Data were extracted from the monthly prescribing information published by the NHS Business Service Authority, under the Open Government Licence.

This was a retrospective, observational study set in England. The data were from all NHS GP practices in England between 1 January 2015 and 30 September 2024.

We included the two main forms of vitamin B₁₂, hydroxocobalamin (BNF code 0901020N0)19 and cyanocobalamin (BNF code 0901020D0).20 Only those two forms were available via prescriptions from the NHS. Hydroxocobalamin is an injectable form.2 10 Cyanocobalamin comes mainly in oral formulations, although it can also come in an injectable form. We extracted prescription data separating injectables and tablets. The list of tablet formulations can be found in the online supplemental table S1 and the list of injectable formulations can be found in the online supplemental table S2.

We presented the outcomes in terms of crude prescription counts and rates per 100 000 registered patients. We also presented the average quantity of units per prescription (ampoules or tablets) over time. The number of registered patients per year (to serve as the denominator for yearly rates) was calculated from the monthly numbers of patients by taking an average. It was 57 117 376, 57 674 187, 58 473 838, 59 194 531, 59 908 362, 60 433 996, 60 937 210, 61 771 877, 62 586 477 and 63 302 856 for years 2015–2024, respectively. Rates were used to account for changes in the population sizes over time and the differences between regions. Rates were used also because they allow national and international comparisons between published studies.

The observed monthly prescription counts, rates and units per prescription were visualised in longitudinal plots from January 2015 to September 2024. We stratified the analysis by seven regions in England: Midlands, North East and Yorkshire, North West, South East, South West, East of England and London.21 To estimate the changes in prescribing over time and the effect of the COVID-19 pandemic, we compared prepandemic data from the years 2015 to 2019 to the years from 2020 to 2024. We used linear regression to model the trends between 2015 and 2019 and to predict the expected trends between 2020 and 2024. The smoothed trends in the regional prescribing were calculated using the 3-point moving average method. Yearly counts and rates were summarised using descriptive statistics. For 2024, we held data ending on 30 September 2024. Therefore, to enable comparisons of 2024 with previous years, we estimated the data for this year pro-rata by dividing the total count by 9 and multiplying it by 12.

To process data and undertake statistical analysis, we used R: A Language and Environment for Statistical Computing, Core Team V.4.3.2 (2023), R Foundation for Statistical Computing, Vienna, Austria (https://www.R-project.org/). The function plot_time_series in the package timetk was used for graphics. We made the data set available for download from an Open Science Framework repository.22

No formal patient and public involvement activities were undertaken as part of this study.

We observed rapid changes in counts and rates of prescriptions starting in early 2020 for both injections and tablets (figure 1). In 2020, there were 571 393 (21%) fewer prescriptions for injections than in 2019 (table 1). Using modelled data (taking into account the longitudinal trends), we showed that in 2020 there were 806 031 (27%) fewer prescriptions for injections than there would have been if the pandemic had not occurred. The difference was statistically significant, the observed count in 2020 was 2 171 924 and the predicted was 2 977 956 (95% CI: 2 905 348 to 3 050 565). Although the deficit between observed and predicted prescriptions for injections was getting smaller over time, it remained significant until the end of the study period. In 2024, the difference was 499 845 (13%), 3 194 952 observed and 3 694 797 predicted (95% CI: 3 541 711 to 3 847 884). Therefore, reaching over 3 million in 2024, the number of prescriptions for injections exceeded the prepandemic level of 2.7 million in 2019. However, based on the prepandemic trends, it was still below the predicted count by half a million prescriptions per year.

Correspondingly, we observed 424 963 (43%) more prescriptions for tablets in 2020 than in 2019 (table 2). The predicted difference for 2020 was 299 834 (27%), 1 415 315 observed versus 1 115 481 predicted (95% CI: 1 094 350 to 1 136 612). This trend continued in the following years, reaching over half a million more than expected prescriptions for tablets in 2024. The observed prescriptions count in 2024 was 2 230 729 versus 1 509 901 predicted (95% CI: 1 465 348 to 1 554 454). By 2024, there were over 2 million prescriptions issued for tablets, double the count of 2019.

The changes in prescription counts were reflected in rates. The initial drop in 2020 in prescriptions for injections, by nearly 1000 prescriptions per 100 000 patients, levelled up by 2024 to the prepandemic rates. The observed rate in 2024 was 5047, exceeding the 2019 rate of 4579, but representing a predicted deficit of 926 (16%) prescriptions per 100 000 patients as estimated from the predicted rate of 5973 (95% CI: 5705 to 6241). In contrast, for tablets, in 2020, there were 488 (26%) more prescriptions per 100 000 than expected. This increased in 2024 to 1059 (43%) more prescriptions than expected. The observed rate in 2024 was 3524 versus 2465 predicted (95% CI: 2386 to 2544). The observed rate for tablets doubled from 1653 in 2019 to 3524 prescriptions per 100 000 people in 2024.

Despite the drop in prescriptions for injections and the increase in prescriptions for tablets, there were more prescriptions for injections issued than tablets every year across the study period. In 2023 and 2024 alike, there were 3 million prescriptions for injections and 2 million prescriptions for tablets. However, the proportion shifted over time towards tablets. In 2019, there were three times more prescriptions for injections than tablets, nearly 3 million (at a rate of over 4579 prescriptions per 100 000 patients) versus 1 million (rate 1653). By 2024, this proportion decreased to one and a half more prescriptions for injections than tablets. In 2024, there were over 3 million prescriptions for injections and over 2 million for tablets.

Figure 1 shows that the average quantity of ampoules was steadily decreasing across the study period from around 1.6 to 1.4 ampoules per prescription. This was similar for tablets, the average quantity of tablets per prescription decreased from 55 in 2015 to 48 in 2024. The trend for injections was not affected by the pandemic. However, we observed a temporary increase in the number of tablets prescribed in 2020, with a peak in March 2020 at 58 tablets per prescription.

The regional trends resembled the national trends. The sudden drop in prescriptions for injections as well as the sudden increase in prescriptions for tablets was observed in 2020 for all regions in England. The trends in monthly rates per 100 000 for injections are presented in figure 2 and for tablets in figure 3. The raw monthly counts of prescriptions for injections are presented in the online supplemental figure S1 and for tablets in the online supplemental figure S2. The steady upward trend in prescriptions for injections, after the initial drop in 2020, was observed in all the regions. However, some regions, for example, the North West, did not recover to the prepandemic levels as quickly as the other regions.

The rates of prescriptions varied between the regions. At the end of the study period, London had the lowest number of prescriptions issued for both forms. In London, there were around 19 000 prescriptions issued per month for injections at the rate of around 180 per 100 000 patients and 18 000 prescriptions for tablets at the rate of 170 per 100 000 patients. In comparison, in the South West, which was the region with the highest number of prescriptions, there were nearly 35 000 prescriptions for injections per month at the rate of 550 per 100 000 and around 22 000 prescriptions for tablets and the rate was 360 per 100 000 patients.

We observed a sharp dip in prescriptions for injections of vitamin B₁₂ at the beginning of 2020 and a simultaneous rapid increase in tablet usage nationally and across all regions in England. These changes coincided with the outbreak of the COVID-19 pandemic. From 2020 onwards, the prescriptions for both injections and tablets were increasing steadily with injections returning to the pre-2020 levels by the end of the study period and prescriptions for tablets doubling between 2019 and 2024. Despite the changes, there were more prescriptions for injections than tablets issued across the study period. However, the proportion shifted from three times more prescriptions for injections than tablets in 2019 (3 million vs 1 million) to one and a half times more prescriptions for injections than tablets in 2024 (3 million vs 2 million).

This is the first national study documenting patterns in the prescribing of vitamin B₁₂, focusing on a comparison between tablets and injections before and after the pandemic. An important strength of this study was in the size and completeness of the data set. We accessed prescribing data from all GP practices in England achieving 100% national coverage. Data were nearly real-time, so this study presented current trends. We extracted the data in December 2024 showing trends for up to September 2024. We used a transparent and open research approach. Data used in this study were open access and freely available for reproduction and scrutiny of our analyses and findings. The original data are published monthly by the NHS and this facilitates future updates to monitor further changes in prescribing. With the release of the new NICE guidelines focusing on the diagnosis and management of vitamin B₁₂ deficiency (published in March 2024)15 and as we continue learning from the pandemic, it becomes critical to replicate this study in the future.

We also note some limitations. This data set was based in England only. Similar studies from other countries are needed to understand the global perspective. Data were anonymised and only aggregate prescription counts were available. Therefore, the information on patient demographics and disorders treated with B₁₂ supplements was unknown and could not be considered. Furthermore, we could not access information on B₁₂ supplements purchased by patients online or over the counter. This is a potentially growing trend and could be a source of bias leading to underestimated values for tablets and injections.23 Future studies are needed to evaluate this trend. The information on doses prescribed was not available which makes comparison of quantities challenging. We considered all the tablet formulations, including all strengths. Future research investigating switches between strengths of tablets could be of value. Although we used practice list sizes to calculate rates and enable comparisons with other published studies, our approach was limited because the same-size populations may differ in the characteristics of patients. A methodology for adjusting for age and sex in populations exists but this is imperfect and resource-intensive, so not available for this study.24

We used a data set with 100% national coverage in England to show the changes in prescribing patterns from 2015 to 2024. We accessed the number of prescriptions and number of units (tablets and ampoules) per prescription. We showed a shift from vitamin B₁₂ injections towards tablets. More research is needed to evaluate the effect of these changes on symptoms and quality of life of patients, including fatigue, peripheral neuropathy, pain, impaired sleep, cognition and mental health.25 A survey in the UK, by Seage and Semedo, investigated the impact on people with pernicious anaemia (n=329) who experienced a suspension of injections, with tablets being offered as an alternative.26 They reported uncertainty about the long-term effects on health and well-being and dissatisfaction with healthcare. The study reported a need for more support in the management of pernicious anaemia beyond the pandemic.

The Royal College of General Practitioners advised during the COVID-19 pandemic that selected patients should be taught to self-administer B₁₂ injections.18 However, the British Society for Haematology recommended against patients switching to self-administration during the pandemic due to limited opportunities for support and supervision.17 A UK survey (n=1297) by Tyler et al found that 40% of people with B₁₂ deficiency, including pernicious anaemia, self-medicated via injection (this was defined as buying a supply online without a prescription).23 People reported stigma, let down and the lack of a patient-centred approach in primary care. This eroding trust promoted the need to self-medicate to regain quality of life and negatively affected the perception of healthcare. Most importantly, postponing injections for people on maintenance treatment could expose patients to worsening symptoms and in extreme cases irreversible neurological damage that could potentially be avoided by support and self-administration.27 28

Injections remained the most popular option throughout the study period. There were 3 million prescriptions for injections issued in 2019 and in 2024. In comparison, there were 2 million prescriptions for tablets issued in 2024, the most since 2015 and double the amount in 2019. We also showed that on average there were around 1.4 ampoules and 56 tablets per prescription. This is relevant, because for the long-term management of vitamin B₁₂ deficiency, one ampoule is needed every 2–3 months19 and 56 tablets usually provide 1–2 months of treatment.20 At 1.4 ampoules per prescription, we expect approximately three to four prescriptions per year and at 56 tablets, we expect 6–12 prescriptions per year. Therefore, although we did not have the exact dosage information, we concluded that for oral to be the major route we should expect the number of prescriptions to be twice that of the injections.

Inadequately treated B₁₂ deficiency can lead to serious physical and psychological complications. Traditionally, injections have been considered the best option unless there is evidence of dietary deficit. Many of the B₁₂ deficiency conditions, for example, pernicious anaemia, are chronic and require lifelong supplementation. The administration of injections requires more resources than dispensing tablets. Even before the pandemic, there had already been an ongoing discussion about the limited evidence supporting the benefit of injections versus tablets for B₁₂ deficiency.12–15 25 This included inconclusive evidence comparing the safety, tolerability, clinical effectiveness and cost effectiveness of both types of treatment. The COVID-19 pandemic brought a rapid review in prescribing guidelines17 18 and as we show in this study, greatly accelerated the shift from injections to tablets. However, the pandemic-related changes focused on reducing primary care appointments and protecting patients from COVID-19 infections. Therefore, the challenges associated with the limited evidence in relation to patient outcomes remain pertinent. The COVID-19 pandemic should also accelerate the important opportunities for more research in this area.

Not applicable.

Data were publicly available and fully anonymised, no patients were contacted during this work, thus ethical approval was not required.