The most widely used definition of severe recurrent throat infection comes from the Pittsburgh tonsillectomy trial of Paradise et al. (1984): seven or more episodes in the preceding year, five or more in each of the preceding two years, or three or more in each of the preceding three years, with each episode documented in a clinical record and marked by fever, cervical lymphadenopathy, tonsillar exudate or a positive test for group A streptococcus. Current American guidance uses the same thresholds and recommends watchful waiting for children who fall below them (Mitchell et al., 2019). Scottish guidance takes a similar position on indications for surgery (SIGN, 2010).

The reason for caution about surgery is the same reason for caution about case reports. A Cochrane review found that tonsillectomy gave a modest reduction in sore throat episodes in the year after operation compared with non-surgical management, and that many children who were not operated on improved spontaneously (Burton et al., 2014). Children enter treatment at a time of unusually frequent illness, so some decline would be expected through regression to the mean alone (Barnett et al., 2005).

The controlled evidence does not support a preventive effect. A systematic review of randomised trials of homeopathy for childhood and adolescent ailments concluded that the evidence was not convincing enough to recommend it for any condition (Altunç et al., 2007). A Cochrane review of homeopathic products for preventing and treating acute respiratory tract infections in children found no consistent benefit over placebo and judged much of the evidence to be of low quality (Hawke et al., 2018). A meta-analysis of individualised prescribing across conditions found few trials that could be regarded as reliable (Mathie et al., 2014). A case report in this field therefore starts from a low prior probability that the remedy is active.

The CARE guideline sets out the items a clinical case report should contain, including a timeline, the diagnostic assessment, the intervention, follow-up and outcomes, and the patient's perspective (Gagnier et al., 2013), and an accompanying document explains each item with examples (Riley et al., 2017). A supplement for homeopathic cases, HOM-CASE, adds items on the type of homeopathy, the medicine and its potency, the rationale for selection and the assessment of possible causal attribution (van Haselen, 2016). Case reports have a recognised place in medicine, chiefly in describing the unexpected and in generating hypotheses; they are not a means of establishing that a treatment works (Vandenbroucke, 2001).

The report should begin from the medical record and not from parental recollection. For each episode in the two or three years before the first homeopathic prescription, the clinician preparing the report should extract the date, the findings on examination, whether a throat swab or rapid antigen test was done, and what was prescribed. Episodes that were never seen by a clinician should be listed separately and labelled as undocumented. The baseline can then be set against the Paradise thresholds, and the reader can see whether the child's illness was severe, moderate or within the ordinary range for age. Clinical scoring of individual episodes, for instance with the McIsaac modification of the Centor score (McIsaac et al., 1998), helps to distinguish probable streptococcal infection from viral sore throat, a distinction on which the Infectious Diseases Society of America guideline places great weight (Shulman et al., 2012).

The homeopathic intervention should be described so that another practitioner could understand exactly what was given: the remedy by its full name, the manufacturer, the potency and scale, the form, the schedule and every later change, together with the features of the child on which the choice rested. The practitioner's training and the length and number of consultations belong here too. Equal care is owed to everything else the child received. Antibiotic courses, analgesics, referrals to an ear, nose and throat service, changes in nursery or school attendance, exposure to tobacco smoke at home and vaccination history can all bear on the frequency of infection, and omitting them invites the reader to assume that nothing else changed.

Outcomes should be chosen and written down at the first consultation. The primary outcome should be the number of clinician-documented episodes per year, defined as at baseline. Secondary outcomes can include antibiotic courses, days absent from school or childcare, and a validated quality-of-life measure such as the Pediatric Quality of Life Inventory (Varni et al., 2001). A parent-held diary is useful for capturing milder episodes, but entries should be kept distinct from clinic-confirmed ones. Follow-up contacts should be scheduled at fixed intervals over the two years and not only when the child is unwell, so that periods of good health are recorded as carefully as periods of illness.

Before drafting, the author should set out the explanations that compete with a remedy effect and say what the record shows about each. The list is short and predictable: natural remission with age, regression to the mean from a peak year, the effect of concurrent conventional treatment, changes in exposure such as leaving a nursery, and more lenient counting of episodes once treatment has begun. Structured causality tools can discipline this step. The algorithm of Naranjo et al. (1981) was devised for adverse drug reactions and scores features such as timing, response to withdrawal and re-challenge, and alternative causes; HOM-CASE asks authors of homeopathic reports to make a comparable assessment explicit (van Haselen, 2016). In a remitting childhood illness most of these criteria cannot be met, and the honest result will usually be that causation is possible but unproven.

Formal single-patient trials, in which treatment and placebo periods alternate under blind conditions, were developed for stable chronic symptoms that return when treatment stops (Guyatt et al., 1986). They suit an episodic condition with a falling background rate poorly, and the framework does not pretend to offer an equivalent.

The manuscript should follow the CARE checklist item by item, with the HOM-CASE items added, and should include a timeline figure that shows every documented episode before and after the first prescription on the same scale. The title and abstract should describe the case as adjunctive and should not use words such as cure or successful treatment. Written consent from a parent or guardian, and the child's assent where age allows, are required for publication, in keeping with the principles of the Declaration of Helsinki (World Medical Association, 2013).