Discussion

This study evaluates the costs and benefits of the pan-London seasonal influenza programme offering eligible individuals their flu vaccine in community pharmacies as an alternative to local GP delivery. To assess flu vaccine coverage across London before and after the start of the pharmacy initiative, we analysed two data sets, ImmForm (GP reports) and Sonar (pharmacy reports). To evaluate the London-specific costs from the NHS and pharmacy perspectives of seasonal flu vaccination in pharmacies and GPs and the opinions of vaccine providers, we conducted an online survey administered to pharmacists and GPs across London. Our results do not show any evidence for increased vaccine uptake as a result of the pan-London flu vaccine pharmacy initiative. The cost to the NHS per vaccine dose of pharmacy delivery for the London initiative was lower than that of GP delivery. While we found evidence that the programme offered convenience and choice to patients, we found several disadvantages of the programme, most notably the incomplete reporting of pharmacy vaccine delivery into both recording systems.

To calculate incomplete reporting of vaccine delivery in the GP ImmForm system, we compared the number of vaccine doses that were noted to be delivered at somewhere other than the GP practice with the number of doses delivered to patients at pharmacies. If the place of vaccine dose administration was not correctly registered in ImmForm, then our results will have underestimated the number of vaccine doses that are recorded in ImmForm. Nevertheless, this potential issue further highlights the inefficient and information loss-prone nature of maintaining two recording systems with manual entry. Moreover, it is unclear how pharmacies ascertain whether individuals are eligible for vaccination. For a national programme with increased pharmacy vaccine uptake to remain efficient, it will be important to easily identify patient eligibility. For a national system, it may be helpful to identify advantages and failings of other vaccination reporting software, such as the Information Immunization System (IIS) used in the USA.13

Previous studies suggest that pharmacy-provided vaccines may increase uptake of seasonal influenza vaccine in high-risk groups. In a survey among 503 high-risk, non-elderly people in a pilot in Sheffield, 19% of vaccine recipients at a pharmacy stated that they would not have received the vaccine otherwise,14 an estimate consistent with a similar Canadian study.5 In our study, if 19% of people who received the vaccine at a pharmacy, that is, 19% of 10% of the total eligible population, would not have done so at their GP, the overall increase in vaccine coverage would be less than 2%. Such a change would likely be too small a difference to detect against the background fluctuations in coverage. We calculated the vaccine coverage for all seasons from ImmForm data. The true vaccine coverage is likely to be higher due to evidence of under-reporting of pharmacy-administered doses in ImmForm data. However, even accounting for these changes, our results remain consistent. Nonetheless, our analysis does not include the possibility of a secular trend of decreasing vaccine uptake in GP practices, which may obscure increases in coverage because of pharmacy administration. Moreover, it is likely that over time, pharmacy-administered vaccines will become more widely known and a rise in coverage observed. Our surveys indicate that pharmacy patients report convenience as the main factor in the decision to seek the flu vaccine at a pharmacy rather than their local GP, a finding consistent with other studies. For instance, people trust pharmacists to administer vaccines,4 ,15 they are happy with the vaccination service they receive5 ,14 and they believe that pharmacies offer a convenient way to receive the vaccine.14 ,16 Indeed, one study reports that some people would rather pay out of pocket at the pharmacy instead of receiving the vaccine for free in primary care, largely due to convenience and accessibility.16 In our analysis, we observed a relatively high use of the pharmacy among carers. This result was also seen in a pilot programme using community pharmacies on the Isle of Wight, where there was an increase in uptake among carers and healthcare workers.17 A higher uptake among these groups (under 65 years and healthy) was perhaps expected as in the USA it has been shown that people who use the out-of-hours service provided by the community pharmacists (evenings and weekends) mostly consist of those under 65 years, and those without clinical conditions.3 This result suggests that the group using pharmacies the most are healthy individuals for whom vaccination allows only indirect protection of the elderly or sick. To streamline this route of vaccination, it may be possible to provide pharmacy vouchers with carer passports. Conversely, low pharmacy uptake for risk groups with historically low vaccine uptake, such as individuals with liver disease, neurological disorders or immunosuppression, provides increasing inequity for these groups.

Studies that have previously assessed physician opinion of regarding non-physician administered vaccine provision in the USA are broadly consistent with our findings: with between 60% and 70% of physicians agreeing that pharmacists and other providers have adequate training and provide a more convenient service to the patients.18 Moreover, 70% of physicians report incomplete vaccine documentation as a result of non-physician vaccine provision,18 and 90% of physicians have reported that this incomplete documentation is a barrier to patient referral.19

Promotion of the pharmacy initiative to eligible patients was agreed between NHS England and a steering group from London Pharmaceutical Committee (LPC). Promotion was predominantly by posters displayed in pharmacy windows, most of which were designed by the LPC, but some pharmacies used general pharmaceutical company vaccine campaign posters. Complementary outreach was achieved by other methods including website promotion (http://www.myhealthlondon.nhs.uk), twitter, photo shoots (eg, deputy Mayor of London receiving flu vaccine) and promotion via NHS trust occupational health departments.

Our results suggest that if the reported time use data in the surveys are accurate and the purchase price of vaccines is at list price, then London pharmacists would have needed to recoup £5.51 in order to break even on flu vaccine provision. There are three reasons how this shortfall is either recouped or non-existent. First, the actual time spent administering the vaccine or completing the data input may be less than reported. For example, under the assumption that only 1 min per dose is spent inputting data, the amount needing to be recouped by pharmacists drops by £2.18–£3.33 (table 3). As pharmacies adapt to the administration system, the duration of their patient consultation will likely decrease. For example, reducing the total time for vaccine administration and data input by 40% would allow pharmacies to break even. Moreover, the time required by pharmacists to administer and complete data entry per dose seems high relative to flu vaccinations in clinics,20 suggesting that bottom-up costing is likely to overestimate the costs of vaccine delivery. Overestimation of time spent administering vaccination or recording administration may arise due to the perception among pharmacists that our survey may influence remuneration decisions. Second, in practice, pharmacists will likely not pay the list price of their chosen vaccine, instead negotiating directly with the manufacturers for the best deal. For instance, if pharmacists wanted to minimise their expenditure based solely on the list price, they would choose Influvac (Abbott) retailing at £5.22 per dose, leaving a profit after reimbursement of £0.68. Thus, as many other brands were chosen, the negotiated price for pharmacy stock must provide more than a £0.68 per dose advantage. Third, pharmacists will hope to generate income by the increased footfall in their pharmacy. Even if the total cost is not recouped, the pharmacy may be willing to pay for the benefit of developing stronger patient confidence in their healthcare provision. We did not attribute any additional recurrent or investment costs of the GP delivery programme above the flu vaccine service fee and the purchase reimbursement price. Hence, the costs of the GP programme is likely underestimated as we did not include the costs of any flu communication material provided to GPs and the costs of managing vaccine data supplied by GPs. However, the cost of the pharmacy option may also be underestimated as it does not include additional costs for GP practices to input data from pharmacists, or contact patients who have already received the vaccine but are not recorded on the system. This latter cost may be relatively high due to the level of under-reporting of pharmacy uptake on the GP ImmForm system. The majority of vaccine doses in GP practices are administered by nurses or healthcare assistants, and as the salary of a nurse in a GP practice is only two-thirds of a pharmacist10 and even less for a healthcare assistant, and most of the vaccine delivery cost results from healthcare staff time, one would expect personnel costs for GP-based vaccine administration to be less costly than those for pharmacy-based vaccine administration.

Many GPs currently offer a call reminder service for the 65 years and over and a call and recall reminder service for other risk groups. If the costs for this reminder service are recouped in the service fee reimbursed to GPs per patient vaccinated, then decreasing the number of reimbursed service fees but maintaining the full call service for all eligible patients increases GP practice costs per vaccine dose. As GPs are paid for the vaccination coverage they achieve (irrespective of whether the practice delivered the vaccines themselves) under the Quality and Outcomes Framework (QOF), GP practices have an incentive to report complete uptake figures via ImmForm. However, there is no similar incentive to ensure complete reporting of pharmacy-delivered vaccine doses to GP practices. Moving to a single recording database system that is accessible to both pharmacists and GPs would align the incentives of recording by pharmacies for reimbursement with recording for complete patient health records. In the London pharmacy initiative, GPs and pharmacists received vaccine purchase reimbursement from the NHS differently; while GPs receive the list price for whatever vaccine stock they use, pharmacists receive a fixed fee regardless of the brand chosen. This reimbursement structure provides different incentives for vaccine brand choice. Leaving aside the health-related preferences of the healthcare provider, GPs would financially benefit from achieving a difference between the list vaccine price and the negotiated vaccine price. Pharmacists would also benefit from a large difference between the list vaccine price and the negotiated vaccine price, but any profit they receive would ultimately be contingent on the negotiated price to reimbursement fee difference. Therefore, with a fixed fee reimbursement contract, pharmacists will have less incentive to offer the more expensive quadrivalent vaccine (Fluarix Tetra). Indeed, across London, <1% of pharmacists offered the Fluarix Tetra compared with 5.5% of GPs. Previous proposals by the Department of Health have sought to procure flu vaccines centrally to streamline the vaccine programme and reduce vaccine purchase costs have been met with resistance by GPs.21 The current national pharmacy programme is equally unpopular with GPs,6 who are concerned with loss of income. Commissioning any re-evaluated vaccine programme is subject to various trade-offs: most notably maintaining incentives for vaccine provision, keeping vaccine purchase and reimbursement costs low, and ensuring high quality and complete reporting. Further, it is possible that there will be emerging issues when scaling up the pharmacy programme to a national level: more premises offering flu vaccines—such that the number of doses offered per premises decreases, coupled with the potentially opportunistic nature of pharmacy vaccination will likely result in more wasted flu doses per box compared with a limited number of providers, with a stable number of patients.

The ‘Community Pharmacy Seasonal Influenza Vaccination Advanced Service’, allowing pharmacies to administer flu vaccines, has been introduced in autumn/winter 2015.22 The programme differs from the pan-London initiative in a number of key ways: first, pharmacists are reimbursed for vaccine purchase in the same way as GPs, such that they claim for the list price (plus VAT) of the vaccine they choose to administer; second, they receive the same service payment as GPs of £7.64 per dose; third, pharmacists are paid an additional £1.50 per dose to cover training, revalidation and waste disposal.23

This study is subject to several limitations that must be considered before generalising the results countrywide. First, the coverage and survey data were London-specific, and these may not be representative of the country. Second, the survey response rate was very low, particularly for pharmacists. Such a low response may be indicative of a biased sample, and more importantly, not provide a large enough sample size for precise estimates of costs, which are calculated from self-reported activity durations. To achieve more accurate estimates of costs from all perspectives, it may be beneficial to conduct a larger, nationwide time use survey. Third, survey responses and coverage data are from 1 or 2 years, respectively. While more data may provide a more precise estimate of outcomes, it is also worth noting that a longer period of pharmacy-based vaccine delivery may increase the awareness of such a programme, and therefore ultimately vaccine coverage may increase over time. Furthermore, the analysis did not account for any underlying secular trends in the uptake of flu vaccination that may obscure any increase due to the pharmacy initiative. Fourth, the survey did not question patients and, notwithstanding the previous patient survey in London, there may be some interesting aspects of pharmacy delivery that have not been reported. For example, our GP survey found that GPs were concerned about the loss of healthcare opportunity for their patients, but given that nurses are more likely to administer flu vaccines in GP surgeries, and that pharmacists were probably more likely to have more time to administer the jabs, this may be a concern not echoed for patients. Finally, surveys were carried out retrospectively, and thus pharmacists may be prone to misreport activity durations on time use surveys. A more accurate survey would be conducted through the vaccination season in real time.

In summary, our findings indicate that the total cost per dose of seasonal flu vaccination is slightly lower when it is administered by pharmacies compared with GPs. However, we found no evidence of increased vaccine coverage as a result of the pharmacy initiative. Further, our results suggest that maintaining two reporting systems for GPs and pharmacies leads to substantial under-reporting of vaccine uptake and unreliable GP vaccine uptake data.