HATRIC trial
Trial design
We conducted a phase II double-blind feasibility trial with UK GP practices cluster-randomised to give liquid or tablet preparation (for logistical reasons), and within each practice, eligible patients were individually randomised to Pelargonium sidoides root extract (EPs®7630 – Kaloba®) or matched placebo. The detailed protocol for this trial has already been published [33]. This trial adheres to CONSORT guidelines.
Participants
We recruited patients from 20 GP practices in southern England. We included adults aged 18 years and over, presenting to their GP with an acute cough (≤21 days’ duration) as their main symptom, and with symptoms localising to the lower tract (e.g. sputum, chest pain, dyspnoea, wheeze). This illness definition has been used in other studies in this population [6, 34]. We excluded patients with suspected pneumonia on the basis of focal chest signs (focal crepitations, bronchial breathing) and systemic features (severe breathlessness, high fever, vomiting, severe diarrhoea), patients with serious illness who required hospital admission, exacerbations of COPD, serious comorbidities, and pregnant women (or at risk of pregnancy – defined as any woman of reproductive age not using the combined oral contraceptive pill, a hormonal intrauterine device, a hormonal contraceptive injection, or a subcutaneous hormonal implant).
Eligible patients were invited to participate in the trial by their GP or nurse and were given a patient information sheet during their GP appointment. If they consented, they were then given the study medication and the study diary on the spot and were advised that a trial coordinator would be in touch soon regarding their diary. The randomisation list for sites and patients was generated by Southampton Clinical Trials Unit using the command ralloc in Stata v15 [35]. We used block randomisation (with varying block size) in a 1:1 allocation ratio of placebo to Kaloba® treatment. Treatment packs were sent to sites in sets of four and each patient was allocated the next available sequentially numbered patient pack at their site. Neither the patient nor the doctor or nurse knew to which treatment (Kaloba® or placebo) they had been randomised. Patients were free to withdraw consent from the study at any time without providing a reason.
Interventions
Patients on experimental treatment were given a root extract of Pelargonium sidoides DC (Geraniaceae). The product used was EPs®7630, manufactured by Dr. Willmar Schwabe GmbH & Co. KG, Germany (named Kaloba®). The extraction solvent used was 11% ethanol (w/w), such that 10 g (= 9.75 mL) of oral solution contains 8.0 g extract from the roots of Pelargonium sidoides DC (1: 8–10), and one film-coated tablet contains 20 mg of extract (as dry extract, 1: 8–10). The verum medication provided for the study is taken from production lots sold in the market. All test methods and specifications for the materials, intermediates and the final product are part of the registration dossier and therefore approved by the respective authorities. The liquid also contained glycerol and ethanol (120 mg / 1 ml). The tablets contained the following excipients: Maltodextrin, microcrystalline cellulose, lactose monohydrate, croscarmellose sodium, precipitated silica, magnesium stearate, and the film coating (hypromellose, macrogol, iron oxide yellow E172, iron oxide red E172, titanium dioxide E171, talc, simeticone, methylcellulose and sorbic acid).
The extract was obtained from dried roots of Pelargonium sidoides extracted with 11% ethanol w/w, resulting in a drug extract ratio of 1: 8–10 for the liquid extract and the dry extract as well (obtained from the liquid extract by drying). Roots of Pelargonium sidoides were collected in South Africa (e.g. Eastern Cape). The dried material was tested in an array of phytochemical and biochemical methods to confirm the quality and identity of the herbal material. Pharmacognosy was done by the quality control department of Dr. Willmar Schwabe GmbH & Co. KG, Germany (Dr. H. Hentrich and lab technicians specifically skilled in pharmacognostic test procedures). A voucher specimen of every lot is deposited in the department of Pharmacognosy to be retained for ten years. In addition, the herbal material is tested with respect to purity (e.g. heavy metals, and microbiological quality).
Quality testing
The extract (EPs® 7630) used in the herbal medicinal product (Kaloba® manufactured by Dr. Willmar Schwabe GmbH & Co. KG, Germany) is classified by the European Pharmacopoeia as “other extract” and therefore not adjusted to a particular content of constituents. Independent of this formal classification, the constituents of this herbal active ingredient have been described in detail [36]. Approximately 80% m/m of the extract are assigned to six major groups of constituents, oligomeric prodelphinidines (commonly designated in this context as polyphenols) being the most significant group (approximately 40% of the dried extract).
The liquid herbal medicinal product used in the study (Kaloba® manufactured by Dr. Willmar Schwabe GmbH & Co. KG, Germany) is tested and released compliant with drug GMP and the European Pharmacopoeia according to written, authorized and validated analytical procedures with respect to identity, content (ethanol, glycerin, extract; HPLC) and microbiological quality. The film-coated tablets are tested with the same methods, adapted to the respective dosage form with respect to extract content (HPLC), uniformity, disintegration and microbiological purity. The respective placebo preparation is tested with the same procedures, except for herbal drug content (i.e. absence of active ingredient) and in addition for visual appearance in comparison to the batch of herbal drug product used as medication. The quality testing is performed by the Quality Control laboratory of Dr. Willmar Schwabe GmbH & Co. KG, Germany by a laboratory technician with extensive long-term experience in the testing of herbal medicinal products. Personnel manufacturing or performing quality tests on investigational medicinal products including placebo preparations have been specifically skilled and trained in this field.
The liquid placebo was designed to match the appearance of the liquid herbal medicinal product. This was achieved by using the identical solvent composition (water, glycerol, ethanol) and by replacing the herbal active ingredient with colouring and flavouring. The placebo tablets were also designed to match the appearance of the active tablets, by using the identical film-coating composition and a coloured granulation for manufacturing of the uncoated tablets matching the tablet core of the herbal medicinal product.
The dosage of the liquid (Kaloba® or placebo) was 30 drops (approximately 1.5 ml) three times daily, to be taken 30 min before meals. For tablets (Kaloba® or placebo), the dose was one 20 mg tablet three times daily, to be taken 30 min before meals. Patients were advised to continue this treatment daily until 2–3 days after symptoms had resolved but that treatment duration should not exceed 2 weeks. This is the dosage recommended by the manufacturer and on the traditional herbal registration.
In addition to the trial treatment, GPs were allowed to select the clinically appropriate prescribing strategy as per practice policy, no antibiotics or to prescribe antibiotics, either to be taken immediately, or as a “delayed” prescription.
Outcomes
Baseline data were collected at the GP practices where patients were recruited, and medical records were reviewed after one month to extract data on NHS resource usage. Patients were given a diary to complete for 4 weeks, including their symptoms, treatments taken, out-of-pocket expenditure, days off work related to their acute bronchitis and quality of life measurements (EQ5D). All patients were asked to complete the EQ-5D-5L questionnaire at baseline (day 1) and day 7 and approximately half of the patients were also randomly selected to be given the questionnaire at additional time points on days 2 and 4, in order to assess the acceptability of collecting quality of life data more frequently. Patients were telephoned at day 1 or 2 to check for any problems with diary completion, and again at days 14 and 28 days to prompt diary completion and return. If diaries were not returned, or were returned incomplete, a brief telephone interview to collect the key data was undertaken after 35 days.
Feasibility outcomes were: recruitment rate, withdrawal rate from the study, return rate of patient diaries, percentage of completion of patient diaries, compliance with medication according to diary data and returned medication, type of antibiotic prescription given (i.e. immediate/delayed/not given), percentage of patients who took antibiotics, time to antibiotic usage, mean symptom severity (at days 2–4), percentage of patients resolved, time to resolution of symptoms and duration of treatment with herbal medication. The minimum key outcome dataset from the diaries was considered to be [1] antibiotic use in the 28 days post randomisation, [2] no longer experiencing moderate symptoms for two consecutive days and [3] study medication use. A health economic study was a part of the feasibility study which aimed to develop the methods of data collection both for quality of life and for usage of key resources for the design of the future phase III trial.
We explored several possible outcome measures for clinical effectiveness (Table 3). “Duration of symptoms rated as moderately bad or worse” was used in the largest clinical trial of treatment for acute bronchitis [6]. The “last day” definitions allow for the possibility that the illness may fluctuate in severity over the 28-day period and the “first day” definitions do not. Therefore, the last day will tend to be later than or equal to the first day. “Proportion of patients with symptom resolution at day 7” was used in the Cochrane review [30].
Statistical methods
The intended sample size was 160 patients overall (40 patients in each of the 4 arms) recruited from 20 GP practices in the UK. No formal sample size calculation was carried out however, ignoring clustering, using a 95% confidence interval approach and an expected proportion of 50% of eligible patients randomised into the trial (to give the worst-case scenario), it can be shown that this sample size would allow us to predict the recruitment rate to within 8% using nQuery Advisor v7.0. Accounting for the clustering based on an intra-cluster correlation (ICC) of 0.05 and an expected proportion of 50% of eligible patients randomised into the trial, this sample size allows us to predict the recruitment rate (number of eligible patients randomised into the trial) to within 13%, given an average cluster size of 8, and 20 recruiting sites.
A detailed statistical analysis plan was developed prior to the analysis. All analyses were conducted using STATA version 15 [35]. We aimed to present descriptive feasibility data rather than to test hypotheses and patients were analysed as randomised.
Health economic methods
The economic study was designed from the NHS and a Personal Social Service perspective. In addition, we collected personal costs to test whether a societal perspective should be considered in future trials. Key resources included the costs of the intervention and NHS service use including medication, primary care consultation, outpatient attendance, A&E visits, and hospitalisation. Data on NHS resource usage was extracted through a review of medical records in the 28 days following recruitment. Out-of-pocket spending and days off work related to lower respiratory tract infection were collected through patients’ self-reported diaries. Quality of life data were measured by giving the EQ-5D-5L questionnaire to all patients at baseline (Day 1) and Day 7 and half of the patients were randomly asked to complete the questionnaire at additional time points on Day 2 and Day 4. This aimed to assess the acceptability of more frequently collecting quality of life data in future trials.
HATRIC-Q nested qualitative study
We conducted a nested qualitative study of patients who met inclusion criteria (including those who did not consent to participate in the clinical trial) and health professionals. Patients were asked for their consent to pass their contact details to the qualitative researcher, who contacted them by telephone to conduct a semi-structured interview, using an interview guide. Health professionals were invited by email. Consent was received verbally by telephone. We aimed for a maximum variation sample, based on pre-specified criteria, such as ethnicity, gender, age, employment status, site, trial drug (liquid drops or tablets), diary status (complete, partly completed, by recall, not returned), and recruitment performance (for the staff interviews). The patients and health professionals who had participated in the trial were asked about trial materials and trial procedures, while those patients who declined to participate were asked for their reasons for non-participation. Interviews were audio-recorded and transcribed verbatim. We also explored the interviewees’ experience of LRTI and its treatment with antibiotics, as well as their attitudes towards delayed prescribing of antibiotics and their views on herbal medicine. Only the results pertinent to trial design are reported here.
Transcripts were analysed using inductive thematic analysis, following the framework approach [37]. The transcripts were compared within and between each other to search for themes, which were then reviewed, defined and named. The analysis followed three key steps:
-
1.
Familiarization: repeated readings of transcripts and listening to recordings assisted familiarisation with the data, identification of initial codes and interesting pieces of data
-
2.
Categorization: grouping similar responses into meaningful categories
-
3.
Classification: assessing relationships between categories and more abstract grouping in order to explain the data
Standard methods were used to safeguard rigor, including multiple coding by DS, RY, MW and GL to check the validity and consistency of coding.
Ethical and other approvals
HATRIC was registered on the ISRCTN registry (ISRCTN17672884) on 16 August 2018.