An observational study was conducted in the prefecture of Heraklion, Crete, Greece (February – June 2016). Sixteen pharmacies, purposively selected, participated as recruiting sites. The preparation of the extract is extensively described elsewhere .
Patients presenting at the pharmacies with symptoms of upper respiratory tract infection, who voluntarily requested and choose to purchase the extract as part of their treatment were recruited. Eligibility for inclusion in the study was based on the Jackson criteria for the identification of patients suffering from common cold . In order also to increase the possibility that the infection was viral, and not of any other type, the presence of fever was considered as a perquisite for inclusion in the study. As such, inclusion criteria were: age over 18 years, sudden emergence of fever within 48 h from the visit in the pharmacy and presence of at least one of the symptoms: cough, sore throat, nasal discharge or ingestion, headache, muscle pain, sweating, rigors and fatigue. Exclusion criteria were: age under 18 years, absence of fever within 48 h from the visit to the pharmacy, daily consumption of aspirin (≥100 mg), presence of malignancies, immune system disorders, liver damage and pregnancy.
Setting as primary outcome the day patients reported that their symptoms ceased completely and assuming 5% probability of type-I error, a sample size of 185 patients had 90% power to detect an average time of symptoms’ cessation of 6 days, with standard deviation of ±1.2 days and accuracy of 0.17 days.
Upon presenting at the pharmacies, eligible patients were invited to complete an anonymous questionnaire (Additional file 1: Questionnaire.doc), developed based on the Wisconsin Upper Respiratory System Survey (WURSS-21), which assessed symptoms’ severity . This tool has been previously used with reliable results . Symptoms were recorded at the day of visit at the pharmacy (day 1), as well as at the fourth (day 4) and seventh (day 7) day of extract intake, via phone calls performed by the pharmacists.
Symptoms were classified as local (cough, sore throat, itchy throat, hoarseness, nasal congestion, excreta, sneezing and headache) and general (fatigue, muscle pain and rigors). They were scored in a Likert scale, with 0 denoting absence of symptom and 7 denoting maximal severity of the respective symptom. For each patient, three indexes were calculated: The score of local symptoms, the score of general symptoms and the total score of symptoms, defined as the sum of general and local symptoms’ scores. Presence of fever was recorded as a dichotomous variable (yes/no).
Collected socio-demographic and clinical characteristics included age (years), gender (male/female), BMI (Kg/m2), smoking (currently/former/never) and presence of chronic conditions (yes/no). Any supplementary medication received for the treatment of patients’ upper respiratory tract infection was also recorded (type and days of intake). The day patients reported feeling better and the day patients reported that their symptoms ceased completely were recorded as well.
Sample characteristics were summarized using descriptive statistics. For comparisons between categorical variables Pearson’s χ2 tests were performed. Paired Samples T-test and McNemar tests were used for comparisons of paired differences. To assess the severity of all symptoms over the observation period, the Area under the Curve (AuC) was computed. Multiple linear regression models were performed with dependent variables: the AuC, the day patients reported feeling better and the day patients reported their symptoms ceased completely. Independent variables were the socio-demographic and clinical characteristics and supplementary medication intake (yes/no). Linear mixed models with dependent variables the total score of symptoms and presence of fever and independent variables the day of observation (1, 4, 7), supplementary medication intake (by day of observation), socio-demographic and clinical characteristics were used to assess symptoms’ alleviation over time. Confidence level was a = 0.05. Statistical software used were IBM SPSS version 21 and Stata SE version 12.