per group. To allow for dropouts, we used a sample size of 52 per group.
All data were analysed using the Statistical Package for the Social Sciences (windows version 11.5; SPSS, USA) for Windows by a statistician blinded to the allocation of treatments to groups. When comparing measures at the end of the study for the two groups, ANCOVA was used with the baseline as covariate. The data from non-repeated measures such as blood tests was analysed using t-tests. Outcome measures with nominal responses were analysed using χ2 tests. All p-values were obtained via 2-sided tests and were assessed at α = 0.05. The age variable was used as a covariate for outcome analyses as the two treatment groups were not comparable on this variable. Intention to treat was employed in all relevant data analyses.