In the present study, no effect of repeated one-page ADR information letters on the overall reporting rate could be detected. The results are in contrast to the findings by Castel et al.
, who reported an increased reporting rate after distribution of a drug safety bulletin [7
]. One explanation for the divergent findings may be that the methodologies to evaluate differences in reporting rate differ between the studies; we used a randomized controlled design whereas Castel et al.
used a time series methodology. Nevertheless, other interventions may be more useful when the number of ADR reports is to be increased, such as education [6
] or detailed feedback to the reporting physician [12
Our results show that more high quality reports were received from intervention than control units, that is, the reports more often concerned ADRs which should be reported according to Swedish regulations. Thus, repeated ADR information letters may represent a valuable means to increase the number of reports which should actually be reported. One may speculate that a combination of such letters with an educational intervention may be even more effective as regards this aspect, since the latter also has been shown beneficial [6
]. Indeed, from a pharmacovigilance perspective, it is of importance that ADR reporting is focused on ADRs which are most likely to contribute to increased drug safety knowledge. ADR reports concerning well-known non-serious conditions may thus be of limited value since these contribute little to ADR signals. On the contrary, these reports constitute background noise, which may make detection of ADR signals more difficult at least as far as statistical signal detection methods within ADR databases are concerned, e.g. Proportional Reporting Ratios (PRR) [15
] and Bayesian Confidence Propagation Neural Network (BCPNN) [16
Varenicline was the most frequently reported substance in high quality reports. This substance was registered for smoking cessation in 2006 and thus the reports concerned a new drug. In addition, these reports also concerned unexpected ADRs. Indeed, the majority of high quality reports concerned unexpected ADRs. This finding may not be surprising since primary healthcare personnel probably observe serious ADRs less frequently than hospital personnel due to the definition of a serious ADR; any untoward medical occurrence that, at any dose (i) results in death; (ii) requires inpatient hospitalization or prolongation of existing hospitalization; (iii) results in persistent or significant disability/incapacity; or (iv) is life-threatening [3
Interestingly, 84 out of 151 primary healthcare units (56%) did not report any ADR during 2008. This finding supports a high degree of under-reporting [4
], and may indicate that primary healthcare personnel is an important target for interventions for improved reporting of ADRs.
Significantly more questionnaire responders had received and read the ADR information letters in the intervention group. However, the figures were generally low, indicating that ADR information letters are not prioritized reading for healthcare personnel in clinical practice. Interestingly, many questionnaire responders in the control units reported having received and read the ADR information letter. The intervention thus seems to have spilled over to the control units. Physicians and nurses may work in more than one primary healthcare unit, that is, both in the intervention group and in the control group. Furthermore, the units all belong to the same organization and information may thus easily pass from one unit to another. Another explanation for the finding that the ADR information letters were read by personnel in the control units is that ADR information from other sources, e.g. the pharmaceutical industry, may have been administered to the primary healthcare during 2008.
An important limitation of the present study is the small number of reports received from the small number of primary healthcare units. Indeed, given the available number of primary healthcare units in the region and the final results, the power of the present study to detect differences between the groups ended at 17%.
Another limitation is that our definition of quality of ADR reports was quite strict and related only to the Swedish regulations on ADR reporting. Thus, the information content of the report as regards other important aspects were not evaluated, such as factors of importance for the assessment of the strength of the relationship between the drug/s and the event/s, i.e. time to ADR onset, and response to dechallenge and rechallenge.