In this observational comparative effectiveness study, no significant long-term outcome differences were seen after 36 months in children with atopic eczema when treated conventionally compared to homoeopathic treatment; neither were short-term differences seen (at six or 12 months 
). Both groups substantially improved during the observation period and in both groups every tenth patient used corticoids. Patients in the conventional groups showed a trend towards earlier improvements. Costs, however, were higher in the homoeopathic group.
The design of the study (observational, usual-care and multicentre setting) allows evaluation of a therapy’s comparative effectiveness considering the patients’ own preferences and therapy choices. In our study the treatment was individualised and to reflect a more realistic care additional medication was not forbidden. Data collection was performed using a variety of sources including the affected child, their parents, study doctors and external blinded raters, to improve the objectivity and validity of the study outcomes.
The aim of this study was to reflect the real world situation and to compare conventional and homeopathic care provided by physicians in a usual care setting. Thus, we chose to take patients’ and/or parents’ therapy preferences into account, making randomisation not possible. The observational design resulted in relevant baseline differences between the two groups. In the homoeopathic group severity of disease appeared higher compared to the conventional group. Eczema conditions according to the SCORAD were less favourable (especially the intensity and extent of atopic eczema) and the disease duration was longer. In addition, parents in the homoeopathic group were older, had a higher social class background and a higher treatment expectation. To take baseline differences into account, we adjusted our analyses for these factors. However, it is possible that other unknown and unmeasured factors might have influenced the results. If other confounding factors were present but not accounted for, or if the performed adjustments were not sufficient (e.g. due to broad value categories or measurement error), residual confounding might be present. If adjustments did not sufficiently balance disease severity, then results might be biased in favour of either the conventional or the homoeopathic group. Therefore, the non-randomised design is a clear limitation of our study regarding the internal validity of our results 
At six, 12 and 36 months SCORAD severity was comparable in both groups, although patients in the conventional group took more conventional medication in the first year, e.g. corticosteroids, antihistamines or pimecrolimus and tacrolimus 
. Baseline use of pimecrolimus and tacrolimus was comparable in both groups, in contrast to less use of corticosteroids in the homoeopathic group than in the conventional group. While at six months, the use of these anti-inflammatory drugs was lower in the homoeopathic group, after 36 months around ten percent of patients in both groups used corticoids. Overall medication use decreased during the trial in both groups compared to baseline values. In both groups the frequency of basic skin care was comparable. However when interpreting these results, one should also take into account that atopic eczema can improve spontaneously in young children.
Although disease specific costs in both groups appear lower in the third year compared to the first year, the long-term costs were more than twice as high in the homeopathy group compared to the conventional group. Costs were mainly driven by doctors’ fees and paying for medical aids. As described in the methods section, the mean costs during months 7–12 of the study were used to estimate the costs for the outpatient contact at the long-term follow-up for the main analysis. We chose this method as it also seems to be a conservative approach, in assuming that the intensity of medical contact (especially for homeopathic treatment) is much higher during the first year of treatment, and that the use of these costs for estimating follow-up expenditures might lead to an overestimation of costs. Data from another prospective observational study that included 3981 patients with different diagnoses showed that children with atopic eczema 
visited their homeopathic doctor between month 7–12 40% more often (1.5±1,7 times) than a year later (0.9±1.2 times in months 19–24). These findings were used as the basis for a sensitivity analysis, resulting still in higher costs in the homoeopathic group.
The substantial and statistically significant cost-differences between the groups found during the first year of treatment were stable over time. The follow-up between 31 and 36 months lead to a comparable result compared to the analysis in the first year. While interpreting these differences, potential limitations should be kept in mind, particularly in regard to how the outpatient costs were estimated. Within the 31–36 months follow-up detailed therapeutic documentation was not available. However, the method of calculating the outpatient costs during the first year of the study was based on this kind of documentation.
In this long-term observational study after three years, while unable to rule out residual confounding but taking patient preferences into account, treatment at homoeopathic doctors was similar, yet not superior to treatment at conventional doctors for children with mild to moderate atopic eczema, but still had higher costs.