Article Text

Download PDFPDF

Association of socioeconomic deprivation with opioid prescribing in primary care in England: a spatial analysis
  1. Magdalena Nowakowska1,2,3,
  2. Salwa S Zghebi1,4,
  3. Rosa Perisi1,2,
  4. Li-Chia Chen5,
  5. Darren M Ashcroft1,3,5,
  6. Evangelos Kontopantelis1,2
  1. 1 NIHR School for Primary Care Research; Centre for Primary Care and Health Services Research; Manchester Academic Health Science Centre (MAHSC), The University of Manchester, Manchester, UK
  2. 2 Division of Informatics, Imaging and Data Sciences; School of Health Sciences; Faculty of Biology, Medicine and Health; Manchester Academic Health Science Centre (MAHSC), The University of Manchester, Manchester, UK
  3. 3 NIHR Greater Manchester Patient Safety Translational Research Centre, The University of Manchester, Manchester, UK
  4. 4 Division of Population Health, Health Services Research and Primary Care; School of Health Sciences; Faculty of Biology, Medicine and Health; Manchester Academic Health Science Centre (MAHSC), The University of Manchester, Manchester, UK
  5. 5 Centre for Pharmacoepidemiology and Drug Safety; Division of Pharmacy and Optometry; School of Health Sciences; Faculty of Biology, Medicine and Health, Manchester Academic Health Science Centre (MAHSC), The University of Manchester, Manchester, UK
  1. Correspondence to Magdalena Nowakowska, NIHR Greater Manchester Patient Safety Translational Research Centre, The University of Manchester, Suite 5, 5th Floor, Williamson Building, 176 Oxford Road, Manchester M13 9PL, UK; Magdalena.nowakowska{at}


Background The increasing trends in opioid prescribing and opioid-related deaths in England are concerning. A greater understanding of the association of deprivation with opioid prescribing is needed to guide policy responses and interventions.

Methods The 2018/2019 English national primary care prescribing data were analysed spatially. Prescribing of opioids in general practice was quantified by defined daily doses (DDD) and attributed to 32 844 lower layer super output areas (LSOAs), the geographical units representing ~1500 people. Linear regression was used to model the effect of socioeconomic deprivation (quintiles) on opioid prescribing while accounting for population demographics and the prevalence of specific health conditions. Adjusted DDD estimates were compared at each deprivation level within higher organisational areas (Clinical Commissioning Groups, CCGs).

Results In total, 624 411 164 DDDs of opioids were prescribed. LSOA-level prescribing varied between 1.7 and 121.04 DDD/1000 population/day. Prescribing in the most deprived areas in the North of England was 1.2 times higher than the national average for areas with similar deprivation levels and 3.3 times higher than the most deprived areas in London. Prescribing in the most deprived areas was on average 9.70 DDD/1000 people/day (95% CI 9.41 to 10.00) higher than the least deprived areas. Deprivation-driven disparities varied between individual CCGs. In the most unequal CCG, prescribing in the most deprived areas was twice that in the least deprived areas.

Conclusion Opioid prescribing varied substantially across England and deprivation was strongly associated with prescribing. This paper provides evidence for guiding policy interventions and allocation of resources to areas with the highest levels of opioid prescribing.

  • Prescribing
  • deprivation
  • spatial analysis
  • primary care
  • socio-economic

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Opioids are regarded as one of the most effective means to control acute pain. However, their effectiveness for managing chronic non-cancer pain is limited and has associated risks of dependence, overdose and addiction.1 Globally, there were an estimated 53.4 million users of opioids in 2017 and use of opioids accounted for 66% of the drug use disorder related deaths.2 Between 2008 and 2018, opioid prescribing in England increased by 34%3 and 231 151 477 prescription items were prescribed in English primary care in 2018/2019.4

Existing evidence suggests that the level of opioid prescribing differs across individuals and communities.2 5 In the USA, opioid prescribing was higher in regions with a larger population, lower socioeconomic status and higher availability of physicians, when accounting for differences in the prevalence of injuries, surgical procedures or conditions requiring analgesics.6 Similar associations between opioid prescribing and markers of socioeconomic deprivation were found in Canada,7 Australia8 and Europe.3 9–12 In England, higher levels of prescribing were found in the North,3 9 10 mirroring the well-established North-South inequalities.13

People living in more deprived areas are more likely to experience non-cancer pain,14 have more complex health needs15 and suffer adverse effects of opioid use.16 Several interventions have been developed to improve opioid prescribing practices and improve patient outcomes with mixed levels of success.17 In order to address the issues effectively and efficiently, a greater understanding of the distribution of opioid prescribing, while accounting for variations within the health services' administrative bodies, is needed. Locating areas of higher risk can help ensure that resources are in place to deal with increased demand for services such as opioid dependence treatment and hospital admissions due to opioid overdose.

Nationwide studies of opioid prescribing in English primary care have focused on prescribing at the general practice (GP) or Clinical Commissioning Group (CCG) level.3 9 10 CCGs are administrative bodies responsible for delivering health services in England. They cover between ~100 000 to over 500 000 people with communities varying in characteristics; therefore, spatial variation in opioid prescribing within CCGs is expected. Analyses at the CCG level showed that deprivation was correlated with increased opioid prescribing.9 At the GP level, increased levels of prescribing were associated with the number of registered patients, rurality and deprivation.3 However, a comparison of four major cities in England showed that the relationship between opioid prescribing patterns and their predictors, including deprivation, differs between regions.9

The region-specific relationships between opioid prescribing, and its potential predictors, like deprivation, are unknown. The aim of this study was to estimate opioid prescribing at the lower layer super output area (LSOA) level, which represents ~1500 people. In addition, we investigated the variation of prescribing within and between CCGs. Furthermore, we aimed to explore the extent to which disparities in opioid prescribing are associated with population demographics and socioeconomic deprivation. We compared the level of opioid prescribing in areas with the same deprivation level located in different CCGs while controlling for other demographic factors. LSOA is the lowest available measure at which deprivation is measured in England; therefore, a more accurate representation of population deprivation and an analysis at that level can be more informative about the relationship between deprivation and opioid prescribing, than analyses at CCG or practice level.


Study design and data source

This cross-sectional study examined opioid prescribing in the financial year 2018/2019 (1 April 2018 to 31 March 2019) in English primary care. We extracted data on medications prescribed in a GP setting and dispensed during the study period from the monthly NHS Digital Practice Level Prescribing dataset containing details on medicines prescribed in England and dispensed in the community in the UK.18 Data on patients registered with a GP and their home address LSOA come from NHS Digital. Data on population demographics, deprivation and rurality of LSOAs come from the Office for National Statistics (ONS). We extracted data on the prevalence of selected conditions from the 2018/2019 Quality and Outcome Framework scheme data.19 A detailed explanation of the data attribution and all data handling processes is provided in online supplemental material 1.

Supplemental material

Description of the geographical level

We used LSOAs as the geographical unit of analysis. Established following the 2011 census, there are currently 32 844 LSOAs in England, each representing ~1500 people. We analysed LSOA-level data and compared it across the country and within the 195 English CCGs. To attribute the GP level data to LSOAs, we used data on the number of patients registered with the practice to their home address LSOA which we adjusted to reflect the ONS LSOA population data estimates.

Prescription measure

We extracted prescriptions for drugs included in the British National Formulary (BNF) chapter We excluded prescriptions for methadone and sublingual buprenorphine 2 mg and 4 mg as they are primarily used in opioid dependence treatment. Prescriptions for co-codamol containing 8 mg codeine were excluded as it is primarily used for acute pain management and can be purchased without a prescription. For each prescription, we converted the strength of the opioid into defined daily doses (DDDs) using the WHO definition.21 For each area, the outcome measure, DDD/1000 people/day, was calculated as the sum of DDDs attributed to the area, divided by the total number of residents, multiplied by 1000, and divided by 365.

Socioeconomic deprivation measure

Socioeconomic deprivation was measured using the LSOA-level 2019 Index of Multiple Deprivation (IMD) data. IMD assigns each LSOA a score derived from seven weighted indices: income, employment, education, health, crime, barriers to housing and services, and living environment. LSOAs are then ranked based on the total score. For this analysis, we used IMD quintiles to measure the socioeconomic deprivation of the LSOA, from least (IMD 1) to most (IMD 5) deprived.

Disease prevalence and demographics measures

We extracted practice-level data on the prevalence of conditions previously linked to9 or expected to drive opioid prescribing (cancer, palliative care, depression, severe mental illness (SMI), rheumatoid arthritis, and obesity). We attributed these data to LSOA using the same method as for prescribing data. Additional information at the LSOA-level was obtained from ONS data collections: rurality/urbanity classification, the proportion of females, and the proportion over 65 years old.


We used spatial analysis to explore the geographical distribution of opioid prescribing across England. Firstly, we present the LSOA-level of opioid prescribing across the country with choropleth maps. We compared the average LSOA-level prescribing across deprivation quintiles in England and within five main regions of the country: London, South East, South West, Midlands and East of England, and North of England. To determine whether LSOAs with similar levels of prescribing tend to be clustered next to each other, we measured autocorrelation using Moran’s I statistic. For this study, we determined a space limit of 100 km radius from LSOA population-weighted centres beyond which we assumed there is no association between LSOAs. The relationship between these LSOAs was weighted based on the row standardised inverted distance matrix. To assess the statistical significance of Moran’s I, we calculated p values using a Monte Carlo approach with 999 permutations.

We investigated the predictors of high prescribing and the association between deprivation and LSOA-level opioid prescribing within different CCGs using linear regression. We developed three models. Model 1 included the following predictor variables: prevalence of the seven selected conditions, proportion of females, proportion of the population over 65 years old, LSOA size, LSOA rurality, and deprivation quantile. Model 2 included all predictor variables from model 1 and a categorical variable for the CCG to which the LSOA belonged. To assess the importance of CCG in predicting levels of opioid prescribing we compared estimated R2 values across models. We estimated the effects of the predictor variables using model 2. To identify areas with significantly high or low prescribing levels, which cannot be explained by the variables included in the model, we mapped the residuals from this model. We estimated the spatial autocorrelation for the residuals using Moran’s I to establish if geographical patterns exist in the unexplained prescribing levels. Model 3 included an interaction term between the categorical variable CCG and IMD quintile in addition to variables included in model 2. We used the model 3 and the margins command in Stata to estimate the predicted average value of opioid prescribing at each IMD level within each CCG while controlling for all other variables. The difference between the estimated average prescribing in areas of each deprivation level was calculated to assess the level of disparities within a CCG. We present the estimated difference between the most and least deprived areas in the article. Comparisons of the least deprived areas and all remaining IMD levels are presented in online supplemental material.

We conducted all analyses in R v.3.6.022 and Stata 15.23


In total, 624 411 164 DDDs of opioids were prescribed in England in the financial year 2018/2019 with a national average of 30.61 DDD/1000 people/day. The mean (SD) size of an LSOA was 1701 (419) patients with 82.9% of LSOAs classified as urban. CCG covered on average 168.4 LSOAs (min=41, max=681). Descriptive statistics are presented in the online supplemental material 1 table SM1.

Geographical distribution of opioid prescribing

Within LSOAs, the average level of prescribing was 31.28 DDD/1000 people/day (14.92). On a national level, prescribing in the most deprived areas was on average 1.7 times higher than in the least deprived areas (figure 1). Prescribing in the least deprived areas in the North of England (median 32.44, lower quartile (LQ) 26.51, upper quartile (UQ) 39.75) was 1.4 times higher than the national average for the least deprived areas (23.43, LQ 17.43, UQ 30.90) and 2.8 times higher than in the least deprived areas in London (11.38, LQ 8.45, UQ 14.58). The highest levels of prescribing were found in the most deprived areas in the North of England (47.96, LQ 38.54, UQ 57.45) which was 1.2 times higher than the national average for the most deprived areas (40.36, LQ 28.72, UQ 51.82) and 3.3 times higher than the most deprived areas in London (14.42, LQ 11.82, UQ 17.23) (figure 1).

Figure 1

Opioid prescribing at defined daily doses per 1000 people per day at the lower layer super output area level stratified by region and deprivation quintile. DDD, defined daily doses; LSOA, lower layer super output area; IMD, Index of Multiple Deprivation.

LSOA-level prescribing varied between 1.17 and 121.04 DDD/1000 people/day. The between-LSOA variations of prescribing levels differed between CCGs (figure 2). The mean prescribing in the most uniform CCG was 28.73 (1.16) DDD/1000 people/day and in the most varying CCG it was 69.28 (21.80) DDD/1000 people/day.

Figure 2

Lower layer super output area-level opioid prescribing variation within and between Clinical Commissioning Groups (CCG). The red line indicates the national average for LSOA-level opioid prescribing. The dark blue dots indicate the within CCG average LSOA-level opioid prescribing. The grey boxes indicate the within CCG first and third quartiles (the 25th and 75th percentiles) of LSOA-level opioid prescribing. The light blue whiskers represent the highest and smallest value no further than 1.5*IQR from the hinge. The light blue dots represent the within CCG outliers. CCG, Clinical Commissioning Group; DDD, defined daily doses; LSOA, lower layer super output area.

Prescribing was not distributed randomly across space with Moran’s I of 0.57 (p≤0.001), suggesting that prescribing at the LSOAs level is moderately clustered. Higher prescribing LSOAs were located in the Midlands, the North, Cornwall (South West of England) and the coastal areas (figure 3).

Figure 3

Distribution of lower layer super output area level opioid prescribing in financial year 2018/19 in defined daily doses per 1000 people per day. The grey lines represent borders of Clinical Commissioning Groups.

Predictors of high prescribing

The linear regression model 1 explained 50% of the variation in prescribing levels which increased to 79% for model 2 (table 1). The prevalence of cancer (−0.78, 95% CI −0.98 to −0.58) and SMI (−8.98, 95% CI −9.45 to −8.52) were negatively associated with opioid prescribing. The prevalence of rheumatoid arthritis had a strong positive effect on prescribing (4.08, 95% CI 3.23 to 4.93), whereas the remaining conditions had a small but positive association. The proportion of females, people >65 years old, and LSOA size had a small effect on prescribing. Prescribing in urban areas was on average 1.85 DDD/1000 people/day (95% CI 1.64 to 2.08) higher than in rural areas. A positive association was observed between socioeconomic deprivation and opioid prescribing. Prescribing in the most deprived areas (IMD 5) was on average 9.70 DDD/1000 people/day (95% CI 5.97 to 6.5) higher compared to the least deprived areas (IMD 1).

Table 1

Linear regression modelling opioid prescribing measured in DDDs/1000 people/day at an LSOA level

Supplemental material

Supplemental material

Analysis of the residuals of model 2 showed that there were no obvious geographical patterns that were not captured in the analyses, with very low Moran’s I of 0.09 (p≤0.001). A significant difference in the residuals for LSOAs within and between CCGs suggests marked variation in the levels of prescribing which cannot be explained by the predictor variables included in the model (online supplemental material 1 figure SM2).

Based on model 3, we calculated the estimated average LSOA-level prescribing per each IMD level, adjusted for all other variables. The estimated averages for all CCG and IMD levels are available in online supplemental material 1 figure SM3. In some CCGs, the difference between the most and the least deprived areas was unavailable as there were no LSOA with a specific IMD level within that CCG. In five CCGs, the estimated prescribing in the least deprived areas was higher than in most deprived areas (figure 4). For example, in North Lincolnshire CCG the average prescribing in the least deprived areas (number of LSOAs in CCG=13) was 4.05 DDD/1000 people/day higher than in the most deprived areas (n=13) (46.22 DDD/1000 people/day in IMD 1 vs 42.17 DDD/1000 people/day in IMD 5) (online supplemental material 4). The highest difference in absolute numbers was estimated for Lincolnshire East CCG where prescribing in the most deprived areas (n=33) was estimated as 86.59 DDD/1000 people/day. In contrast, in the least deprived areas (n=6) it was estimated as 43.15 DDD/1000 people/day.

Figure 4

Estimated difference in opioid prescribing between the most and least deprived areas in English Clinical Commissioning Groups (CCGs) in the financial year 2018/19. In each CCG, the difference was estimated by subtracting the estimated average opioid prescribing, adjusted for demographic variations, in lower layer super output areas from the most deprived quintile from the least deprived areas. Opioid prescribing was measured in defined daily doses per 1000 people per day.

Supplemental material



Our study provides evidence that levels of opioid prescribing differ significantly across small geographical areas in England, even after adjusting for population structure. Differences between LSOAs occurred across the country and within individual CCGs. Striking differences between prescribing in the North of England and London were visible even when comparing areas with the same IMD quantile. At a national level, deprivation was a significant predictor of LSOA-level opioid prescribing. However, the inequalities between the most and least deprived areas differed significantly between CCGs.

Comparison with existing literature

To our knowledge, this is the first English study of the geographical variations in opioid prescribing at such a small geographical level. Previous studies analysed variations in CCG-level opioid prescribing. An analysis of 2015 prescribing data found average prescribing levels at 36.9 DDD/1000 registrants/day, slightly higher than our estimated average of 31.28 DDD/1000 people/day.9 This may reflect the previously suggested decrease in opioid prescribing after 2015.3 However, the exact level of change is likely to be higher due to methodological differences and our correction for over-registration. Highest prevalence of prescribing in the north of the country mirrors the wider inequalities enrooted in the previously documented economic divide between the North and the South.13 As previous research pointed out, London’s population is significantly different from the rest of the country.24 High numbers of healthy, young workers and increased social mobility affect the health needs and, consequently, demand and supply of health services of the local population. Further research may enhance our understanding of how the relationship between environmental factors and opioid prescribing may demonstrate differently in London.

On a national level, we found prescribing in the most deprived areas to be 1.7 times higher than in the least deprived areas. Similarly, a recent report by Public Health England found that the proportion of the population registered with GPs in England receiving a prescription for opioids was ~9% in the least deprived compared to ~15% in the most deprived quantile.5 Associations between socioeconomic deprivation and opioid prescribing were also found in other countries. A study conducted in Australia found that in 2015 the average amount of prescription opioids used was 20.28 DDD/1000 population/day in the highest disadvantage areas compared to 9.87 DDD/1000 population/day in the low disadvantage areas.8

The effect of socioeconomic deprivation differed between regions and CCGs. In the Midlands and the East of England prescribing in the most deprived areas was ~48% higher than in the least deprived areas, whereas the difference in London was ~27%. Within CCGs, variations in opioid prescribing by deprivation levels varied from higher prescribing observed in the most deprived areas to over 40 additional DDD/1000 people/day prescribed in the most deprived areas compared to the least deprived areas. Communities with comparable measures of socioeconomic deprivation have previously been shown to experience different health outcomes.25 The effect of deprivation also differed when GP-level opioid prescribing was compared in four English urban areas.9 In general, a higher level of opioid prescribing is expected in more deprived areas due to the higher prevalence of pain14 and more people receiving long-term opioid treatment.26 However, the difference between the most and least deprived areas was not consistent across the country. Differences between areas of the same level of deprivation could be linked to variation in individual factors such pain coping abilities27 and levels of health literacy,28 as well as contextual factors such as income inequality and group diversity,29 natural environment, social capital,30 and cultural stigma around the treatment of chronic pain.31 Although IMD is a routinely used measure of socioeconomic deprivation, it may not capture all relevant aspects of living in deprived areas that may be important in understanding supply and demand for opioids. This may explain why in a small number of CCGs, we found higher prescribing in the more affluent areas compared to those more deprived. Qualitative research, exploring the lived experiences of socioeconomic hardship in different areas of the country, may enhance our understanding of the association between opioid prescribing and deprivation. Due to the lack of geographically representative patient level data, we were unable to account for specific opioid prescribing patterns, such as long term and high dose prescribing, which could help us further understand the variations found in this study.

In light of the high variation between areas of the same levels of measured deprivation, future research could consider greater emphasis on understanding the community specific experience of deprivation and its effects on opioid prescribing. Such understanding could help policymakers develop interventions that address the specific needs of a community and can create lasting conditions in which both the demand and supply of opioids can be optimised. CCGs with high prescribing levels across the areas may benefit from wider structural changes such as improving availability and accessibility of specialised pain treatment services which have been found to vary across England.32 CCGs with elevated prescribing levels observed in some but not all areas could benefit from practice and community based interventions which could lead to a reduction in opioid use and improvement in health outcomes of people treated with opioids.33 In the face of no one-size-fits-all intervention successful at improving opioid prescribing patterns,17 evidence provided in this study can guide policy designers in ensuring that appropriate resources are allocated where they can bring the most benefits.

Strengths and limitations

This study used national data for opioids prescribed in England to explore the association between prescribing and deprivation at a low geographical level.

Some limitations need to be acknowledged. Firstly, prescription data were attributed to LSOAs uniformly, weighted on the number of patients registered with the practice living in a particular LSOA. It was not possible to assess the actual number of prescriptions that were distributed to people living in an LSOA. It is likely that prescriptions are not distributed uniformly across all patients, and a small proportion of patients receive a high proportion of all opioid prescriptions. Secondly, we were unable to assess whether in certain regions opioids are being prescribed at a higher dose or duration. Thirdly, we used DDDs as a standardised measure of opioid prescriptions. By doing so, we were able to aggregate all opioid prescriptions while taking into account their potency; however, the actual prescribing practices may differ to those recommended.34

Implications and conclusion

Understanding the spatial distribution of opioid prescribing may provide important guidance for predicting demand and planning services. The results of this study suggest that the relationship between deprivation and opioid prescribing is complex and varies across the country. Our analysis shows the importance of examining opioid prescribing at a low-geography level and comparing areas with different socioeconomic and demographic characteristics. Interventions may be needed to tackle the deprivation driven disparities in opioid prescribing. These should be designed to meet the specific need of the communities, including understanding and addressing the underlying causes of high prescribing. Further analysis of the relationship between socioeconomic deprivation and opioid prescribing, including the social, cultural and medical mediators, which may explain how deprivation leads to increased opioid prescribing, may provide valuable evidence for policymakers and clinicians to improve opioid-related health outcomes.

What is already known on this subject

The prevalence of opioid prescribing in English primary care is not equal across communities and variations between large geographical units have been observed. Previous research suggests that socioeconomic deprivation may be associated with these variations; however, research on a smaller geographical level is needed to improve understanding of locations with highest opioid prescribing and the role of socioeconomic deprivation.

What this study adds

Regional analysis showed that areas with the highest levels of deprivation in the North of England had 1.2 times higher prescribing than the national average for areas with similar deprivation levels and 3.3 times higher prescribing than most deprived areas in London. Although deprivation is positively associated with opioid prescribing on a national level, the strength of this relationship varies between Clinical Commissioning Groups (CCGs), with some showing higher levels of inequalities than others. After adjusting for demographical differences, in some CCGs opioid prescribing was higher in the more affluent areas compared to the most deprived ones, whereas in others, on average, 44.3 more defined daily doses per 1000 people per day were prescribed in the most deprived compared to the least deprived areas. This study provides evidence for guiding policy development and implementation to optimise opioid prescribing and delivering support in communities most affected by elevated levels of opioid prescribing. Furthermore, it highlights that some geographical areas experience vastly different levels of opioid prescribing even with similar deprivation levels. This suggests that the relationship between socioeconomic deprivation, as measured by Index of Multiple Deprivation, and opioid prescribing is not uniform across the country. The geographical mapping of areas experiencing the highest levels of prescribing may guide policymakers in the allocation of resources, such as additional training and pain management services, and inform clinicians of prescribing practices in their local areas. Furthermore, it provides a foundation for further research into the complex relationship between opioid supply, demand, and the socioeconomic environment.



  • Twitter Evangelos Kontopantelis @dataevan.

  • Contributors All authors contributed to the design of the study. MN prepared and analysed the data. MN wrote the manuscript and MN, SSZ, RP, LCC, DMA and EK all critically edited the manuscript. All authors approved the manuscript before submission.

  • Funding This study was funded as part of a PhD studentship from the National Institute for Health Research (NIHR) School for Primary Care Research (SPCR) and NIHR Greater Manchester Patient Safety Translational Research Centre. The study represents independent research by the NIHR. The views expressed in this publication are those of the authors and not necessarily those of the National Health Service, the NIHR or the Department of Health and Social Care. The study funders had no role in the study design, data collection, analysis or interpretation, in the writing of the paper or in the decision to submit the paper for publication. MN had full access to all the data in the study and had final responsibility for the decision to submit for publication.

  • Map disclaimer The depiction of boundaries on the map(s) in this article does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. The map(s) are provided without any warranty of any kind, either express or implied.

  • Competing interests LCC and DMA have received funding for another study of opioid utilisation patterns from Mundipharma Research Ltd. MN, SSZ, RP, and EK have no conflict of interest to declare.

  • Patient consent for publication Not required.

  • Ethics approval This study involves analysis of aggregated, anonymous and publicly available data. No ethical approval was sought or required for this study.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement All data used in this study are publicly available from the sources cited in the manuscript.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Linked Articles