Examining the burden of opioid prescribing for non-cancer pain considering socio-economic differences, in Wales. A retrospective database study examining trends between 2005 and 2015

Objectives: To use a proxy-measure of oral morphine equivalent dose (OMED) to determine trends in opioid burden in people with non-cancer pain and explore differences related to deprivation status. Design, setting and participants: Retrospective cohort study using cross-sectional and longitudinal trend analyses of opioid prescribing data from 78% of Welsh Primary Care General Practices, whose data is shared with the Secure Anonymised Information Linkage (SAIL) databank. Anonymised data for the period 2005 to 2015, for people aged 18 or over, without a recorded cancer diagnosis and who received at least one prescription for an opioid medicine was included. Primary and Secondary outcomes: A proxy-measure of oral morphine equivalence dose (OMED) was used to describe trends in opioid burden over the study period. OMED burden was stratified by 8 drug groups and deprivation, based on the quintile measures of the Welsh Index of Multiple Deprivation 2011 (WIMD2011). Result: In the 11 years examined, 22 641 424 prescriptions for opioids were issued from 345 primary care general practices in Wales. Daily OMED per 1000 population increased by 94.7% (from 16 266 mg to 31 665 mg). Twenty-eight percent of opioid prescribing occurred in the most deprived quintile. More than 100 000 000mg more OMED was prescribed in the most deprived areas of Wales, compared to the least deprived. Codeine prescribing accounted for 35% of the OMED burden in Wales over the study period. Conclusions: Whilst opioid prescription numbers increased 44% between 2005 and 2015, the OMED burden nearly doubled, with a disproportionate OMED load in the most deprived communities in Wales. Using OMED provides an insightful representation of opioid burden, more so than prescription numbers alone. Socio-economic differences are likely to affect pain presentation, access to support services and increase the likelihood of receiving an opioid prescription.


Introduction
The number of prescriptions for opioid medicines issued in the United Kingdom (UK) has increased substantially over the last twenty years (1)(2)(3)(4)(5)(6). In particular, prescriptions for 'strong' opioids, such as morphine, oxycodone and fentanyl, have seen a higher increase than those classed as 'weak', like codeine and dihydrocodeine (1,2,6). Increases in prescribing persist, even when evidence to support using these medicines for people living with pain that is not acute, cancer or end-of-life related, is largely missing (7)(8)(9)(10)(11).
dataset was formed using one line per opioid prescription, linked to individual patient demographic data using unique Anonymised Linkage Fields (ALFs). The ALF allows cross-linking between different existing datasets, providing a record of all healthcare interactions for any individual whose data is available to SAIL. Primary Care General Practice data was linked to the Welsh Index of Multiple Deprivation 2011 (WIMD2011), based on the Local Super Output Areas (LSOAs) listed on the PCGP records.

Opioid prescriptions
Prescriptions are automatically assigned read codes at issue as a consistent means to identify data of interest (1,3,43,44). Read codes are a thesaurus of clinical terms used to record all interactions, diagnoses and interventions in Primary Care settings in Wales. A list of read codes was compiled for all prescribable oral and transdermal opioid medicines used for analgesia, including combination products, e.g.
paracetamol and codeine (co-codamol), using the NHS Information Authority's Clinical Terminology Browser and accessed via the SAIL secure gateway. Products licensed for the management of misuse and injectable opioids, which are reserved for palliative care, were excluded.
Only data for people aged 18 years or older between 2005 and 2015 without a recorded cancer diagnosis (identified using read codes for cancer diagnoses or treatment) at any time between 2004 and 2015 were included in the analysis.
All data were subjected to repeated cross-sectional sampling to determine prescribing trends over the study period.

Oral morphine equivalent dose
Oral morphine equivalent dose (OMED) is normally calculated by multiplying the dose for each prescription (or a combination of doses, e.g. 10mg + 30mg tablets to give a total of 40mg per dose), by the equi-analgesic ratio of the opioid in question (1). The number of days' supply provided by each prescription is then divided by the numerical daily dose (NDD) taken from the free text on the prescription (e.g. 'one tablet to be taken twice daily'). This way, an OMED can be calculated per prescription or per individual over longer time periods (e.g. annual OMED) (1,2). However, at the time of this study, dispensing data (42) could not be anonymously linked into SAIL datasets. Whilst the prescribed product, including strength, was recorded, details such as directions or quantity prescribed was unavailable to calculate the prescribed oral morphine equivalent dose (OMED) for each individual. Therefore, each opioid product was allocated its OMED value based on conversion tables available for clinical practice (8,43) and dose and then multiplied by the recommended dose per day, as available from the British National Formulary (46) or the summary of product characteristics (47). The OMED for each product was multiplied by the number of prescriptions issued each year to determine the annual totals. Results were stratified by drug, with less frequently used medicines grouped as 'other' opioids. This included diamorphine (oral preparations), dipipanone, hydromorphone, meptazinol, methadone (tablets), pentazocine, pethidine and tapentadol.
A secondary analysis used proxy-OMED totals for products which had a daily dose of 120mg OMED or more, classed as 'high dose' opioid prescribing (8).

Deprivation scores
The Welsh Index of Multiple Deprivation (WIMD) is the official measure used by Welsh Government to determine relative deprivation of areas within Wales (40). The WIMD is a weighted total score of deprivation based on income (23.5%), employment (23.5%), health (14%), education (14%), geographical access of services (10%), community safety (5%), physical environment (5%) and housing (5%). Scores are not linear, so areas in group 2 are not twice as deprived as those in group 4. Indices are published every 3 years. The 2011 Index was used for this study. Data are presented in quintiles, with WIMD1 being the most deprived areas and WIMD5 the least deprived.

Measuring utilisation
The number of prescriptions and number of patients per year, per drug were calculated in repeat, cross-sections for each year and further stratified into drug and deprivation group. Data were standardised to annual population size using data All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint from the Office of National Statistics (ONS) (48) and StatsWales (49). Deprivation data was adjusted by each quintile's annual population.

Data analysis
Data was extracted from the study tables within SAIL using Structured Query Language-code searches in Eclipse software. Trends were reported standardised per 1000 population. Percentage change rate of number of prescriptions issued and number of people receiving prescriptions over the study period were also noted.
Data was stratified into 8 groups. The 7 most commonly prescribed drugs were buprenorphine, codeine, dihydrocodeine, fentanyl, morphine, oxycodone and tramadol. The 'other' group included dextropropoxyphene, meptazinol, pethidine and tapentadol. Statistical analysis was conducted using SPSS 25 software and figures drawn using Excel 16.30.

Patient and public involvement statement
There was no direct patient involvement in development and design of this study.
However, the SAIL databank has members of the public who provide advice and give recommendations on safeguarding and ethical approval via a Consumer Panel.
Panel members also provide input to the IGRP, which approves all data applications.   Table 1: Daily oral morphine equivalent dose (mg) issued on prescription, given as annual totals and adjusted to population, stratified by drug All rights reserved. No reuse allowed without permission.

Prescribing
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the Codeine was the most commonly prescribed opioid, with just under 12.5 million prescriptions issued and the highest annual total OMED prescribed for the study duration ( Figure 1). Its OMED per 1000 population increased by 79%, from 5916 mg to 10 581 mg. Tramadol was the second most commonly prescribed opioid in Wales.
Its use increased by 74%, from 3397 mg to 5905 mg OMED per 1000 population, although annual total OMED reduced from 2014 ( Figure 1).
Large increases were noted in 'strong' opioids (morphine, oxycodone, fentanyl and buprenorphine) during the study (Figure 1). Morphine OMED increased by 397%, from 1422mg to 7063mg per 1000 population. By 2015, morphine was prescribed around 3 times the equivalent dose of either oxycodone (increased 349%, from 569mg to 2554mg per 1000 population) or fentanyl (increased 131%, from 1164mg to 2 691mg per 1000 population). Morphine OMED was around 6 times that of buprenorphine, which itself increased 171%, from 422 to 1142mg per 1000 population.
Overall, 71% of the total opioid burden in the areas of Wales covered by the SAIL databank was accounted for by 3 drugs; codeine (35%), tramadol (22%) and morphine (14% author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint Despite the large increases in the number of prescriptions issued, OMED per prescription reduced for the four main 'strong' opioids, morphine (-21%, from 86mg to 68mg), oxycodone (-5%, from 105mg to 100mg), fentanyl (-21%, from 186mg to 147mg) and buprenorphine (-62%, from 98mg to 37mg), (Table 1). Buprenorphine OMED reduction resulted from falling use of twice-weekly, higher dose patches (>35micrograms per hour), which outweighed the higher number of prescriptions for lower dose weekly patches (5-20 micrograms per hour).
Whilst the number of prescriptions for 'other' opioids decreased by 91% (from 56 to 5 per 1000 population) over the study period, the OMED per prescription more than doubled from 27mg to 58mg per prescription. Prescription numbers notably increased from 2011 onwards, when tapentadol was released onto the UK market.
Codeine, dihydrocodeine and tramadol dose per prescription increased over the 11 year period.

High dose opioid prescribing
Based on the strength of preparation prescribed, five groups of drugs demonstrated dosing of 120mg OMED and over between 2005 and 2015 (Table 1). When compared to the total OMED prescribed in the study period, statistically significant differences were established between them (p=0.000, H=45.7, 2 =0.8). High-dose fentanyl OMED was confirmed to be significantly larger than OMED for buprenorphine (p<0.001), morphine (p<0.001) and 'other' opioids (p<0.001), as was oxycodone OMED when compared to morphine (p<0.005) and 'other' opioids (p0<.001).
Whilst morphine (335% increase from 1103mg to 4795mg per 1000 population) and oxycodone (340% increase from 377mg to 1659mg per 1000 population) saw similar percentage increases, the total OMED for high-dose morphine was nearly three times that of oxycodone throughout the study (Figure 2). The large percentage increase in 'other' opioids prescribed at 120mg OMED or above (1497% from 9mg to 141mg per 1000 population) was entirely due to tapentadol prescribing from 2011 onwards. Tapentadol prescribing tended to be at high OMED levels.

Opioid prescribing trends by deprivation
All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint The trends in annualised daily OMED of all oral and transdermal opioids stratified by the Welsh Index of Multiple Deprivation (WIMD2011) are illustrated in Figure 3. Over the 11 years of the study, the most deprived areas (WIMD1) examined were prescribed 100 711 696mg more OMEDs than the least deprived (WIMD5) ( Table 2).
Between 2005 and 2015, OMED doubled in all but the least deprived (WIMD5) areas (Table 2), with 28% (176 824 265mg of 622 969 068mg) of the total OMED prescribed issued in the most deprived areas of Wales. In contrast, 12% (76 112 569mg) were prescribed in the least deprived areas. Throughout the study, the OMED prescribed in WIMD1 areas remained more than twice those noted in WIMD5 areas ( Table 2) for both total OMED (mg) and OMED per 1000 population. Despite large percentage increases in all quintiles, the total OMED prescribed in each area was significantly different (p<0.001, H=34.5, 2 =0.61). The least deprived areas had significantly lower prescribed OMED than quintiles 1 to 3 (WIMD1 p<0.001, WIMD2 p<0.001 and WIMD3 p<0.05). The total OMED prescribed in the most deprived areas (WIMD1) was significantly higher compared to that in WIMD4 areas (p<0.05).
All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the .

Oral or transdermal opioids
Products with a daily >120mg OMED

OMED per prescription
OMEDs per prescription were similar in all quintiles, despite the large differences in the actual number of prescriptions issued (Figure 3).

High-dose opioid prescribing by deprivation
The total OMED prescribed for products with a daily OMED of 120mg and over tripled in each deprivation quintile between 2005 and 2015 ( Table 2). As observed with all opioid prescribing, the most deprived areas had the greatest high-dose OMED prescribed. The total of 37 943 319mg prescribed over 11 years was 2.24 times more than in the least deprived quintile, where 16 958 635mg were prescribed in the same period. The differences between high-dose OMED in each quintile were found to be statistically significant with small effect size (p=0.002, H=17.1, 2 =0.26).
All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint After adjustment for population size per quintile, the large differences in OMED per 1000 population dependent on the level of deprivation were maintained with significant differences, although small effect size (p=0.001, H18.5, 2 =0.29). OMED per 1000 population was significantly lower in the least deprived areas of Wales compared to quintiles 1 to3, all with p<0.05. In 2015, around 17% more high-dose OMED was prescribed in WIMD1 areas (11574mg per 1000 population) than in the next most deprived WIMD2 (9863mg per 1000 population). Even though high-dose OMED between 2005 and 2015 tripled in WIMD5 areas, OMED prescribed was still 156% higher in WIMD1 areas at the end of the study (Figure 4).

Discussion
The results of this retrospective database study demonstrate a marked increase in The data demonstrates the disproportionate opioid burden felt by people living in the most deprived areas of Wales. Increasing deprivation was linked to the receipt of higher doses, higher OMED and a higher opioid burden per person. Whilst rises in percentage terms were similar in all WIMD2011 quintiles, significant differences were consistent between the most and least deprived areas throughout the 11 years examined without evidence that the trend was beginning to change.
Despite a low OMED per prescription, codeine is used widely throughout Wales and places a major burden on the population. Morphine was the most burdensome strong opioid, although both oxycodone and fentanyl had higher OMED (mg) per prescription issued.
This study identified rising trends in opioid prescribing in Wales, similar to those previously reported in other parts of the UK (1,3,6,32,33,50). All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org /10.1101/19012625 doi: medRxiv preprint Examining trends by prescription numbers alone is likely to underestimate the opioid burden within a population. Curtis et al demonstrated a 34% growth in prescription numbers resulted in a 127% increase in OMED burden (6). Similarly, as shown in this study, a 44% increase in prescription numbers in Wales translated into a 95% increase in opioid burden using the proxy-OMED measure described.
Codeine is referenced as a 'weak' opioid (8,46) and its risks may therefore be downplayed. However, deaths in Wales associated with codeine, including as part of compound preparations (e.g. co-codamol), increased from 12 in 2005 to 16 in 2015 (51). Furthermore, following the end of the study period, codeine-associated deaths increased a further 50%, with 24 reported fatal incidents in 2018, making codeine responsible for 15% of opioid associated deaths in Wales in 2018 (n=164) (51).
Since codeine is also available to purchase over-the-counter (OTC), prescribing data is an underestimate of population exposure to the drug (31, 52) Higher OMED per prescription for non-morphine strong opioids may point to prescribers misunderstanding oral morphine equivalence (53-55) and using higher doses of non-morphine opioids than are recommended (8). Tapentadol was responsible for a rapid increase in OMED for 'other' opioids since its introduction to the UK market in 2011. This has been noted in England as well (6). By the end of the study period, high dose tapentadol OMED per prescription was greater than that of morphine.
This study has demonstrated substantial increases in opioid prescribing, with higher levels in more deprived populations, which was also previously reported in England (2,33) and internationally (56-59. Furthermore, much of the focus of the opioid crisis in the US has been on rural, deprived communities where access to, and affordability of, healthcare services and social support are poor (35,56,60).
Increased levels of prescribing in areas of high socio-economic deprivation have been linked to greater reported pain intensity (32). However, limited evidence supports the notion that opioids are effective at reducing pain, particularly in the longer term (8,18,61). High-dose opioids (above 120mg OMED) have also been associated with increased levels of pain (62,63). In the context of this and previous All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint studies (2,6,32,33, the implications of the increased prescribing of opioids in more deprived areas are concerning as it exposes the most vulnerable people to higher levels of medicines that may be ineffective at best and may even cause additional health and well-being complications (11).
This study is the first to utilise a large dataset including routine data from 78% of Primary Care practices in Wales to map opioid prescribing trends (4). Large sets of prescribing and diagnostic data have been validated as an accurate means for conducting healthcare population research (64,65) as they reduce recall bias and regional variation. Moreover, any person registered with included practices and prescribed an opioid medicine was included in the analysis, thus avoiding selection bias. Furthermore, this is the first study conducted using Welsh data that utilised an OMED measure to better understand the burden of opioid prescribing on the population. Using the linkage systems within SAIL datasets, data from people with a recorded cancer diagnosis, could be excluded from analysis. While recent studies of opioid prescribing in the UK based on large datasets merely assumed that the majority of prescribing was attributable to persisting, non-cancer pain based on the longevity of prescribing and the dose forms used (2,6,50), this data confidently reflects prescribing for non-cancer pain.
However, prescribing data alone provides an indication of intention to treat but does not confirm consumption. It also does not give an indication of the diagnosis or how long an individual might have been using the medication. Moreover, the data presented here did not identify people receiving more than one opioid medicine and who would have a higher individual OMED burden.
It was not possible to access linked dispensing data, which would provide the detail required to more accurately calculate OMED. This resulted in the development of a proxy measure which provides a daily estimation of OMED, multiplied by the number of prescriptions as described above, and required a number of assumptions to be made. Without knowing prescription duration, i.e. how many daily doses were provided per prescription, the annualised OMEDs presented here may be an underestimation of the opioid burden. However, the trends described are similar to those reported elsewhere in the UK (1-3,6,33,50). All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint For this study, opioid medicines were identified by read codes and accuracy of the data extraction depended on the inclusivity of the coding used. Incomplete coding lists would result in data being lost to the analysis and could lead to an underrepresentation of prescribing. Also, oral and transdermal medicines were selected on the likelihood of them being used for pain management rather than the management of misuse disorder. For example, high dose sublingual buprenorphine and methadone solution were excluded. However, in rare cases, these may be prescribed as analgesia while other included opioid preparations could be used as part of misuse management. Again, this could result in an over or underestimation of prescribing. However, similar rationales for deciding which opioid products to include in analysis of Primary Care prescribing was adopted by other UK-based authors (1,3,6,33).
Using a measure of opioid burden, such as oral morphine equivalent dose gives a more accurate measure of the impact of opioid prescribing on a population. In Welsh Government recently set out plans to improve access to persistent pain management support across Wales (67). It focusses on improving access to nonpharmacological and self-management support such as pain management programmes (68,69), psychological input (70) and physiotherapy (71) which have been demonstrated to have longer-term benefits and do not carry the same risks as opioids. However, as with other guidelines in this area (72,73), the document does not include a strategy to address health inequalities in pain management support per se (64). Given the disproportionate use of opioids in the most deprived areas, it All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint would seem essential to tackle the root causes of those differences, if anything is to improve. More research is needed to better understand the reasons opioids are still prescribed in the face of mounting evidence of harm including current concerns around dependence and misuse (8,11,31). Furthermore, greater understanding of why people living with pain continue to use opioids in spite of poor effect or additional health concerns is required.
Supporting non-pharmacological management is essential to enable people to live well with pain (74) but can be difficult to encourage in practice (75). McCrorie and colleagues (2015) concluded that problematic opioid prescribing can result from failing to meet the self-perceived needs of people living with pain and from prescribers feeling unable to negotiate alternatives (76). This was echoed by Finestone who described opioid prescribing as a surrogate for poor access to pain management support (77).
It would seem that more targeted use of resources into Primary Care, and particularly in areas of greater deprivation, would be a prudent way of improving lives and tackling the use of opioid medicines that can cause so much harm.

Contributors
ED conceived of and designed the study, collated the read-codes used for data extraction, devised the OMED proxy-measure and coded the extracted data, undertook the data analysis, drafted and revised the manuscript. BS oversaw the study design and data analysis and critically revised the manuscript. MJ oversaw the study design and statistical analysis and critically revised the manuscript. CP and JR oversaw the study design and critically revised the manuscript. All authors read and approved the final manuscript.

Funding statement
This work was supported by Pharmacy Research UK grant number PRUK-2016-PA1-A. ED's PhD is partly supported by funding from Research Capacity Building Collaboration.

Patient and Public Involvement statement
This research was done without patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patient relevant outcomes or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or accuracy. All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint   author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

List of figures
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19012625 doi: medRxiv preprint