Prescription Opioid Registry Protocol in an Integrated Health System

Abstract

Objective

To establish a prescription opioid registry protocol in a large health system and to describe algorithms to characterize persons using prescription opioids, opioid use episodes and concurrent use of sedative/hypnotics.

Study Design

Protocol development and retrospective cohort study.

Methods

Using Kaiser Permanente Northern California (KPNC) electronic health record data, we selected patients using prescription opioids in 2011. Opioid and sedative/hypnotic fills, and physical and psychiatric comorbidity diagnoses were extracted for years 2008 to 2014. Algorithms were developed to identify each patient’s daily opioid and sedative/hypnotic use, and morphine daily dose equivalent. Opioid episodes were classified as long-term, episodic, or acute. Logistic regression was used to predict characteristics associated with becoming a long-term opioid user.

Results

In 2011, 18% of KPNC adult members filled at least one opioid prescription. Among those patients, 25% used opioids long-term and their average duration of use was more than 4 years. Sedative/hypnotics were used by 76% of long-term users. Being over 80 years of age, white, living in a more deprived neighborhood, having a chronic pain diagnosis, and use of sedative/hypnotics were predictors of initiating long-term opioid use.

Conclusion

This study established a population-based opioid registry that is flexible, and can be used to address important questions of prescription opioid use. It will be used in future studies to answer a broad range of other critical public health issues relating to prescription opioid use.

INTRODUCTION

Prescription opioid use has increased dramatically in the past two decades, with associated increases in opioid misuse/abuse and opioid overdose. These are among the most commonly prescribed medications,¹ with 259 million prescriptions written for opioid pain relievers in the United States in 2012.² However, effectiveness of these medications for long-term use has not been established,³ and risk of opioid-related abuse and overdose has led to a prescription opioid epidemic.⁴ Close to 2 million Americans had opioid abuse or dependence in 2014.⁵ From 2003 to 2013, the proportion of drug abuse treatment admissions for non-heroin opiates tripled.⁶ More than 18,000 people had a fatal overdose in 2014 related to prescription opioids, more than four times the number in 1999.⁷ Sedative/hypnotics are frequently involved in overdose deaths, and their concurrent use with opioids is of high concern.⁸

The goal of the overall project was to use electronic health record (EHR) data from Kaiser Permanente Northern California (KPNC) to develop a patient prescription opioid registry with the potential to ultimately address several research inquiries: the characterization of opioid use and opioid users, identification of prescription opioid misuse, predictors of opioid overdose, and to describe patients’ services utilization and costs. It draws on our, and others, previous research.^9–11 Our objective here is to describe our protocol to develop the registry and to address the initial research questions: 1) characterize all persons who used prescription opioids in 2011; 2) analyze their opioid use, and concurrent use of sedative/hypnotics; and 3) identify predictors of becoming a new long-term user of opioids. We provide detail and context for our methodological approach, which can be a foundation for future analyses and we hope a methodological resource for other research teams addressing these questions.

MATERIALS AND METHODS

Setting

KPNC is a nonprofit, integrated health care delivery system providing comprehensive health services to approximately 3.8 million members in Northern California. The membership reflects the region’s general population, although it under-represents persons with very low levels of education and income.¹² Membership includes enrollees from Medicare, Medicaid and State health insurance program.

Data Sources

Membership, outpatient pharmacy, and medical encounter data are archived in KPNC’s EHR. Demographic (e.g. age, race/ethnicity, sex, address), membership status, health services, diagnostic and pharmacy data are regularly extracted from the EHR and stored in a Virtual Data Warehouse (VDW).¹³ The VDW is a distributed data model that includes EHR data and other data such as mortality and census data. Although the current study only uses KPNC data, the VDW is a feature of the Health Care Systems Research Network (http://www.hcsrn.org/en/), whose data are harmonized across 19 health systems.

KPNC pharmacy data in the VDW include generic name, strength, directions for use, date dispensed, quantity dispensed, days-supply, prescriber identification number, and National Drug Code (NDC). Surveys have found that over 90% of members obtain all or almost all of their prescription medications through KPNC pharmacies.¹⁴ KPNC does not have policies that impose mandatory opioid dosage limits or prior authorization.

Clinical diagnostic and health services utilization data include hospitalizations, and Emergency Department (ED) and outpatient clinic visits (primary and specialty care). Mortality data, including date of death and underlying cause of death are created from the EHR and death certificates. The VDW tumor registry contains information on all new cancers for KPNC members diagnosed after January 1, 1997.

Registry Inclusion

We extracted all opioid fills made at KPNC outpatient pharmacies during 2011 (Supplemental Table 1). Similar to prior research,^8,9,15,16 we focus on formulations with higher likelihood of abuse, and those used primarily to treat pain. Thus we excluded opioid formulations used primarily as antitussives, anesthetics, antihistamines, antidiarrheals, or opioid agonists/antagonists. Eligible registry participants were ≥19 years of age (on January 1, 2011) with at least one opioid fill during 2011, and without a cancer diagnosis between January 1, 1997 and December 31, 2014.

To compare the demographics of persons with and without opioid use, we selected a comparison cohort of all persons who were: KPNC members during 2011; aged ≥19 years on January 1, 2011; without a cancer diagnosis between 1997 and 2014, and without an opioid fill in 2011.

For each person, we extracted gender, race/ethnicity, birth date, and home address as of January 1, 2011. Using member’s home address in combination with the 2006–2010 American Community Survey collected by the U.S. Census Bureau,^17,18 a neighborhood deprivation index (NDI) was generated at the census tract level.¹⁹ Chi-square tests and t-tests were used to identify differences (at p<0.05) between the groups.

Registry Structure

For the persons who used prescription opioids, we extracted all outpatient opioid fills between January 1, 2008 and December 31, 2014. This allowed at least three years of opioid use before and after 2011, and facilitated analyses related to history and subsequent course of opioid use. We constructed a dataset of daily use: one record per person per day for each day of those seven years; for each day, variables indicated if the person was considered to be using opioids, opioid type, and the morphine equivalent milligrams used. To identify opioid use on any given day, we assumed: 1) persons used opioids according to the “days-supply” variable – i.e. according to provider instructions; 2) if a person had ≥ 7 days remaining on a prior fill at dispensation, the new fill was assumed to be used concurrently with the prior fill, otherwise, the new fill was assumed to be used consecutively; and 3) stock from three fills could be used concurrently, or be “held” for future use.

Opioid Episode Definition and Classification

Using the daily use dataset, we created episodes of opioid use, defined as the period from the start of any opioid use until a gap in use of more than 180-days.^9–11 For each episode, we calculated: 1) episode duration; 2) number of fills; 3) prescribed days-supply of opioids filled during the episode; 4) mean morphine daily dose equivalent (MDDE) across the duration of the episode, by opioid type; 5) MDDE “as prescribed”, by opioid type (where the denominator is prescribed days-supply). Because episodes can include gaps in use, MDDE across episodes will typically be lower than MDDE as prescribed, since the latter’s denominator does not include gaps in use.

Episodes were classified into three mutually-exclusive types: 1) acute, 2) episodic, and 3) long-term, as in previous research.^9–11 “Acute” episodes were those lasting <90 days. “Episodic” episodes were of ≥90 days, and during which the person was dispensed <120 days-supply of opioids, and during which there were <10 opioid fills. “Long-term” episodes were of ≥90 days with either ≥120 days-supply of opioids or ≥10 opioid fills. When describing episodes spanning any part of 2011, we retained only the most “severe” episode per person and classified the person according to that episode type. Thus, all persons who had a long-term opioid episode covering any part of 2011 were classified as “persons with long-term opioid use”. Among the remaining, those with at least one episodic episode were “persons with episodic opioid use”, and the remaining persons were “persons with acute opioid use”.

Comorbid Health Conditions, Mortality, and Sedative/Hypnotic Use

To examine comorbidities, we extracted from the VDW all diagnoses (International Classification of Diseases, 9th Revision, Clinical Modification [ICD9]) associated with healthcare encounters. As in prior research with other complex patient populations,²⁰ we identified whether persons received a diagnosis for one or more of thirteen chronic conditions: arthritis, asthma, congestive heart failure, chronic obstructive pulmonary disease, chronic pain, diabetes, epilepsy, end-stage renal disease, hypertension, ischemic heart disease, osteoporosis, Parkinson’s Disease, and stroke. We also identified persons with the following psychiatric and substance-use disorders:²⁰ attention deficit disorders, anxiety disorder, autism, bipolar disorder, dementia, depression, other psychoses, personality disorder, schizophrenia, opioid abuse/dependence, and non-opioid substance use disorders (excluding tobacco). Date of death and underlying cause of death were also extracted, as was KPNC membership information for each month from 2008 to 2014.

For persons with opioid use, we extracted all KPNC pharmacy fills for sedative/hypnotic medications (Supplemental Table 2) during 2008–2014. Using days-supply and the same approach as for opioid fills (without calculating dose), we identified daily use of sedative/hypnotics for each person. These records were merged with the opioid episodes to calculate the number of days of sedative/hypnotic use during each opioid episode.

Predictors of Initiating Long-Term Opioid Use

Persons who use opioids long-term are at highest risk of adverse events compared to persons who use shorter term.²¹ To identify predictors of becoming a long-term user within three to four years of starting opioid use, we identified “opioid naïve” persons without an opioid fill prior to their first fill of 2011; therefore at least three years without prior use of prescription opioids. We required continuous KPNC membership from 2008–2014, or death, subsequent to their first 2011 fill (allowing gaps in membership of ≤3 months). In the year prior to their first opioid fill, we examined medical diagnoses, sedative/hypnotic use, and four measures of utilization as proxies for severity and propensity to use resources: 1) number of inpatient days; 2) number of ED visits; 3) number of office visits, and; 4) number of non-opioid prescription fills.

We conducted logistic regression analysis with the dependent variable indicating whether or not the person subsequently had a long-term opioid episode from initial fill to December 31, 2014. In addition to sex, age (in seven groups), race/ethnicity, and NDI quartile, we included the following covariates measured in the year before the first opioid fill: dichotomous indicators for each of 13 chronic medical condition; dichotomous indicators for any psychiatric disorder, opioid abuse/dependence, and non-opioid abuse/dependence; and psychiatric conditions; a dichotomous indicator of sedative/hypnotic use; the four utilization measures; and a class variable indicating the KPNC clinic where the patient received most of their care.

RESULTS

Among 2,480,030 adult KPNC members in 2011, 455,693 (18.3%) had at least one opioid fill at a KPNC pharmacy. Persons with opioid use were different from persons without opioid use on every characteristic, including being more likely to be female, older, white, have a chronic medical or psychiatric condition, and to have a diagnosis of opioid abuse/dependence or non-opioid substance use disorder (Table 1).

Table 1.

Description of Prescription Opioid Persons With and Without Prescription Opioid Use, 2011.^a

Characteristic	Persons with prescription opioid use (n=455,693)	Persons without prescription opioid use (n=2,024,337)
Gender, n (%)
Female	267,755 (58.8)	1,031,995 (51.0)
Male	187,938 (41.2)	992,342 (49.0)
Age, mean (med)	50.35 (17.3)	45.52 (16.7)
Age Group, n (%)
19 – <30	64,866 (14.2)	436,364 (21.6)
30 – <40	74,636 (16.4)	398,724 (19.7)
40 – <50	86,527 (19.0)	407,002 (20.1)
50 – <60	95,702 (21.0)	371,768 (18.4)
60 – <70	69,879 (15.3)	240,725 (11.9)
70 – <80	39,082 (8.6)	106,717 (5.3)
80+	25,001 (5.5)	63,037 (3.1)
Race/Ethnicity, n (%)
Asian	41,920 (9.2)	356,773 (17.6)
Black	43,378 (9.5)	126,310 (6.2)
Hispanic	80,540 (17.7)	342,260 (16.9)
Pacific Islander	2,203 (0.5)	11,445 (0.6)
Native American	2,825 (0.6)	8,230 (0.4)
Multi-racial	19,092 (4.2)	53,735 (2.7)
Other or unknown	10,958 (2.4)	214,133 (10.6)
White	254,777 (55.9)	911,451 (45.0)
Member months in 2011, mean (median)	11.48 (1.8)	10.84 (2.8)
Died in 2011, n (%)	4,761 (1.0)	8,243 (0.4)
Neighborhood deprivation, n (%)b
Quartile 1, least deprived	101,499 (22.3)	523,074 (25.8)
Quartile 2	132,226 (29.0)	579,802 (28.6)
Quartile 3	127,979 (28.1)	524,327 (25.9)
Quartile 4, most deprived	87,079 (19.1)	343,444 (17.0)
Missing	6,910 (1.5)	53,690 (2.7)
Diagnosed conditions, n (%)c
Any chronic medical conditiond	287,436 (63.1)	627,085 (31.0)
Arthritis	150,481 (33.0)	214,147 (10.6)
Asthma	62,178 (13.6)	110,894 (5.5)
Congestive heart failure	16,903 (3.7)	19,913 (1.0)
Chronic obstructive pulmonary disease	29,437 (6.5)	46,477 (2.3)
Diabetes	64,286 (14.1)	138,849 (6.9)
Epilepsy	4,432 (1.0)	8,050 (0.4)
End-stage renal disease	3,108 (0.7)	2,054 (0.1)
Hypertension	162,064 (35.6)	359,712 (17.8)
Ischemic heart disease	30,708 (6.7)	49,051 (2.4)
Osteoporosis	18,340 (4.0)	36,099 (1.8)
Chronic Pain	64,421 (14.1)	32,421 (1.6)
Parkinson’s disease	1,789 (0.4)	3,777 (0.2)
Stroke	7,364 (1.6)	10,105 (0.5)
Psychiatric disorders
Attention deficit disorders	4,491 (1.0)	10,529 (0.5)
Anxiety	56,311 (12.4)	96,373 (4.8)
Autism	112 (<0.1)	1,043 (<0.1)
Bipolar disorder	7,371 (1.6)	10,882 (0.5)
Dementia	4,580 (1.0)	8,709 (0.4)
Depression	76,352 (16.8)	116,252 (5.7)
Other psychoses	2,694 (0.6)	4,105 (0.2)
Personality disorder	2,550 (0.6)	2,770 (0.1)
Schizophrenia	1,555 (0.3)	4,140 (0.2)
Any psychiatric disorder	113,794 (25.0)	197,813 (9.8)
Opioid abuse/dependence	4,918 (1.1)	1,880 (<0.1)
Non-opioid substance use disorders	18,130 (4.0)	24,408 (1.2)

^aOpioid use refers to any fill at a KPNC pharmacy for an opioid. Some persons without opioid “use” may have had a fill at a non-KPNC pharmacy. All characteristics between opioid patients and persons without opioid use were statistically different at p<0.05 based on chi-square (for categorical variables) or T-test (for continuous variables). KPNC: Kaiser Permanente Northern California.

^bNeighborhood deprivation index was created using data from the 2006–2010 American Community Survey collected by the US Census Bureau, and was generated through principal components analysis of eight variables at the census tract level, including percentages of males in the neighborhood working in management and professional occupations, residents living in crowded housing (more than one person per room), households in poverty, households headed by females with dependents, households receiving public assistance, households earning <$30,000 per year, residents ≥25 years of age with less than a high school education, and residents ≥16 years of age who are unemployed. Quartiles based on all census tracts within KPNC service area.

^cPatients receiving at least one diagnoses for these conditions during the year 2011.

^dIncludes all the conditions listed below except the psychiatric health disorders, opioid abuse/dependence, and non-opioid substance use disorders.

The 455,693 persons with opioid use had 474,045 unique episodes occurring in some part of 2011. Due to the 180-day gaps used when creating episodes, few persons had more than one episode in 2011: 18,352 persons had two episodes covering any part of 2011, and the rest had one. After retaining the most “severe” episode per person, there were 112,089 long-term opioid episodes, 71,011 episodic episodes, and 272,593 acute episodes (Table 2). Long-term users were on average 56 years of age and 61% were female. Forty percent of persons with long-term opioid use received at least one diagnosis for a psychiatric disorder in 2011, 3% were diagnosed with opioid abuse/dependence, and 7% were diagnosed with a non-opioid substance use disorder.

Table 2.

Characteristics Among Persons with an Opioid Episode Spanning Any Part of 2011, by Episode Type.^a

Characteristic	Type of opioid use
	Long-term (n=112,089)	Episodic (n=71,011)	Acute (n=272,593)
Female, n (%)	68,658 (61)	42,751 (60)	156,346 (57)
Age, mean (median)	56.41 (55.98)	51.32 (51.15)	47.61 (46.67)
Age Group, n (%)
19 – <30	5,429 (5)	9,310 (13)	50,127 (18)
30 – <40	11,363 (10)	11,039 (16)	52,234 (19)
40 – <50	20,984 (19)	13,359 (19)	52,184 (19)
50 – <60	30,012 (27)	14,758 (21)	50,932 (19)
60 – <70	22,285 (20)	11,481 (16)	36,113 (13)
70 – <80	13,207 (12)	6,793 (10)	19,082 (7)
80+	8,809 (8)	4,271 (6)	11,921 (4)
Race/Ethnicity, n (%)
Asian	3,037 (3)	5,104 (7)	33,779 (12)
Black	11,288 (10)	7,993 (11)	24,097 (9)
Hispanic	14,493 (13)	13,123 (18)	52,924 (19)
Pacific Islander	256 (<1)	370 (1)	1,577 (1)
Native American	972 (1)	470 (1)	1,383 (1)
Multi-racial	6,133 (5)	3,287 (5)	9,672 (4)
Other or unknown	1,370 (1)	1,250 (2)	8,338 (3)
White	74,540 (67)	39,414 (56)	140,823 (52)
Member months in 2011, mean (median)	11.48 (12.00)	11.52 (12.00)	11.47 (12.00)
Neighborhood deprivation, n (%)
Quartile 1, least deprived	21,543 (19)	15,116 (21)	64,840 (24)
Quartile 2	32,303 (29)	20,350 (29)	79,573 (29)
Quartile 3	33,162 (30)	20,252 (29)	74,565 (27)
Quartile 4, most deprived	23,268 (21)	14,273 (20)	49,538 (18)
Missing	1,813 (2)	1,020 (1)	4,077 (1)
Diagnosed conditions, n (%)b
Any chronic medical conditionc	92,894 (83)	49,355 (70)	145,187 (53)
Arthritis	52,274 (47)	28,155 (40)	70,052 (26)
Asthma	19,870 (18)	11,350 (16)	30,958 (11)
Congestive heart failure	7,287 (7)	3,121 (4)	6,495 (2)
Chronic obstructive pulmonary disease	12,482 (11)	5,014 (7)	11,941 (4)
Diabetes	22,565 (20)	11,388 (16)	30,333 (11)
Epilepsy	1,654 (1)	779 (1)	1,999 (1)
End-stage renal disease	1,025 (1)	758 (1)	1,325 (<1)
Hypertension	56,453 (50)	27,679 (39)	77,932 (29)
Ischemic heart disease	11,715 (10)	5,497 (8)	13,496 (5)
Osteoporosis	6,988 (6)	3,092 (4)	8,260 (3)
Chronic Pain	41,238 (37)	8,768 (12)	14,415 (5)
Parkinson’s disease	640 (1)	324 (<1)	825 (<1)
Stroke	2,602 (2)	1,387 (2)	3,375 (1)
Psychiatric disorders
Attention deficit disorders	1,392 (1)	788 (1)	2,311 (1)
Anxiety	22,288 (20)	9,556 (13)	24,467 (9)
Autism	13 (<1)	7 (<1)	92 (<1)
Bipolar disorder	3,352 (3)	1,204 (2)	2,815 (1)
Dementia	1,709 (2)	839 (1)	2,032 (1)
Depression	31,890 (28)	13,069 (18)	31,393 (12)
Other psychoses	1,229 (1)	406 (1)	1,059 (<1)
Personality disorder	1,291 (1)	433 (1)	826 (<1)
Schizophrenia	716 (1)	223 (<1)	616 (<1)
Any psychiatric disorder	44,437 (40)	19,461 (27)	49,896 (18)
Opioid abuse/dependence	3,656 (3)	434 (1)	828 (<1)
Non-opioid substance use disorders	7,372 (7)	3,042 (4)	7,716 (3)

^aAll opioid fills were extracted from 2008 to 2014 for persons with at least one opioid fill in 2011. Using these fills, opioid-use episodes were constructed. For this table, only one episode (the most “severe”) per person was retained. For episodes that began prior to 2008, the episode start date will be identified here as being the first fill after January 1, 2008. For episodes that ended, or continued, after 2014, the episode stop date will only include fills through December 31, 2014. Therefore, some episodes are “left” truncated, and some are “right” truncated. All characteristics between long-term opioid patients and episodic opioid patients were statistically different at p<0.05 based on chi-square (for categorical variables) or T-test (for continuous variables), except for autism. All characteristics between episodic opioid patients and acute opioid patients were statistically different at p<0.05.

^bPatients receiving at least one diagnoses for these conditions during the year 2011.

^cIncludes all the conditions listed below except the psychiatric disorders, opioid abuse/dependence, and non-opioid substance use disorders.

On average, long-term episodes were 1,609 days long and included 54 opioid fills (Table 3). Because the opioid data spanned 2008 to 2014, it is possible that some episodes began prior to 2008 or continued after 2014, thus episode durations may be underestimated. Among long-term opioid episodes, 27% began prior to July 1, 2008 and ended after June 30, 2014, and may be both “left” and “right truncated”. Another 17% of long-term episodes may have been left truncated only, and 24% may have been right truncated only.

Table 3.

Characteristics of Opioid Episodes Spanning Any Part of 2011.^a

Episode characteristic	Mean (median), except where noted otherwise
	Long-term episodes (n=112,089)	Episodic episodes (n=71,011)	Acute episodes (n=272,593)	All episodes (n=455,693)
Days durationb	1608.68 (1576.00)	226.29 (179.00)	11.90 (6.00)	438.07 (17.00)
Opioid fills during episode	53.87 (42.00)	3.86 (3.00)	1.28 (1.00)	14.62 (2.00)
Prescribed days supply of fills during episodec	1144.98 (787.00)	30.85 (22.00)	7.83 (6.00)	291.13 (10.00)
MDDE across episode, n (%)d
<20	66,218 (59)	68,773 (97)	45,963 (17)	180,954 (40)
20 to <50	26,746 (24)	2,012 (3)	143,659 (53)	172,417 (38)
50 to <120	11,915 (11)	218 (<1)	76,741 (28)	88,874 (20)
120+	7,210 (6)	8 (<1)	6,230 (2)	13,448 (3)
MDDE across episode, by opioid type, mgd
Schedule III opioids	15.16 (9.57)	4.36 (2.83)	36.28 (25.00)	26.11 (20.00)
Short-acting schedule II opioids	4.54 (0.00)	0.87 (0.00)	4.36 (0.00)	3.86 (0.00)
Long-acting schedule II opioids	19.27 (0.00)	0.15 (0.00)	0.21 (0.00)	4.89 (0.00)
Total	38.98 (15.25)	5.37 (3.34)	40.85 (30.00)	34.87 (21.43)
MDDE as prescribed, n (%)e
<20	18,319 (16)	7,196 (10)	20,459 (8)	45,974 (10)
20 to <50	64,807 (58)	48,188 (68)	157,484 (58)	270,479 (59)
50 to <120	23,772 (21)	14,672 (21)	87,792 (32)	126,236 (28)
120+	5,191 (5)	955 (1)	6,858 (3)	13,004 (3)
MDDE as prescribed, by opioid type, mge
Schedule III opioids	35.32 (30.00)	38.12 (30.00)	43.37 (30.00)	40.53 (30.00)
Short-acting schedule II opioids	61.98 (48.00)	65.17 (52.50)	68.08 (57.50)	64.12 (50.23)
Long-acting schedule II opioids	101.11 (60.00)	49.93 (30.00)	54.26 (30.00)	97.66 (60.00)
Total	46.53 (33.88)	39.33 (31.67)	45.09 (33.33)	44.54 (33.33)
Episodes during which sedative/hypnotics were used, n (%)	85,262 (76)	29,063 (41)	44,271 (16)	158,596 (35)
Among opioid episodes with sedative/hypnotic use
Days duration of opioid episodes	1675.85 (1675.00)	251.61 (198.00)	17.02 (7.00)	951.80 (581.00)
Proportion of days in episode that sedative/hypnotics were used	0.35	0.24	0.57	0.35

^aAll opioid fills were extracted from 2008 to 2014 for persons with at least one opioid fill in 2011. Using these fills, opioid-use episodes were constructed. For this table, only one episode (the most “severe”) per person was retained. For episodes that began prior to 2008, the episode start date will be identified here as being the first fill after January 1, 2008. For episodes that ended, or continued, after 2014, the episode stop date will only include fills through December 31, 2014. Therefore, some episodes are “left” truncated, and some are “right” truncated. All characteristics between long-term opioid patients and episodic opioid patients were statistically different at p<0.05 based on chi-square (for categorical variables) or T-test (for continuous variables). All characteristics between episodic opioid patients and acute opioid patients were statistically different at p<0.05 except MDDE as prescribed, long-acting schedule II opioids.

^bEpisode duration includes gaps in use of up to 180 days. As noted above, episodes may be right or left truncated.

^cThe days supply prescribed during the period. Due to the allowance of gaps in use and/or the concurrent use of multiple opioid fills on the same day, prescribed days supply for an episode can be greater or less than the episode duration.

^dMDDE: Morphine daily dose equivalent. Total dose assumed to be used during the episode, divided by episode duration (which may include gaps in use).

^eTotal dose of this opioid-type assumed to be used during the episode, divided by the prescribed days supply for this opioid-type.

Among persons with long-term use, the mean MDDE was 38.98 mg, with the highest MDDE for long-acting, Schedule II opioids. However, use of those opioids was highly skewed (median MDDE was zero), with only 28% of long-term users using any long-acting, Schedule II opioids. Among persons with episodic and acute use, on the other hand, Schedule III opioids had the highest mean MDDE. Persons with episodic use used at lower levels than long-term users, and tended to have substantial gaps between fills, and therefore had a much lower mean MDDE (5.37 mg.). Persons with long-term use also had higher mean MDDE than episodic or acute users.

Sedative/hypnotics were used by 76% of persons with long-term opioid use during their episodes, and for an average of 34% of episode days. Among persons with acute use, 16% used sedative/hypnotics during their (much shorter) acute episode.

Among all persons using prescription opioids in 2011, 175,558 (39%) were opioid naïve. Of these, 85,305 had continuous KPNC membership from 2008–2014 (n=81,809), or until death (n=3496), and were the analytic sample for initiating long-term opioid use (Table 4). Multivariate analysis indicated that persons ≥80 years of age were more likely to become long-term users than persons <age 50. Compared to Whites, Asians and Hispanics were less likely to become long-term users (OR:0.35; CI: 0.30–0.40, and OR: 0.62; CI: 0.55–0.69, respectively). Persons in more deprived neighborhoods were more likely to become long-term users than those in the least deprived neighborhoods (most deprived neighborhood; OR: 1.46; CI: 1.30–1.64).

Table 4.

Persons with and without Subsequent Long-term Opioid Use After an Initial Opiod Fill.^a

Characteristic	Had a long-term opioid episode between 2011 and 2014 (n=3605)	Did not have a long-term opioid episode between 2011 and 2014 (n=81,700)
Gender, n (%)
Female	1,988 (55.1)	44,785 (54.8)
Male	1,617 (44.9)	36,915 (45.2)
Age, mean (median)b	60.75 (61.04)	52.73 (52.65)
Age Group, n (%)b
19 – <30	147 (4.1)	8,430 (10.3)
30 – <40	242 (6.7)	11,981 (14.7)
40 – <50	490 (13.6)	15,620 (19.1)
50 – <60	833 (23.1)	17,582 (21.5)
60 – <70	802 (22.2)	14,136 (17.3)
70 – <80	613 (17.0)	8,594 (10.5)
80+	478 (13.3)	5,357 (6.6)
Race/Ethnicity, n (%)b
Asian	200 (5.5)	12,130 (14.8)
Black	295 (8.2)	6,157 (7.5)
Hispanic	420 (11.7)	13,888 (17.0)
Pacific Islander	11 (0.3)	367 (0.4)
Native American	17 (0.5)	339 (0.4)
Multi-racial	195 (5.4)	3,255 (4.0)
Other or unknown	25 (0.7)	1,475 (1.8)
White	2,442 (67.7)	44,089 (54.0)
Neighborhood deprivation, n (%)b
Quartile 1, least deprived	824 (22.9)	22,223 (27.2)
Quartile 2	1,122 (31.1)	24,680 (30.2)
Quartile 3	1,022 (28.3)	21,292 (26.1)
Quartile 4, most deprived	597 (16.6)	12,475 (15.3)
Diagnosed conditions in the year prior to initial 2011 opioid fill, n (%)c
Any chronic medical conditionb,d	2,668 (74.0)	42,256 (51.7)
Arthritisb	1,311 (36.4)	18,032 (22.1)
Asthmab	457 (12.7)	7,310 (8.9)
Congestive heart failureb	203 (5.6)	2,176 (2.7)
Chronic obstructive pulmonary diseaseb	371 (10.3)	3,627 (4.4)
Diabetesb	700 (19.4)	9,495 (11.6)
Epilepsy	26 (0.7)	509 (0.6)
End-stage renal diseaseb	24 (0.7)	273 (0.3)
Hypertensionb	1,873 (52.0)	25,731 (31.5)
Ischemic heart diseaseb	369 (10.2)	4,646 (5.7)
Osteoporosisb	207 (5.7)	2,697 (3.3)
Chronic Painb	322 (8.9)	2,085 (2.6)
Parkinson’s diseaseb	34 (0.9)	316 (0.4)
Strokeb	107 (3.0)	1,215 (1.5)
Any psychiatric disorderb	890 (24.7)	11,030 (13.5)
Opioid abuse/dependenceb	9 (0.2)	51 (<0.1)
Non-opioid substance use disordersb	174 (4.8)	1,364 (1.7)
One or more sedative/hypnotic fills, n (%)b,e	1,118 (31.0)	12,384 (15.2)
Number of hospital days, mean (median)b,e	1.35 (0.00)	0.66 (0.00)
Number of Emergency Dept. visits, mean (median)b,e	0.29 (0.00)	0.23 (0.00)
Number of office visits, mean (median)b,e	9.02 (6.00)	6.78 (5.00)
Number of pharmacy fills, mean (median)b,e	18.57 (15.00)	10.06 (6.00)

^aPersons included were those who had no outpatient fills for opioids (“opioid naïve”) at a KPNC pharmacy between January 1, 2008 and their initial 2011 opioid fill. KPNC: Kaiser Permanente Northern California.

^bPersons with a long-term opioid episode between index fill and December 31, 2014 were significantly different from persons who did not have a long-term opioid episode, at p<=0.05.

^cPatients receiving at least one diagnoses for these conditions in the year prior to their initial 2011 opioid fill.

^dIncludes all the conditions listed below except “any psychiatric disorder”, “opioid abuse/dependence” and “other substance abuse”.

^eMeasured in the year prior to initial 2011 opioid fill.

Numerous conditions were associated with long-term opioid use: (Table 5), including chronic pain (OR: 2.47; CI: 2.17–2.82), non-opioid substance use disorders (OR: 2.25; CI: 1.89–2.69), psychiatric disorders (OR: 1.20; CI: 1.10–1.31), and arthritis (OR: 1.41; CI: 1.31–1.52). Use of sedative/hypnotics was associated with increased odds of becoming a long-term user (OR: 1.68; CI: 1.54–1.82, versus no use). Even after adjusting for diagnosed conditions, inpatient hospital days and use of non-opioid medications in the prior year remained predictive of long-term user. On the other hand, outpatient office visits in the prior year were associated with lower odds of becoming a long-term user.

Table 5.

Predictors of Becoming a Person with Long-Term Opioid Use.^a

Characteristicb	Odds Ratio (95% Confidence Interval)c
Gender
Female	0.94 (0.87,1.01)
Male	REF
Age Group
19 – <30	0.54 (0.44,0.67)*
30 – <40	0.64 (0.53,0.77)*
40 – <50	0.79 (0.67,0.92)*
50 – <60	0.95 (0.83,1.09)
60 – <70	0.91 (0.80,1.04)
70 – <80	0.94 (0.82,1.07)
80+	REF
Race/Ethnicity
Asian	0.38 (0.33,0.44)*
Black	0.91 (0.80,1.04)
Hispanic	0.65 (0.58,0.73)*
Pacific Islander	0.73 (0.40,1.35)
Native American	0.99 (0.60,1.63)
Multi-racial	0.97 (0.83,1.13)
Other or unknown	0.49 (0.33,0.73)*
White	REF
Neighborhood deprivation
Quartile 2	1.18 (1.07,1.30)*
Quartile 3	1.24 (1.12,1.37)*
Quartile 4, most deprived	1.30 (1.15,1.47)*
Missing	1.05 (0.75,1.47)
Quartile 1, least deprived	REF
Diagnosed conditionsd
Arthritis	1.41 (1.31,1.52)*
Asthma	0.93 (0.84,1.04)
Congestive heart failure	0.81 (0.68,0.97)*
Chronic obstructive pulmonary disease	1.24 (1.09,1.41)*
Diabetes	0.91 (0.82,1.01)
Epilepsy	0.61 (0.41,0.93)*
End-stage renal disease	1.04 (0.66,1.62)
Hypertension	1.27 (1.17,1.39)*
Ischemic heart disease	0.84 (0.74,0.96)*
Osteoporosis	1.10 (0.94,1.29)
Chronic Pain	2.47 (2.17,2.82)*
Parkinson’s disease	1.08 (0.74,1.57)
Stroke	0.94 (0.76,1.17)
Psychiatric disorders	1.20 (1.10,1.31)*
Opioid abuse/dependence	1.73 (0.78,3.86)
Non-opioid substance use disorders	2.25 (1.89,2.69)*
Sedative/hypnotic fill (any vs none)e	1.68 (1.54,1.82)*
Hospital days (per 10)e	1.17 (1.09,1.26)*
Emergency Dept. visits (per 10)e	0.43 (0.18,1.00)*
Office visits (per 10)e	0.92 (0.88,0.96)*
Pharmacy fills (per 10)e	1.33 (1.29,1.37)*

^aPopulation were persons with an opioid fill in 2011 and no prior fill after January 1, 2008. Persons were also required to have been continuous KPNC members for the entire seven years from 2008 to 2014, or to have died after their initial fill. KPNC: Kaiser Permanente Northern California.

^bAlso included in the model as an adjuster variable was the KPNC clinic at which the patient received most of their health care. There were 49 such clinics; results not shown.

^cResults of a logistic regression model in which the dependent variable was a flag indicating that the patient had a long-term opioid episodes subsequent to their initial opioid fill.

^dPatients receiving at least one diagnoses for these conditions in the year prior to initial 2011 fill.

^eMeasured in the year prior to initial 2011 opioid fill.

DISCUSSION

This study developed a protocol for an EHR-based prescription opioid registry that can be used to address important research questions about prescription opioid use in non-cancer patients on a population level. The current paper also addressed initial questions about the characteristics of prescription opioid users and what predicts initiating long-term use.

Consistent with prior literature,^3,22,23 persons who used opioids were older, more likely to be white, and were more clinically complex patients, with more medical and psychiatric conditions and substance use disorders than persons not using opioids. Further, a considerable portion of patients were using opioids long-term. Patients using opioids long-term are especially important to identify, since duration of use is associated with abuse, overdose and other adverse events.^{3,16,21,24–26} Like prior studies,^22,27 we found persons with long-term opioid use to be more likely than those with shorter term use to have higher daily dosages, chronic medical or psychiatric conditions, and opioid or other substance use disorders. Given the current epidemic of misuse and overdose, identifying long-term users with population-based data can help health systems identify patients early, monitor them, and refer to specialty services (e.g. substance use treatment, pain management) as needed.

Our analysis of “new users” of opioids indicated that only 4.2% went on to use long-term within three years, although at any given time the percent of long term opioid users is quite high (25%). Although individual risk is low, at a population level this is consistent with the high level of adverse events observed in recent years.

Predictors of developing long-term use included chronic pain, sedative/hypnotic use, psychiatric disorders, and non-opioid substance use disorders. Concurrent use of sedative/hypnotics and opioids has been shown to be associated with a substantial increased risk of death from drug overdose.^8,25,26,28 Federal and health system guidelines have focused on reducing high daily dosages, and also on restricting concurrent opioid and sedative/hypnotic use.^2,29 Persons who lived in more deprived neighborhoods were also more likely to develop long-term opioid use; to our knowledge, a relationship not previously identified. The relationship of opioid prescribing and socioeconomic status (SES) has not been extensively studied. Our data do not contain information on pain severity or control. However, findings may suggest that individuals residing in more deprived neighborhoods (which may also be a proxy for individual deprivation) have more complex health status, or fewer non-medication treatment alternatives available – hypotheses deserving further study.

There is increased interest in using registries to address critical clinical and policy questions.³⁰ A goal of this project was to develop a protocol that can serve as a reference for other clinical and research teams addressing similar questions. Study algorithms can be used in health systems with pharmacy dispensation data and encounter data. For example, because our approach used the VDW, investigators from 19 other health systems in the Health Care Systems Research Network can also use the VDW to similarly address important questions about prescription opioid use. We recognize this is not without challenges, and would require adaptations, particularly for systems that have dissimilar EHR data elements or claims data. However, by sharing details about our methodology we hope to contribute to developing harmonized approaches across systems to address the opioid epidemic.

Our study has limitations. Our measures of opioid and sedative/hypnotic use depends on pharmacy dispensation data, which is commonly used in the literature, and which we consider a reasonable proxy for use. Uncertainty also exists about calculating use for overlapping fills. However, in contrast to some prior studies, we make explicit our assumptions for overlapping fills. The vast majority of KPNC members fill prescriptions at KPNC pharmacies,¹⁴ but we miss potential non-KPNC pharmacies fills. Although all registry members filled an opioid prescription at KPNC, it is possible that persons using opioids may be more likely to seek opioid prescriptions externally. Identification of medical and psychiatric conditions, and substance use disorders, is based on diagnoses recorded in the EHR as part of routine care, thus persons with more visits may have more opportunity to receive a diagnosis. Also, there can be truncation of episodes that began prior to 2008 or continued post 2014 and therefore possible under-estimation of long-term episode duration. These limitations are similar to other EHR data-based studies. Finally, generalizability to other systems may be limited, although study algorithms can be adapted.

Future analyses will leverage the prescription opioid registry and its algorithms to examine prescription opioid misuse, fatal and non-fatal overdose, and health service utilization and cost. Thus, with this same registry, we will be able to address a broad range of critical public health issues relating to prescription opioid use.

Take-away Points.

We describe a protocol for developing a prescription opioid registry using electronic health record data, and algorithms to characterize persons using prescription opioids, opioid use episodes and concurrent use of sedative/hypnotics.

• In 2011, 18% of adult members of a large integrated health plan filled at least one opioid prescription.

• Among these patients, 25% used opioids long-term, and average duration of use was more than 4 years. Sedative/hypnotics were used by 76% of long-term users.

• Being younger, white, living in a more deprived neighborhood, having a chronic pain, arthritis, COPD, hypertension, or psychiatric disorder diagnosis, and use of sedative/hypnotics were predictors of becoming a long-term opioid user.

List of abbreviations

CDC

Centers for Disease Control and Prevention

CI

95% Confidence Interval

ED

Emergency Department

EHR

Electronic Health Record

HCSRN

Health Care Systems Research Network

ICD9

International Classification of Disease, 9^th Revision, Clinical Modification

KPNC

Kaiser Permanente of Northern California

MDDE

morphine daily dose equivalent

NDC

National Drug Code

NDI

Neighborhood Deprivation Index

OR

Odds Ratio

VDW

Virtual Data Warehouse