Appendix Table 1.

Cohort Flow Descriptions

Step	Description	Number of patients	Number of person-years
	Prevalence cohort
1	All opioid users	12,930,228
2	At least 90 days opioid use (opioid use periods)a	1,298,048	2,570,004
3	Continuous eligibility for entire calendar year and previous yeara	727,693	1,580,955
4	At least 18 years old at time of first opioid use, had known gender, region	721,843	1,570,054
5	Had no acute pain conditions for entire calendar year	621,449	1,228,044
	Incidence cohort
1	All opioid users	12,930,228
2	At least 90 days opioid use (opioid use periods)a	1,298,048	2,570,004
3	Continuous eligibility for 1 year prior to start of long-term opioid use	665,377	1,311,672
4	No prior opioid use in 1 year prior to start of long-term opioid use	49,893	50,426
5	At least 18 years old at time of first opioid use, had known gender, region	49,508	50,040
6	First opioid prescription for at least 28 days	35,682	36,194
7	Had no acute pain conditions in prior to first opioid use	28,790	29,202

^aLTOT users could be counted for multiple years if they had at least 90 days of opioid use in each calendar year. For users with only 1 episode of ≥90 days use which spanned across 2 calendar years, the earlier year was used as the year of use, if it included at least 30 days of use. Otherwise, the later year was selected. For example, an LTOT user with 90 days’ use starting December 1, 2012 and ending February 28, 2013 would count for 2012 because the episode had 31 days’ use in 2012. An LTOT user with 90 days use starting December 3, 2012 and ending March 2 would count for 2013 because the episode included only 29 days of use in 2012.

LTOT, long term opioid therapy.