Objective To evaluate the first year of the demonstration scheme for the reporting of suspected adverse drug reactions (ADRs) by community pharmacists.
Design Reports submitted by community pharmacists between April, 1997, and March, 1998, were compared with reports from general practitioners (GPs) in the regions in which the demonstration scheme was carried out. The comparison included cumulative data on the nature of the reports submitted, as well as completeness and causality assessments of individual reports.
Results Ninety-six reports were submitted by community pharmacists. These were found to be comparable to those submitted by GPs, with regard to the proportion relating to serious reactions and black triangle drugs. Furthermore, community pharmacists submitted higher proportions of reports of ADRs associated with herbal products and reports of suspected generic inequivalence. A lower proportion of community pharmacist reports required follow up. There were no statistically significant differences between reports submitted by community pharmacists and GPs with regard to the causality and completeness assessments.
Conclusions Despite the relatively small number of reports submitted during the first year of the demonstration scheme, this analysis indicates that community pharmacists submit reports of suspected ADRs which are comparable to those submitted by GPs. This suggests that the nationwide extension of the yellow card scheme to include ADR reporting by community pharmacists would be beneficial. Furthermore, community pharmacists can make an additional valuable contribution to the yellow card scheme in areas which are traditionally not widely reported, such as ADRs associated with herbal products.
One of the objectives of regulatory pharmacovigilance is the long-term monitoring of medicines as used in normal clinical practice to identify previously unrecognised drug safety hazards or changes in adverse effect profiles. Spontaneous suspected adverse drug reaction (ADR) reporting is one of the principle sources of information used in regulatory pharmacovigilance to identify signals of such hazards. The UK spontaneous reporting scheme, known as the yellow card scheme, has been in operation since 1964. Reports of suspected ADRs are primarily submitted on a voluntary basis directly by doctors, dentists, coroners and hospital pharmacists. Reports are also received via the pharmaceutical industry, which has a statutory obligation to report suspected ADRs.1
At the time of the extension of the yellow card scheme to include hospital pharmacists in April, 1997, a demonstration scheme for the reporting of suspected ADRs by community pharmacists was launched.2,3 This scheme aimed to evaluate the potential contribution of this group of reporters. The scheme was located in and run by the four regional monitoring centres (RMCs) of the Committee on Safety on Medicines (CSM).
The demonstration scheme had two components. The first, in the Mersey and West Midlands RMCs, required reporting community pharmacists to be working in a community pharmacy with the facility to maintain patient medication records. The second, in the Northern and Welsh RMCs, had no requirement relating to patient medication records. Community pharmacists in these regions were eligible to report if they worked within any community pharmacy or were based in a GP practice within the region.
Community pharmacists reporting under either component of the demonstration scheme were requested to report suspected ADRs in accordance with the UK ADR reporting guidelines.4 They were also requested to focus their reporting on areas such as over-the-counter products and herbal remedies, where there is limited reporting by doctors and where community pharmacists may have a special role.
At the time of the launch, information packs were provided to community pharmacists in the areas covered by the RMCs. These included background information on the yellow card scheme, the UK ADR reporting guidelines and supplies of the reporting forms to be used by community pharmacists. In addition, individual RMCs provided education on the demonstration scheme to groups of community pharmacists around the time of the launch.
In this paper, data on suspected ADR reports submitted by community pharmacists during the first year of the demonstration scheme have been compared with those submitted by GPs, in order to determine any differences between the types of report submitted by the two sets of reporters. This comparison has included cumulative data on the nature of the reports submitted, as well as completeness and causality assessment of individual reports.
All spontaneous reports received via the national yellow card scheme from April 1, 1997, to March 31, 1998, were identified. A total of 17,769 spontaneous reports were received during this time: 3,621 reports were received via the RMCs, of which 96 (3 per cent) were submitted by community pharmacists and 1,975 (55 per cent) by GPs.
Analysis of cumulative data All reports received via the RMCs from community pharmacists during the first 12 months of the demonstration scheme were analysed to determine the proportion of reports relating to serious and fatal suspected ADRs, and proportions of reactions associated with newly introduced drugs, identified by an inverted black triangle symbol in the BNF (t; "black triangle" drugs). In addition, the number of reports followed up by the RMC to obtain additional information from the original reporter in order to faciliate assessment of the report, and the proportion of reports for which follow up was successful, were also determined. The data for community pharmacists were compared with cumulative data for reports received from GPs via the RMCs during the same time period.
Statistical analysis of the cumulative data was carried out on STATA, using the chi-squared test; P-values are quoted for statistically significant comparisons. Where a comparison found no statistically significant difference using the chi-squared test, the 95 per cent confidence intervals for the difference between the two groups were calculated.
Assessment of individual reports All reports submitted via the RMCs by community pharmacists in the first 12 months of the demonstration scheme were assessed individually for completeness and causality together with a randomly selected 5 per cent sample (98 reports) of GP reports submitted via the RMCs during the same period. Both the completeness and causality assessments were carried out blind.
Completeness was assessed by examining individual reports for the presence or absence of individual "key" data elements, such as patient age and sex, suspect drug name, suspect reaction, outcome of the reaction. Missing data elements were given the score –1, and the completeness score for individual reports calculated. The lowest possible score was –14, and a score of 0 represented a complete report. Each report was evaluated separately by two experienced scientific assessors from the post-licensing division of the MCA. The score for each report was averaged and the mean completeness score for each subset of reporters calculated; a t-test was used to compare the means.
Causality was assessed using the Naranjo "ADR probability scale", an established method first published in 1981.5 The Naranjo algorithm requires a series of questions to be answered for each report. The answers are then scored and the total score for each report calculated. The total score for a report indicates the probability category (definite, probable, possible, doubtful) to which the report is allocated. Each report was assessed separately by an experienced medical and an experienced scientific assessor from the post-licensing division of the MCA.
Overall, there was no difference between the proportions of reports of serious reactions submitted by community pharmacists and GPs (40 per cent v 40 per cent; 95 per cent confidence interval for the difference –10 per cent to 10 per cent). The proportion of reports for black triangle drugs was 10 per cent lower for community pharmacists (29 per cent) compared with GPs (39 per cent), with the difference approaching statistical significance (P=0.055).
There was no statistically significant difference between community pharmacists and GPs in the proportions of reports relating to fatal suspected ADRs (0 per cent v 0.8 per cent; 95 per cent confidence interval for the difference –1.2 per cent to 0 per cent). However, it should be noted that this is based on a very small number of reports and the power of the analysis to detect a difference is therefore very low.
Although the number of reports of suspected ADRs associated with herbal products submitted by both groups of reporters was low, the difference between community pharmacists and GPs was statistically significant, with pharmacists submitting a higher proportions of reports for herbal products compared with GPs (4.2 per cent v 0.4 per cent; P<0.001). Similarly, community pharmacists submitted a statistically significantly higher proportion of reports of suspected generic inequivalence compared with GPs (3.1 per cent v 0.6 per cent; P<0.01).
The proportion of GP reports followed up for additional information was statistically significantly higher than for community pharmacist reports (16 per cent v 7.3 per cent; P<0.05). Although there was a 21 per cent difference in the proportion of reports followed up successfully (36 per cent [GPs] v 14 per cent [community pharmacists]; 95 per cent confidence interval for the difference –47 per cent to 5 per cent), the low absolute number of reports in the community pharmacist group meant that this difference did not reach statistical significance.
It had been intended to compare cumulative data on community pharmacist reports from the two components of the demonstration scheme, in order to evaluate the relevance of the pharmacist's facility to maintain patient medication records on reporting. However, the small number of reports involved in the comparisons did not allow for meaningful statistical comparison between the two arms of the study.
The completeness assessment showed no statistically significant difference between reports submitted by community pharmacists and GPs, with reports from both subsets of reporters having an average of one missing data element per report (–1.16 v –0.90; t=–1.47; P>0.05).
|
|
|
Although the total number of reports submitted by community pharmacists during the first 12 months of the demonstration scheme was low, the analysis of the reports is nevertheless encouraging. Compared with GPs, community pharmacists submitted a similar proportion of reports of serious reactions. There was a suggestion of a lower proportion of reports for black triangle drugs being submitted by community pharmacists. The reason for this is not known but it is possible that since many black triangle drugs may be initiated within a hospital setting, community pharmacists have less interaction with patients over such drugs, whereas GPs will be kept informed by the hospital staff of the treatment given to their patients. Alternatively, the GP may have submitted reports for such drugs after the community pharmacist has contacted them to discuss an adverse reaction.
A higher proportion of GP reports was followed up with the original reporter for additional information, compared with community pharmacist reports. Since the decision to follow up an individual report is based on an algorithm which takes into account the completeness of a report, this might suggest that community pharmacists are submitting more complete reports than GPs. However, the assessment of individual reports for completeness did not support this, with completeness scores for GP and community pharmacist reports showing no statistically significant difference. The reason for the difference in the proportion of reports followed up is not known, however, it is possible that GP reports are more commonly followed up because it is perceived that they are more likely to be able to provide any additional information required. The far larger proportion of reports followed up successfully by GPs compared with community pharmacists would support this possibility.
The assessment of causality using the Naranjo algorithm showed that community pharmacist reports did not differ from GP reports. However, it is notable that the majority of reports for both groups of reporter were clustered in the "possible" causality category, with no reports at all appearing in the "definite" category. The Naranjo algorithm was developed based on an assessment of published case reports. For each case, it is necessary to answer a series of questions relating to specific information such as blood concentrations of the suspect drug, effects of administration of a placebo or specific antagonist of the suspect drug, and the results of rechallenge. Although such information may be available for published, and therefore often very well documented cases, the amount of information provided on yellow card reports is variable, and rarely contains the level of detail that would be seen in a published case report. This lack of information on yellow cards leads to reports accruing low scores and being clustered into one category. This algorithm appears therefore to have low sensitivity for cases where less detailed information is available, such as spontaneous ADR reports.
It had been originally intended to analyse reports from community pharmacists with regard to the reporting of ADRs associated with over-the-counter (OTC) products. Although the reports submitted included reports for drugs which can be supplied without a prescription, a lack of information on the reports with regard to source of the drug (OTC or prescription) or the specific product concerned meant that this was not possible. However, many OTC products are obtained from a pharmacy, so it is reasonable to assume that the community pharmacist is well placed to inform patients about these products and be made aware of adverse reactions which the patient may have experienced.
The current analysis has found that community pharmacists submitted a greater proportion of reports for herbal products and reports of suspected generic inequivalence compared with GPs (10 and five times as many, respectively), despite the small number of reports involved.
Overall, the findings of this study suggest that community pharmacists submit suspected ADR reports which are generally similar to those submitted by GPs. Furthermore, community pharmacists can make an additional valuable contribution to the yellow card scheme in areas which are traditionally not not widely reported, such as ADRs associated with herbal products, OTC remedies and reports of suspected generic inequivalence.
Acknowledgments We are grateful to all the staff at the Committee on Safety of Medicines regional monitoring centres in Merseyside, Newcastle upon Tyne, Wales and the West Midlands for their hard work in running the demonstration scheme. In addition, we acknowledge the assistance of Mr Shaun Delaney, Miss Kate Foy, Miss Kavita Kalsi, Dr Elliot Brown, Dr Munir Pirmohamed and Dr Robin Ferner in the causality and completeness assessments.
| 1. Waller PC, Coulson RA, Wood SM. Regulatory pharmacovigilance in the United Kingdom: current principles and practice. Pharmacoepidemiol Drug Safety 1996;5: 363-75. |
| 2. Committee on Safety of Medicines/Medicines Control Agency. Extension of the yellow card scheme to pharmacists. Curr Prob Pharmacovigilance 1997;23:3. |
| 3. Pharmacists' adverse drug reaction reporting to start on April 1. Pharm J 1997;258:330-1. |
| 4. British National Formulary (no 37). London: British Medical Association and Royal Pharmaceutical Society of Great Britain. March 1999: 9. |
| 5. Naranjo CA, Busto U, Sellers EM, Sandor P, Ruiz I, Roberts EA et al. A method for estimating the probability of adverse drug reactions. Clin Pharmacol Ther 1981;30:239-45. [Medline reference] |
Dr Davis is ADROIT signal generation co-ordinator and Ms Coulson is specialist in pharmacovigilance information management at the Medicines Control Agency. Correspondence to Dr Davis at Room 1016, Medicine Control Agency, Market Towers, 1 Nine Elms Lane, London SW8 5NQ