Examining patterns in opioid prescribing for non

To examine trends in strong opioid prescribing in a primary care population in Wales and identify if factors such as age, deprivation and recorded diagnosis of depression or anxiety may have influenced any changes noted. Trend, cross-sectional and longitudinal analyses of routine data from the Primary Care General Practice database and accessed via the Secure Anonymised Information Linkage (SAIL) databank. A total of 345 Primary Care practices in Wales. Anonymised records of 1,223,503 people aged 18 or over, receiving at least one opioid prescription between 1 January 2005 and 31 December 2015 were analysed. People with a cancer diagnosis (10.1%) were excluded from the detailed analysis. During the study period, 26,180,200 opioid prescriptions were issued to 1,223,503 individuals (55.9% female, 89.9% non-cancer diagnoses). The greatest increase in annual prescribing was in the 18–24 age group (10,470%), from 0.08 to 8.3 prescriptions/1000 population, although the 85+ age group had the highest prescribing rates across the study period (from 149.9 to 288.5 prescriptions/1000 population). The number of people with recorded diagnoses of depression or anxiety and prescribed strong opioids increased from 1.2 to 5.1 people/1000 population (328%). The increase was 366.9% in areas of highest deprivation compared to 310.3 in the least. Areas of greatest deprivation had more than twice the rate of strong opioid prescribing than the least deprived areas of Wales. The study highlights a large increase in strong opioid prescribing for non-cancer pain, in Wales between 2005 and 2015. Population groups of interest include the youngest and oldest adult age groups and people with depression or anxiety particularly if living in the most deprived communities. Based on this evidence, development of a Welsh national guidance on safe and rational prescribing of opioids in chronic pain would be advisable to prevent further escalation of these medicines. This is the first large-scale, observational study of opioid prescribing in Wales. Over 1？million individual, anonymised medical records have been searched in order to develop the study cohort, thus reducing recall bias. Diagnosis and intervention coding in the Primary Care General Practice database is limited at input and may lead to under-reporting of diagnoses. There are limitations to the data available through the Secure Anonymised Information Linkage databank because anonymously linked dispensing data (what people collect from the pharmacy) are not currently available. Consequently, the results presented here could be