Risk of overdose and death following codeine prescription among immigrants

Joel G. Ray, MD, MSc1,2,3

Simon Hollands, MSc1,3

Tara Gomes, MHSc1

Marcelo Urquia, PhD1,2,3

Erin M. Macdonald, MSc1

Ping Li, PhD1

Muhammad M Mamdani, PharmD, MA, MPH1,2,3

David Juurlink, MD PhD1,2

for The Canadian Drug Safety and Effectiveness Research Network

1Institute for Clinical Evaluative Sciences,University of Toronto, Toronto, Ontario

2Department of Medicine,University of Toronto, Toronto, Ontario

3Keenan Research Centre, Li Ka Shing Knowledge Institute,St. Michael’s Hospital, Toronto, Ontario

Contact:Joel G RayDepartment of Medicine, St. Michael’s Hospital30 Bond StreetToronto, OntarioM5B 1W8Tel: (416) 864-6060, Ext 77442Fax : (416) 864-5485e-mail: rayj@smh.ca

ABSTRACT COUNT: 285

MAIN WORD COUNT: 3234

1

Abstract

Background: Immigrants may be at higher risk of adverse drug reactions, in that poor language

proficiency reduces one’s understanding of drug label instructions. Additionally, there are

reports of severe or fatal toxicity due to CYP2D6 ultrarapid hepatic metabolism of codeine to

morphine among some ethnic groups, especially those from Eastern Africa.

Methods: Between 2002 and 2012 we conducted a population-based cohort study among

residents of Ontario, Canada. We used administrative health databases that linked both

immigrants and Canadian-born individuals to both prescription medication use and emergency

department visits and hospital admissions. The primary composite outcome was the risk of drug

overdose or all-cause mortality within 30 days of codeine prescription, comparing patients from

various world regions to Canadian-born individuals. A secondary analysis stratified by codeine

dose and ability to speak English and/or French.

Results: There were 553,504 individuals exclusively prescribed codeine. Relative to an

incidence rate of 57.1 per 100,000 persons-days among Canadian-born codeine recipients, those

who migrated from various World regions were at lower risk of drug overdose or death. For

example, Eastern Africans had an adjusted hazard ratio (HR) of 0.60 (95% confidence interval

[CI] 0.31 to 1.17) upon controlling for potential confounders such as age, sex, income and

physician visits. Patients unable to speak English or French who were prescribed codeine were at

lower risk of the composite outcome relative to those proficient in either language (adjusted HR

0.63, 95% CI 0.54 to 0.74).

Interpretation: Overdose and death following the institution of codeine therapy is not more

commonly observed among immigrants from world regions with a high prevalence of ultrarapid

CYP2D6 status relative to those born in Canada. Lower proficiency in English or French also did

not appear to heighten that risk.

Keywords: Codeine, opioids, immigration, ethnicity, overdose, toxicity, cytochrome P4502D6, CYP2D6 polymorphism, language ability, literacy.

2

Introduction

The avoidance of medication-related errors, including opioid overdose, is an ongoing

international priority <1,2 >. Adverse drug events are common among all age groups, even

among those who are appropriately dosed <3,4>. Patients with low literacy are more likely to

misunderstand prescription medication label instructions than those with adequate literacy <5,6>.

Some new immigrants to countries like Canada, the US and the UK may have less health literacy

because of a different mother tongue, with difficulty following verbal or written information

<7,8>. This may increase the risk of adverse drug events.

For many drugs, susceptibility to adverse drug events also has a biological basis, arising from

differences in hepatic drug metabolism. For example, several reports suggest an increased risk of

severe or fatal opioid toxicity due to enhanced conversion of codeine to morphine among some

non-white ethnic groups, resulting up to a 45-fold increase in the O-demethylated codeine

metabolites, namely, morphine, morphine-3-glucuronide and morphine-6-glucuronide <9,10>.

The enzyme responsible for the O-demethylation to morphine has been identified as cytochrome

P450 isoform 2D6 (CYP2D6) <11>. This polymorphism has a prevalence of 1 per 100 Chinese,

Japanese and Hispanic individuals, 3 per 100 African Americans, and 16 to 28 per 100 persons

of North African, Arab or Eastern African descent <12>. Since codeine is frequently

recommended as a first-choice agent for the treatment of acute or chronic pain, it is of

importance to know whether it has a differing safety profile among various ethnic groups.

There are no epidemiological analyses of the risk of fatal or severe adverse events following

codeine prescription among immigrants, or in relation to their language facility <13>.

Accordingly, we completed a large cohort study of fatal or near-fatal events in opioid-naïve

children and adults newly prescribed codeine, in relation to their world region of origin as well

as their spoken language ability.

Methods

We completed a population-based cohort study in the province of Ontario, Canada. These

individuals had universal access to physician services, hospital care and prescription drug

coverage. The study was approved by the Research Ethics Board of Sunnybrook Health Sciences

Centre, Toronto, Canada.

3

Data sources

We used administrative health databases housed at the Institute for Clinical Evaluative

Sciences (ICES) to identify all study variables. Patient level records in these data sources are

deterministically linked using a unique anonymized identifier. We used the Registered Persons

Database (RPDB) to determine date of birth, sex, postal code and death date (where applicable),

and the Canadian Institutes of Health Information Discharge Abstract Database (CIHI-DAD) to

identify all hospital admissions and discharges using the Canadian International Classification of

Diseases Version 9 (ICD-9) and Version 10 (ICD-10-CA) coding. The Ontario Health Insurance

Plan (OHIP) Database was used to capture all claims for physicians’ services, including the

claim date and diagnosis. Emergency department visits were identified using the CIHI National

Ambulatory Care Reporting System (NACRS) database from July 1, 2000 onward. The Ontario

Drug Benefit Database captures all medication prescriptions for Ontarians aged 65 years and

older, along with younger people deemed eligible for social assistance because of

unemployment, disability, high prescription drug costs relative to net household income and/or

receipt of long-term care or home care services.

The Citizenship and Immigration Canada (CIC) Database contains landing records for every

permanent legal immigrant to Canada who arrived from 1985 onward, including landing date,

country of origin, entry class (Family Class, Refugee Class, Economic Class and Other),

occupational skill level, language ability and educational level at the time of landing. The

MOMBABY dataset uses all DAD inpatient admission records of delivering mothers and their

newborns from fiscal year 2002/03 onward and links mothers and their newborns

deterministically based on the maternal/newborn chart number. The linked MOMBABY dataset

captures all live births in Ontario, such that a child’s ethnicity can be identified according to its

mother’s World region of origin. Income quintile and rurality were defined according to postal

code using Statistics Canada census data.

Participants

Between April 1, 2002 and February 28, 2012, we included individuals aged 6 months to 100

years who had OHIP eligibility for 6 months or longer and were enrolled in the Ontario Drug

Benefit Program at least 6 months preceding the index event. We excluded anyone with any prior

diagnosis (inpatient or otherwise) of drug overdose or drug dependence within 24 months of the

4

index prescription date (see Supplemental File 1 for related diagnostic codes based on the

International Classification of Diseases, 9th [ICD-9] and Tenth [ICD-10] Revisions). To create

an inception cohort of new users, we also excluded those prescribed codeine, another opioid, or

an NSAID within the 6 months preceding the index prescription date.

Exposure

Among eligible participants, we evaluated those newly prescribed oral codeine, in the form of

codeine with or without acetaminophen. Among this cohort, we assessed their World region of

birth, as follows: 1) Canada (including those who may have immigrated prior to 1985), 2)

Southeast Asia, 3) Other Western nations, 4) Caribbean, 5) Eastern Africa, 6) Western, Middle &

Southern Africa, 7) North Africa & Middle East, 8) Latin America and 9) South Asia

(Supplemental File 2). Western, Middle & Southern Africa were grouped together a priori. A

child born in Ontario was defined by the country of origin of his/her mother, since she would be

expected to fill the prescription, follow the instructions and administer the medication.

From the eligible participants not prescribed codeine, we considered, as a tracer exposure

group, those newly prescribed an oral NSAID, as a non-opioid analgesia. Among this cohort we

again assessed their World region of birth. The NSAID group was created because, like codeine,

they are prescribed as an oral analgesia, but NSAIDs are otherwise not associated with

respiratory depression, and they do not undergo CYP2D6 ultrarapid metabolism that might result

in overdose. Hence, the NSAID tracer approach allows one to explore whether the risk

associated with codeine prescription is distinct, and not shared by another commonly prescribed

non-opioid analgesic agent.

New codeine users, and separately, new NSAID users, were further grouped by their self-

described ability to speak English and/or French (Canada’s official languages) at the time arrival

of immigration to Canada. Canadian-born participants were assumed to be either English or

French speaking.

Outcomes

The main study outcome was a composite of any drug overdose or death from any cause within

30 days of index analgesic prescription. Death was determined using the RPDB, and drug

overdose was identified by an Emergency Department diagnosis or an admitting diagnosis at

5

hospitalization (see Supplemental File 1). In secondary analyses, each outcome was examined

individually.

Data analysis

Each participant was followed for 30 days from the index prescription of either codeine or an

NSAID. Individuals who changed from one drug class to the other were censored on that date to

maintain exclusive exposure to one of two drug groups. Baseline participant characteristics were

compared using standardized differences, which are more meaningful than P values for large

samples.

For the main analysis, we calculated the incidence rate of overdose or death within 30 days of

an index prescription of codeine, per 100,000 person-days. We used multivariable Cox

proportional hazards regression to derive a hazard ratio (HR) and 95% confidence interval (CI)

comparing each drug exposure-World region of origin exposure to the Canadian-born-NSAID

group (the referent). The HRs were adjusted for age at cohort entry (continuous, in years), sex,

income quintile, rural vs. urban location of residence, number of distinct medications prescribed

and physician visits within 12 months preceding cohort entry, being a child of an immigrant

mother, and prescription of a CYP2D6 inhibitor (fluoxetine, paroxetine, bupropion, quinidine,

cinacalcet, ritonavir or amiodarone) within 120 before and up to 30 days after the index

prescription of codeine or an NSAID. We repeated this same analysis for the NSAID tracer exposure

group.

We also separately analyzed the 30-day risk of overdose as well as all-cause mortality, again,

for the codeine- and NSAID-exposed cohorts.

For new codeine recipients, we stratified the main analysis by sex, age (< 50 years vs. ≥ 50

years), and codeine dose (< 30 mg or ≥ 30 mg per tablet). The 30-mg threshold was chosen

because it is the standard amount in Tylenol #3

Finally, we assessed the composite outcome among the codeine and NSAID cohorts,

respectively, comparing those who did not speak French and/or English to those who did (the

referent). The same covariates were used as in the main multivariable Cox proportional hazard

model.

All analyses were performed using SAS Version 9.3 (SAS Institute Inc. Cary, NC) and used an

alpha of 0.05 as the threshold for statistical significance. Data suppression was necessary when

6

there were five or fewer individuals per group, in compliance with institutional privacy

regulations.

Results

General cohort

We identified 553,504 persons newly prescribed codeine and 673,740 persons newly

prescribed an NSAID (Table 1). The mean age was 63.6 and 65.5 years, respectively, and 48%

of codeine users and 42% of NSAID users were male. In the preceding 12 months, there was an

appreciably higher mean (SD) number of physician visits among codeine (16.2 [15.1]) than

NSAID recipients (12.8 [11.7]) (Table 1). Among codeine recipients, 32% received a per-tablet

dose under 30 mg and 68% a per-tablet dose of 30 mg or higher, while nearly 30% of NSAID

recipients were prescribed a COX-2 inhibitor.

Overall, 50,247 of codeine (9.1%) and 77,248 of NSAID (11.5%) recipients were immigrants

from another world region or were born to a mother from that world region (Table 1). Among all

immigrants, their average duration of residence in Canada was about approximately 10 years,

and about half declared that they spoke neither English nor French at the time of immigration.

Main analysis

The 30-day rate of overdose or death was 57.1 per 100,000 persons-days among Canadian-born

persons prescribed codeine (Figure 1, squares). Relative to this group, the adjusted HR for the

30-day composite outcome was significantly lower among most immigrant groups compared to

Canadian-born persons. Exceptions to this were Eastern Africa (0.60, 95% CI 0.31 to 1.17), other

Western nations (HR 0.84, 95% CI 0.69 to 1.02) or Western, Middle and Southern Africa (HR

0.76, 95% CI 0.34 to 1.70), where no significant difference in study outcomes between groups

was observed (Figure 1). Similar effects were seen for the secondary outcomes of 30-day

overdose and all-cause mortality in the codeine group (Figure 1, squares).

Among the NSAID tracer cohort the incidence rate of the composite outcome was less frequent

than that observed in the codeine groups (Figure 1, circles). Among NSAID recipients, only the

Southeast Asian and South Asian immigrant groups were at significantly lower risk than the

Canadian-born group.

7

Sub-group analyses

The observed lower relative risk of the composite outcome among most immigrant groups

prescribed codeine was evident across all sub-group analyses (Figure 2). The lower HR was

slightly more pronounced among males (Figure 2, upper), those aged 50 years and older (Figure

2, middle), and those prescribed lower doses of codeine (Figure 2, lower).

Analysis by language

Codeine recipients who did not report speaking English or French at the time of immigration

were significantly less likely to experience the primary outcome of overdose or death (adjusted

HR 0.63, 95% CI 0.54 to 0.74) (Table 2). Among NSAID users, we also observed a lower risk of

the primary outcome among those without English or French as a spoken language compared to

those who spoke English or French (adjusted HR 0.44, 95% CI 0.42 to 0.58) (Table 2)

Discussion

In this study of 1.27 million individuals, immigrants were at lower short-term risk of overdose

or death following codeine prescription relative to Canadian-born individuals. In addition,

immigrants not proficient in English or French had a lower relative risk of adverse outcomes. In

the NSAID tracer cohort, only Southeast Asian and South Asian immigrant groups were at

significantly lower compared to their Canadian-born counterparts.

Some limitations of our work merit emphasis. Despite including 127,495 immigrants or their

offspring, there were instances in the stratified analyses in which data suppression was necessary

due to few events, or there was imprecision around the risk estimates. We only included persons

within the Ontario Drug Benefit Database. For Ontarians aged 65 years and older, this reflects

the entire eligible population. We also note that 7% of all codeine users age 50+ years were

immigrants, while the proportion of immigrants among those under age 50 years was 17.5%. As

seen in Figure 2, the number of Canadian-born persons aged 50+ years prescribed codeine was

about 4.5 times greater than that of Canadian-born persons under 50 years (N = 413,798 vs.

89,459), and for Southeast Asians it was about 5 times higher (N = 10,222 vs. 2003); yet, among

those from South Asia the corresponding ratio was only 1.4, Latin American 1.3, North Africa

and the Middle East 0.7, Eastern Africa 0.3, and the Caribbean 0.9. These differences certainly

reflect the fact that those born in Canada are older than many new immigrant groups. While the

8

majority of participants were over age 65 years, and we adjusted for age, income quintile and

rurality, we acknowledge that our data may not apply to non-seniors who purchase their

medications through self-pay or privately insured means. In addition, there may remain

unaccounted for differences between immigrant and non-immigrant individuals eligible for the

Ontario Drug Benefit Programme, related to unemployment or disability, for example. Together,

these limitations lend caution about drawing conclusions about the risk of codeine-related

overdose or death in the general adult population under age 65 years.

We excluded those prescribed an NSAID or any opioid in the preceding 6 months, or those

diagnosed with drug overdose or drug dependence within the prior 2 years. Within the 30-day

follow-up period, we also censored a participant if they were prescribed a medication from the

other exposure group, or were prescribed any non-codeine opioid. Thus, we maintained

exclusivity between the codeine and the NSAID tracer group.

We did not have data about the indication for initiation of codeine or NSAID. Thus, we cannot

exclude the possibility of confounding by indication, in which codeine or NSAID recipients from

one World region were generally more ill than those from another region. Even adjusting for the

number of physician visits does not directly deal with this issue. Since codeine is often combined

with acetaminophen (paracetamol), it is possible that some adverse outcome events herein were

related to the latter drug. A potentially toxic dose of acetaminophen in a 65 kg person -- 150

mg/kg -- would require ingestion of forty 250-mg tablets (about 10,000 mg), yet, in combination

with 15 mg of codeine per tablet would produce a 600 mg ingestion of the latter, certain to lead

to life-threatening respiratory depression.

We assumed that Canadian-born persons were fluent in English or French, and we knew

nothing about the ethnic composition of the Canadian-born persons. Among the immigrants in

our study, we only knew their declared fluency in one of these two official Canadian languages

at the time of their arrival, and not thereafter, nor did we possess information about their reading

literacy. Given that the median duration residence of the immigrants in our study was 10 years, it

can be assumed that a fair proportion gained enough language fluency to understand basic

medication instructions. Misclassification of the latter would have underestimated the true

observed association between language fluency and the risk of codeine-related overdose or

death.

9

We provide novel information about the risk of codeine-related overdose, including in relation

to race/ethnicity. A Medline search up to November 2013, limited to the title words "codeine"

and "overdose," generated just seven hits, none of which were controlled studies. In a recent

opinion piece, it was concluded "the risk of opioids stems primarily from these drugs, not from

patients." Certainly, our findings support the view that codeine too is associated with a higher

short-term risk of overdose than seen with NSAID use (Figure 1, top panel). A separate

unlimited search on Medline up to November 2013 with the terms ("codeine" and "overdose")

and ("ethnicity" or "immigration" or "immigrant" or "race") produced just one study, which

focused prescription patterns for hydrocodone and oxycodone, but not on adverse events <14>.

A limited number of studies have examined opioid prescribing patterns among racial groups in

the US, but they classified groups as White, Black, Hispanic, or Asian/Other, they did not assess

for adverse events, and codeine accounted for only 11% (n = 5600) of all opioids prescribed

<15>.

Our current findings also suggest that, compared to Canadian-born participants, the risk of

overdose or death is significantly lower among most immigrant groups identified by world

region of origin. This may conceivably be due to the healthy immigrant effect, in which

immigrants are more highly educated than the general population, and/or they maintain healthier

practices that places them at lower risk of disease. Another explanation may be that Canadian-

born individuals partake in riskier behaviors, such as codeine misuse (even without a history of

drug dependence); that they combining of codeine with other prescribed or non-prescribed drugs,

including alcohol; that they have a greater tendency to overdose on other unmeasured drugs; or

they exhibit greater suicidal behavior using prescribed medications. In the 2005 cycle of the

Ontario Student Drug Use Survey of 7th to 12th graders, both illicit drug use and hazardous

drinking was lowest among first-generation youth, which increased with second and then third

generation Canadians <16>. This is reflective of the so-called “immigrant paradox”, in which

acculturation to one’s host country is predictive of increased health risks, such as substance use,

especially among young adults <17>. As we did not have specific information about the context

of the deaths or drug overdoses documented herein, including whether they were related to

codeine (or NSAID) use, or the generational status of the Canadian-born participants, we cannot

fully explain our findings with the current dataset.

10

We did not detect a higher risk of codeine-related adverse events among those originating from

North African, Middle Eastern or Eastern African regions, and who are thought to have the

highest prevalence of the ultrarapid CYP2D6 polymorphism <12,18>. About 29% of Ethiopians,

who originate from Eastern Africa, are believed to have the ultrarapid CYP2D6 polymorphism

<16>. Yet, there are more than 75 allelic variants of the CYP2D6 polymorphism, resulting in an

increase, reduction, or complete loss of activity <19> (See Table 3 in reference 12

[http://theoncologist.alphamedpress.org/content/11/2/126.full.pdf+html]). Systematic, population-based studies of the prevalence of

the ultrarapid CYP2D6 polymorphism, and its impact on drug handling, are lacking. Since

CYP2D6 is responsible for the metabolism of many commonly prescribed drugs, including

codeine <20>, there is a basis for future research in this area, using unbiased and representational

sampling methods. If ethnicity can aid in predicting who is at highest risk of a critical adverse

event following codeine prescription, then altered prescribing of codeine and substitution with

another analgesic agent could be done efficiently and avoid costly and impractical screening for

the CYP2D6 polymorphism. Until then, no recommendations should be made about “safer” drug

prescribing among specific immigrant or native-born ethnic groups, and we refer readers to a

recent clinical practice guideline about CYP2D6 genotyping and safe codeine therapy <21>.

While immigrants in our study were not at higher risk of overdose or death following codeine

prescription, it is surprising that those without proficiency in spoken English or French were also

at lower risk. Hence, more research is needed about the value of medication instructions among

specific immigrants or groups who speak or read another primary language <22>.

Licence for Publication. The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of all authors, an exclusive licence (or non exclusive for government employees) on a worldwide basis to the BMJ Publishing Group Ltd to permit this article (if accepted) to be published in JECH and any other BMJPGL products and sublicences such use and exploit all subsidiary rights, as set out in our licence (http://group.bmj.com/products/journals/instructions-for-authors/licence-forms).

Competing Interest. None declared.

Funding. This project was supported by research funds from Canadian Drug Safety and Effectiveness Research Network (CDSERN) and by the Institute for Clinical Evaluative Sciences (ICES), which is funded by a grant from the Ontario Ministry of Health and Long-Term

11

Care (MOHLTC). The funders had no role in designing the study; collecting, analyzing, or interpreting the data; writing the report; or in the decision to submit the article for publication.

Disclaimer. This study was supported by the Institute for Clinical Evaluative Sciences (ICES), which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care(MOHLTC). The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding sources. No endorsement by ICES or the OntarioMOHLTC is intended or should be inferred.

Authors contributions: Ray: study concept, analysis and interpretation of the data, drafting of manuscript,

manuscript revision, approval of final version Hollands: analysis and interpretation of the data, drafting of manuscript, manuscript

revision, approval of final version Gomes: study design, analysis and interpretation of the data, approval of final version Urquia: study design, analysis and interpretation of the data, approval of final version Macdonald: study design, interpretation of the data, approval of final version Li: study design, analysis of the data, approval of final version Mamdani: study design, analysis of the data, approval of final version Juurlink: study concept, analysis and interpretation of the data, manuscript revision,

approval of final version

What is already known on this subject? Immigrants may be at higher risk of adverse drug reactions due to poor language proficiency or from severe or fatal toxicity due to CYP2D6 ultrarapid hepatic metabolism of codeine to morphine.

What this study adds? Immigrants were observed to be at lower short-term risk of overdose or death following codeine prescription, relative to Canadian-born individuals. Immigrants not proficient in English or French at the time of migration were at lower relative risk of overdose or death following codeine prescription. Hence, no recommendations should be made about “safer” drug prescribing among specific immigrant groups.

Figure legends

Figure 1. Risk of the composite outcome of drug overdose or all-cause mortality (top), overdose (middle) or all-cause mortality (lower) within 30 days being newly prescribed an NSAID (circles) or oral codeine (squares) according to World region of origin. Data are suppressed with 5 or fewer outcome events.

Figure 2. Risk of the composite outcome of drug overdose or all-cause mortality within 30 days being newly prescribed oral codeine, according to World region of origin, stratified by sub-groups. Data are suppressed with 5 or fewer outcome events.

12

References

<1> Institute of Medicine. Preventing Medication Errors. In Aspden P, Wolcott J, Bootman L,

Cronenwett LR, eds. Washington, DC: National Academy Pr; 2006.

<2> O'Hagan J, MacKinnon NJ, Persaud D, et al. Self-reported medical errors in seven

countries: Implications for Canada. Healthc Q 2009;12:55-61.

<3> McPhillips HA, Stille CJ, Smith D, Hecht J, Pearson J, Stull J, Debellis K, Andrade S,

Miller M, Kaushal R, Gurwitz J, Davis RL. Potential medication dosing errors in outpatient

pediatrics. J Pediatr 2005;147:761-7.

<4> Egger SS, Bachmann A, Hubmann N, Schlienger RG, Krähenbühl S. Prevalence of

potentially inappropriate medication use in elderly patients: comparison between general medical

and geriatric wards. Drugs Aging 2006;23:823-37.

<5> Davis TC, Wolf MS, Bass PF 3rd, Thompson JA, Tilson HH, Neuberger M, Parker RM.

Literacy and misunderstanding prescription drug labels. Ann Intern Med 2006;145:887-94.

<6> Tarn DM, Heritage J, Paterniti DA, Hays RD, Kravitz RL, Wenger NS. Physician

communication when prescribing new medications. Arch Intern Med 2006;166:1855-62.

<7> Bowen S. Language Barriers in Access to Health Care. Ottawa: Health Canada, 2001.

Available at http://www.hc-sc.gc.ca/hcs-sss/alt_formats/hpb-dgps/pdf/pubs/2001-lang-acces/

2001-lang-acces-eng.pdf

<8> Zanchetta MS, Poureslami IM. Health literacy within the reality of immigrants' culture and

language. Can J Public Health 2006;97 Suppl 2:S26-30.

<9> Gasche Y, Daali Y, Fathi M, Chiappe A, Cottini S, Dayer P, Desmeules J. Codeine

intoxication associated with ultrarapid CYP2D6 metabolism. N Engl J Med 2004;351:2827-31.

<10> Kelly LE, Rieder M, van den Anker J, Malkin B, Ross C, Neely MN, Carleton B, Hayden

MR, Madadi P, Koren G. More codeine fatalities after tonsillectomy in North American children.

Pediatrics 2012;129:1343-7.

<11> Caraco Y, Tateishi T, Guengerich FP, Wood AJ. Microsomal codeine N-demethylation:

cosegregation with cytochrome P4503A4 activity. Drug Metab Dispos 1996;24:761-4.

<12> Bernard S, Neville KA, Nguyen AT, Flockhart DA. Interethnic differences in genetic

polymorphisms of CYP2D6 in the U.S. population: clinical implications. Oncologist

2006;11:126-35.

13

<13> Warner A, Menachemi N, Brooks RG. Health literacy, medication errors, and health

outcomes: is there a relationship? Hosp Pharm 2006;41:542-51.

<14> Wisniewski AM, Purdy CH, Blondell RD. The epidemiologic association between opioid

prescribing, non-medical use, and emergency department visits. J Addict Dis 2008;27:1-11.

<15> Pletcher MJ, Kertesz SG, Kohn MA, Gonzales R. Trends in opioid prescribing by

race/ethnicity for patients seeking care in US emergency departments. JAMA 2008;299:70-8.

<16> Hamilton HA, Noh S, Adlaf EM. Adolescent risk behaviours and psychological distress

across immigrant generations. Can J Public Health 2009;100:221-5.

<17> Bui HN. Racial and ethnic differences in the immigrant paradox in substance use. J Immigr

Minor Health 2013;15:866-81.

<18> Aklillu E, Persson I, Bertilsson L, Johansson I, Rodrigues F, Ingelman-Sundberg M.

Frequent distribution of ultrarapid metabolizers of debrisoquine in an Ethiopian population

carrying duplicated and multiduplicated functional CYP2D6 alleles. J Pharmacol Exp Ther

1996;278:441-6.

<19> Ingelman-Sundberg M, Evans WE. Unravelling the functional genomics of the human

CYP2D6 gene locus. Pharmacogenetics 2001;11:553-4.

<20> Ma MK, Woo MH, McLeod HL. Genetic basis of drug metabolism. Am J Health Syst

Pharm 2002;59:2061-9.

<21> Madadi P, Amstutz U, Rieder M, Ito S, Fung V, Hwang S, Turgeon J, Michaud V, Koren

G, Carleton BC; CPNDS Clinical Recommendations Group. Clinical practice guideline:

CYP2D6 genotyping for safe and efficacious codeine therapy. J Popul Ther Clin Pharmacol

2013;20:e369-96.

<22> Weiss L, Gany F, Rosenfeld P, Carrasquillo O, Sharif I, Behar E, Ambizas E, Patel P,

Schwartz L, Mangione R. Access to multilingual medication instructions at New York City

pharmacies. J Urban Health 2007;84:742-54.

14

Table 1. Baseline characteristics of participants in the codeine and NSAID exposure groups

Characteristic*Codeine

(N = 553,504)NSAID

(N = 673,740)Standardized

DifferenceAge, years

Mean (SD) 63.6 (20.7) 65.5 (17.2) 0.1≥ 50 445,067 (80.4) 569,248 (84.5) 0.11≥ 66 379,378 (68.5) 482,238 (71.6) 0.07

Male 264,332 (47.8) 285,111 (42.3) 0.11Income quintile (Q)

Q1 (lowest) 136,315 (24.7) 163,244 (24.3) 0.01Q2 116,720 (21.2) 143,082 (21.3) 0.0Q3 102,253 (18.5) 127,219 (19.0) 0.01Q4 97,585 (17.7) 120,579 (18.0) 0.01Q5 (highest) 98,541 (17.9) 117,087 (17.4) 0.01

Rural location 72,891 (13.2) 95,318 (14.2) 0.03No. physician visits, 1 year prior to cohort entry

Mean (SD) 16.2 (15.1) 12.8 (11.7) 0.26≥ 1 Visit 546,085 (98.7) 664,131 (98.6) 0.01

World region of originCanada 503,257 (90.9) 596,492 (88.5) 0.08Southeast Asia 12,225 (2.2) 17,771 (2.6) 0.03Other Western nations 9,342 (1.7) 11,514 (1.7) 0.0Caribbean 3,547 (0.60) 4,612 (0.70) 0.01Eastern Africa 2,962 (0.50) 3,829 (0.60) 0.0Western, Middle & Southern Africa 1,047 (0.20) 1,375 (0.20) 0.0North Africa & Middle East 6,011 (1.1) 9,627 (1.4) 0.03Latin America 3,716 (0.70) 5,674 (0.80) 0.02South Asia 11,397 (2.1) 22,846 (3.4) 0.08

No. (%) of offspring born to an immigrant mother 3,810 (0.70) 2,028 (0.30)No. medications prescribed within 1 year prior to cohort entry

Mean (SD) 4.9 (4.9) 4.4 (4.3) 0.1≥ 1 prescription 393,102 (71.0) 498,667 (74.0) 0.07

Prescription of a recognized CYP2D6 inhibitor 120 before and up to 30 days after cohort entry

28,269 (5.1) 28,315 (4.2) 0.04

NSAID typeCOX-2 -- 198,282 (29.4)Non-COX-2 -- 475,458 (70.6)

Mean (SD) estimated codeine dose per tablet (mg) 25.7 (7.5) --Low Dose (< 30 mg) 176,029 (31.8) --High Dose (≥ 30 mg) 377,475 (68.2) --

15

Characteristic*Codeine

(N = 553,504)NSAID

(N = 673,740)Standardized

DifferenceMedian (IQR) length of time since arrival to Canada

Median (IQR) 10 (4-15) 9 (4-14) 0.08Language ability at the time of immigration***

English only 23,926 (48.5) 38,214 (47.7) 0.03French only 526 (1.1) 908 (1.1) 0.01Both French and English 617 (1.3) 980 (1.2) 0Neither language 24,256 (49.2) 39,951 (49.9) 0.06

Level of education at the time of immigrationSecondary or Less 34,925 (70.8) 56,381 (70.4) 0.01Any education above secondary school 14,403 (29.2) 23,673 (29.6) 0.01

*All data are presented as a number (%) unless otherwise indicated

** Limited only to immigrants born outside of Canada

SD standard deviation; IQR interquartile range

16

Table 2. Risk of the composite outcome of all-cause mortality or diagnosed drug overdose within 30 days of being newly prescribed oral codeine (red font) or an NSAID (black font), according to language ability at the time of immigration

Composite outcome up to 30 days after prescription

Exposure status†Hazard ratio

(95% confidence interval)

Drug Spoken language*

No. (incidence rate per 100,00 person-days) Unadjusted Adjusted**

NSAID English and/or French 2,584 (13.6) 1.00 (referent) 1.00 (referent)No English or French 46 (3.9) 0.29 (0.21 to 0.38) 0.44 (0.42 to 0.58)

Codeine English and/or French 8,691 (55.4) 1.00 (referent) 1.00 (referent)No English or French 174 (23.1) 0.42 (0.36 to 0.49) 0.63 (0.54 to 0.74)

*Language ability in one of Canada’s two official languages, namely, English and/or French. The level of language ability is self-assessed. Canadian-born (non-immigrant) participants were assumed to speak English and/or French, as were their children. Canadian-born children of immigrant mothers were assigned the language of their mothers.

**Adjusted for age, sex, income quintile and rural location at the cohort entry date, number of distinct drugs used in the past 1 year, number of physician visits in the past 1 year, and prescription of a recognized CYP2D6 inhibitor 120 days before and up to 30 days after the index prescription of codeine or an NSAID.

†The interaction term between medication group and language ability was statistically significant (p=0.047) within the adjusted model.

17