A link between a harmful effect and use of a drug should be assessed according to the epidemiological method (see Box 8.10).
Box 8.10 Criteria for analysing causation
|
1. Association - any epidemiological association is observed. Bias and confounding factors should be kept in mind.
2. Temporality - cause should precede the effects. Was the epidemiological study properly designed so that the temporality could be proved?
3. Consistency - repeated observation of an association in different populations under different circumstances (different place and different time).
4. Strength -
• high odds ratio
• high significance level (low p value) or high lower limit of 95% confidence interval of odds ratio
• dose-response relationship.
5. Specificity - for example, thalidomide embryopathy: association is specific.
6. Coherence - coherent with other evidence:
• does not conflict with other clinical evidence
• coherent with evidence from laboratory experiments.
|
Source: Based on criteria published elsewhere.2,3
This does not mean that you cannot assess causation from a single clinical case. If you find one or a few important cases of an adverse reaction reported in a clinical trial, search the medical literature (e.g. PubMed, Embase) for published case series and/or epidemiological analytical study reports. If only one or a few case reports can be found, these may be assessed individually with the help of an algorithm (see Box 8.11).
Box 8.11 Example of an algorithm for assessment of adverse drug reactions4
To assess the adverse drug reaction, answer the following questions with a pertinent score |
| |
(Yes, No, Do not know) |
Example |
1. Are there previous conclusive reports on this reaction? |
(+1, 0, 0) |
0 |
2. Did the adverse event appear after the suspected drug was administered? |
(+2, -1, 0) |
2 |
3. Did the adverse reaction improve when the drug was discontinued or a specific antagonist was administered? |
(+1, 0, 0) |
0 |
4. Did the adverse reaction reappear when the drug was readministered? |
(+2, -1, 0) |
0 |
5. Are there alternative causes (other than the drug) that could on their own have caused the reaction? |
(-1, +2, 0) |
-1 |
6. Did the reaction reappear when a placebo was given? |
(-1, +1, 0) |
0 |
7. Was the drug detected in the blood (or other fluids) in concentrations known to be toxic? |
(+1, 0, 0) |
0 |
8. Was the reaction more severe when the dose was increased, or less severe when the dose decreased? |
(+1, 0, 0) |
0 |
9. Did the patient have a similar reaction to the same or similar drugs in any previous exposure? |
(+1, 0, 0) |
0 |
10.Was the adverse event confirmed by any objective evidence? |
(+1, 0, 0) |
0 |
| |
Total score* |
1 |
* A total score of 9 or more suggests a definite causative link; with a score of 5 - 8 the link is probable; with 1 - 4 it is possible.
This algorithm may help when the authors of a clinical trial have concluded that an adverse event was not related to the drug under investigation. Quite often adverse events that authors have misclassified as unrelated to the treatment can be identified as suspected adverse reactions by using this algorithm. If epidemiological analytical data such as a case-control study suggest a significant relation between an adverse event and a drug, analyse the causation using the criteria listed in Box 8.11.
For example, in clinical trials of the antidiabetic agent pioglitazone, cardiovascular events such as myocardial infarction, severe palpitation necessitating hospital admission and ischaemic strokes were sometimes reported, but investigators often classified them as non-related events.5 According to the algorithm they should be classified as ‘possible’ even though some cause other than the drug could have led to the reaction. Moreover, animal data for pioglitazone show dose-dependent cardiac toxicity with a similar toxicity profile to that seen in clinical trials. These findings suggest a strong link between the drug and the adverse effect and thus the cardiovascular events could be classified as ‘probable’ adverse reactions to pioglitazone.
Suicidal events in the clinical trials of SSRI antidepressant drugs are another notorious example.6 They should be classified at least as ‘possibly related’ for the same reason.
Loading...
