Two percent of patients, who received acupuncture for chronic low back pain or neck pain, experienced adverse effects which required either self-treatment (1.2%), medication/physician treatment (0.6%), or hospital admission (0.03%). Using the guidelines of the European Commission, to describe the adverse effects of medicinal products (European Commission: Enterprise and Industry Directorate-General, 2005), these frequencies of adverse effects from acupuncture would be rated as common (self-treatment), uncommon (medication/physician treatment) and rare (hospital admission). The most frequently reported adverse effects from acupuncture were bleeding, hematoma, and pain. Adverse effects persisted for a median of 3days and were associated with higher costs (on average approximately €125 for the 3months period) if treatment was needed. Overall the costs were 9–11% higher in the group which reported adverse effects which needed treatment compared to those without adverse effects. We assume that the number of adverse effects which require treatment influence the overall cost-effectiveness of acupuncture.
We observed additional costs of about €3400 for the 3month period if hospital treatment for pain associated adverse effects was required. When the costs of the adverse effects of the 3month period were hypothetically extrapolated to the whole study population of patients with chronic low back pain and neck pain treated with acupuncture, every patient would have to spend €2.26 to treat the possible adverse effects associated with acupuncture.
This is the first evaluation of the cost and need for treatment due to adverse effects associated with acupuncture. The strengths of this study include the large sample size incorporating more than 700,000 acupuncture treatments, the detailed evaluation within a usual care setting, the high quality cost data provided by the health insurance companies and the direct reporting of adverse events by patients to minimize underreporting bias. However, although patients' reporting tends to be higher than physicians' reporting on adverse effects even their rates can be too low and a underreporting cannot be fully ruled out (Witt et al., 2006).
However, patient reports have limitations such as recall bias, coincidence of acupuncture and undesirable effects. The main limitation is that there was no waiting list group to control for unspecific adverse effects. Also confounding could play a relevant role. Moreover, because of the very large sample included in ASH, categorization of data was necessary. This might have caused information loss to some extent. However, in order to detect rare adverse effects which result in hospital admissions, an observational study is a suitable design.
The number of adverse effects observed in our study appears to be high, but of note is that most of them were mild. Non-steroidal anti-inflammatory drugs also cause adverse effects (serious gastrointestinal effects in 0.1–2.5% of patients per year (Schaffer et al., 2006) or death in 0.04–0.11% of patients (Tramer et al., 2000)). The observation period for acupuncture was only 3months compared to one year for non-steroidal anti-inflammatory drugs, due to this the figures are not directly comparable. Within this study acupuncture treatment was only available once per year which is similar to acupuncture for chronic low back pain within the statutory health insurance system. Based on the results of our study in a large sample we would not have expected very serious adverse effects such as dead to appear more often in a longer treatment period. Overall acupuncture appears to be a safe treatment.
Costs did not differ for patients with and without adverse effects. Higher costs were only observed if patients had adverse effects which required treatment. Overall the parts of the economic analysis are more difficult to interpret. It is difficult to know how to interpret the results that patients who reported inflammation as a side effect and subsequently treated these themselves can have lower costs than those with no adverse effects. However, the number of patients was small and results from economic analysis with a limited number of patients have to be interpreted with caution, because economic data for single patients is highly variable and is sensitive to confounding. Therefore conclusions regarding the attributable costs from adverse effects due to acupuncture in patients admitted to hospital should also be made with caution. Another limitation arises from the fact that our economic analysis is mainly based on data provided by the participating health insurance companies. Any additional payments (particularly co-payments made by patients) were not available and could not be included.
In conclusion adverse effects due to acupuncture can occur. However, most of them are minor and do not require additional treatment. Only when treatment was required did they result in additional expenses.