Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Checking up on serious adverse events

Internal standards

Suppose for a moment that you want to measure the concentration of a hormone like cortisol in blood or urine. In the absence of some of the superb immunoassays we have nowadays, you would probably argue that since cortisol is lipophilic, we should extract it from the blood or urine into an organic solvent, and measure it chromatographically.

It sounds deceptively simple. The trouble is that when examined in more detail, the efficiency of extraction of cortisol into organic solvent depends on the organic solvent used and the exact nature of what cortisol is being extracted from. So blood is different from urine, and blood samples differ from one another. Indeed, one has to be careful about the organic solvent, and some steroid hormones require freshly distilled ether, not a task undertaken lightly.

Internal standards

Some readers will be thinking that there is too much detail here, but it is necessary to introduce the concept of the internal standard. You see, if we add to our sample a minute but known amount of another steroid that is almost, but not completely, like cortisol, we can stop worrying about variability in extraction efficiency. We simply measure our cortisol concentration against the internal standard. QED.

Bandolier was intrigued to see the same concept used to evaluate the efficiency of reporting of serious adverse events in an enormous study of acupuncture in Germany [1]. It suggested huge under-reporting.


The research involved the German acupuncture cohort study which began in 2001 with 12,000 doctors in private practice, predominantly general practitioners. All were certificated for Chinese acupuncture, and they provide acupuncture services to over 600,000 patients a year. The cohort involved about 200,000 patients seeking acupuncture for chronic pain – headache, back pain, arthritis – over a period of six months.

Patients usually received 10 acupuncture sessions over five weeks. Physicians reported basic data and adverse events on a standard form after the last session. They were instructed on how to complete the form, and payment by insurer depended on completing the form. Adverse events were to be counted only once, and there was no attempt to categorise adverse events by severity.


Data were available for analysis on 190,924 patients receiving 1.8 million acupuncture sessions. Patients were aged from two to 97 years, with 84% aged 40 years or older. There were more females (69%) than males. Most (82%) had 10 sessions of acupuncture.

Minor adverse events were reported in 14,000 patients, with haematoma, temporary aggravation of the ailment, and vasovagal events as the most common specific minor adverse events (Table 1). Localised erythema, thought to be a desired acupuncture reaction, was reported in almost 1 in 4 patients.

Table 1: Adverse events reported in 190,924 patients receiving acupuncture

Adverse event
Rate per 10,000 patients
Localised erythema
1 in 4
All other minor adverse events
1 in 13
Specific adverse events
1 in 19
Aggrevation of ailment
1 in 77
Collapse, dixxiness, nausea, vomiting
1 in 142

By applying German population norms to the acupuncture population from five-year age and sex blocks, the expected number of deaths over the time of treatment was 180. Nine were reported (none obviously related to acupuncture), implying a rate of under-reporting of 95% (or perhaps that acupunture protects you from death?).

Serious adverse events were reported in 45 patients, including the nine deaths. This is an underlying rate of 2.4 per 10,000. In all cases the adverse event was reported to explain early termination of acupuncture sessions. Applying the correction factor for deaths, a twenty-fold higher rate of 47 per 10,000 might have been expected.


This study is not about the safety or efficacy of acupuncture, but about under-reporting of serious adverse events in clinical practice. There is an extensive literature on this, but we tend to forget about it most of the time.

In clinical trials, by contrast, there is extensive monitoring of participating doctors and patients to try to ensure complete reporting of potential harms. It is costly, but demanded for regulatory approval.

And we have not even begun to discuss causation here. Trying to figure out causation in the face of huge under-reporting smacks of the futile, and yet we agonise over it. What this study shows is that, in some cases at least, we might consider the use of an internal standard.


  1. HG Endres et al. An internal standard for verifying the accuracy of serious adverse event reporting: the example of an acupuncture study of 190,924 patients. European Journal of Medical Research 2004 9: 545-551.

previous or next story