Study Design and Population
We conducted a population-based, retrospective cohort study in Ontario, Canada. Our cohort consisted of all residents who moved into a LTC home between April 1, 2010 and February 28, 2019. We used the first assessment performed during the study period to define the cohort and determine residents’ baseline characteristics (index assessment). We excluded those who: 1) were younger than 66 or older than 105 years at the index assessment, 2) were not eligible for the Ontario Health Insurance Plan (OHIP) in the 2 years preceding the index assessment, 3) had an invalid death date (i.e., death date occurred before index assessment). Residents were followed for 1 year or until death, whichever occurred first. This study complied with the Reporting of studies Conducted using Observational Routinely-collected health Data (RECORD) guidelines (Appendix 1).34
Data Sources
We used administrative databases at ICES (ices.on.ca), an independent, nonprofit research institute whose legal status under Ontario’s health information privacy law allows it to collect and analyze health care and demographic data, without consent, for health system evaluation and improvement. We obtained data on LTC residents through the Continuing Care Reporting System (CCRS), which collects information on all LTC residents using the Resident Assessment Instrument Minimum Dataset (RAI-MDS), version 2.0.35 All LTC residents undergo a complete assessment annually, and abbreviated assessments are performed every 3 months (or at more frequent intervals if there is a significant change in health status).
We linked the CCRS to numerous administrative databases using unique encoded identifiers. The Registered Persons Database (RPDB) provided residents’ age, sex and postal code. We linked resident and LTC home postal codes to the 2016 Statistics Canada Census to obtain neighbourhood income quintile and urban/rural status (for both the resident’s dwelling prior to entry into LTC, and the LTC home). The Immigration, Refugees and Citizenship Canada (IRCC) Permanent Resident’s Database was used to identify immigrants who became permanent residents after 1985 (and up until May 2017). We ascertained chronic conditions using algorithms validated by ICES and applied in previous studies (see Appendix 2). Finally, the Ontario Drug Benefit (ODB) database was used to identify all medications prescribed to patients who are covered by the ODB program (i.e., all OHIP eligible residents over the age of 65, receiving home care services, or living in LTC homes).36 These datasets were linked using unique encoded identifiers and analyzed at ICES.
Exposure
We obtained resident language using the primary language collected during resident assessments and recorded in CCRS. During resident assessments, interviewers are instructed to determine the resident’s primary language by listening, observing and, if necessary, asking the resident (or their family member or care provider) to specify their primary language. We defined Anglophones and Francophones as residents’ whose primary language was English and French, respectively, while we defined all other residents as Allophones (which is a term used by Statistics Canada to identify Canadians whose mother tongue is a language other than English or French).22
Next, we derived the language of the LTC home by determining the proportion of residents belonging to each linguistic group within individual LTC homes. For consistency with previous work,37 we derived the language of the home using the frequency of resident languages within each LTC home. To do so, we calculated the person-time representation of each linguistic group (English, French, and all Allophone languages) as a proportion of all languages spoken by residents in the home. We divided the total person-days for each linguistic group by the overall person-days (i.e., overall length of stay in days during the study period). We defined LTC homes as French homes when Francophone residents contributed more than 25% of the person-days, and Allophone homes when Allophone residents contributed more than 50% of the person-days. All remaining LTC homes were defined as English homes. Prior analyses performed by our group revealed that a threshold of 25% had better sensitivity and specificity (85% and 97%, respectively) when compared to a threshold of 50% for the purposes of identifying homes that were designated by law to provide services in French, as per the French Language Services Act.38
Finally, we determined whether each resident was living in a language concordant or discordant LTC home by considering the interaction between the resident’s language and the language of the LTC home. Residents who lived in a LTC home where the language of the LTC home corresponded to their primary language were said to be receiving language-concordant care, while all other residents were said to be receiving language-discordant care.
Covariates
We identified the following baseline covariates: age at time of index assessment, sex, immigration status, neighbourhood income quintile of the resident prior to entry into LTC, urban or rural residence of the resident prior to entry into LTC, geographic region of the resident prior to entry into LTC, geographic region of the LTC home, number of chronic conditions, mental health-related conditions (anxiety, delusions, dementia, depression, hallucinations), active prescription of antipsychotic at time of index assessment (yes/no), number of medications used in the last 7 days (Table 1).
Table 1
Individual level characteristics of the cohort, by language group (n = 198,729)
| Anglophones (n = 162,814) | Francophone (n = 6,230) | Allophones (n = 29,685) |
Age | Mean (SD) | 84.3 (7.5) | 83.8 (7.3) | 85.1 (6.78) |
Median (Q1-Q3) | 85 (79–90) | 84 (79–89) | 86 (81–90) |
Sex | Female | 105,659 (64.9%) | 4,076 (65.4%) | 19,385 (65.3%) |
Male | 57,155 (35.1%) | 2,154 (34.6%) | 10,300 (34.7%) |
Immigration Status (since 1985) | Yes | 2,627 (1.6%) | 50 (0.8%) | 6,191 (20.9%) |
No | 160,187 (98.4%) | 6,180 (99.2%) | 23,494 (79.1%) |
Resident Neighbourhood Income Quintile Before moving to LTC | 1 – Lowest | 45,455 (27.9%) | 1,497 (24.0%) | 8,682 (29.2%) |
2 | 35,072 (21.5%) | 1,359 (21.8%) | 6,727 (22.7%) |
3 | 31,767 (19.5%) | 1,348 (21.6%) | 5,456 (18.4%) |
4 | 26,197 (16.1%) | 1,092 (17.5%) | 4,565 (15.4%) |
5 – Highest | 22,573 (13.9%) | 844 (13.5%) | 3,963 (13.4%) |
Missing Data | 1,750 (1.1%) | 90 (1.4%) | 292 (1.0%) |
Resident Rurality Before moving to LTC | Urban | 137,274 (84.3%) | 4,695 (75.4%) | 28,609 (96.4%) |
Rural | 24,055 (14.8%) | 1,464 (23.5%) | 804 (2.7%) |
Missing Data | 1,485 (0.9%) | 71 (1.1%) | 272 (0.9%) |
Geographic Region of Ontario of Residence Before moving to LTC | Eastern | 14,508 (8.9%) | 3,196 (51.3%) | 1,219 (4.1%) |
Northern | 13,697 (8.4%) | 2,048 (32.9%) | 1,125 (3.8%) |
Southwestern | 134,024 (82.3%) | 971 (15.6%) | 27,221 (91.7%) |
Missing Data | 585 (0.4%) | 15 (0.2%) | 120 (0.4%) |
Number of medications used in past 7-days | Mean (SD) | 10.1 (4.6) | 10.8 (4.7) | 9.5 (4.4) |
Median (Q1-Q3) | 10 (7–13) | 10 (8–14) | 9 (6–12) |
Active Prescription of AP medication at index | Yes | 41,817 (25.7%) | 1,737 (27.9%) | 8,201 (27.6%) |
No | 120,997 (74.3%) | 4,493 (72.1%) | 21,484 (72.4%) |
Number of Chronic Conditions | 0 | 265 (0.2%) | 9 (0.1%) | 55 (0.2%) |
1 | 1,922 (1.2%) | 67 (1.1%) | 310 (1.0%) |
2 | 5,865 (3.6%) | 230 (3.7%) | 1,028 (3.5%) |
3+ | 154,762 (95.1%) | 5,924 (95.1%) | 28,292 (95.3%) |
Mental Health | Anxiety | 14,404 (8.8%) | 885 (14.2%) | 1,936 (6.5%) |
Depression | 35,415 (21.8%) | 1,575 (25.3%) | 5,820 (19.6%) |
Delusions | 6,694 (4.1%) | 204 (3.3%) | 731 (2.5%) |
Hallucinations | 3,815 (2.3%) | 120 (1.9%) | 539 (1.8%) |
Dementia | 105,117 (64.6%) | 4,160 (66.8%) | 19,292 (65.0%) |
Outcomes
Our primary outcome was the occurrence of any PIP-AP during the first year after index assessment (i.e., after April 1, 2010 and before February 28, 2020). PIP-AP was defined according to the 2015 STOPP-START criteria (specifically criterion K2),39 which identifies as potentially inappropriate the use of any neuroleptic (antipsychotic) medications, as they may cause gait dyspraxia or parkinsonism in the elderly.40,41 This study did not make use of the Beers criteria,17 as they are not sufficiently specific to assess antipsychotic use alone. The coding strategy used to identify PIP-AP from administrative databases is described elsewhere.42 Of note, the coding strategy was designed to identify only antipsychotics that were dispensed, thereby minimizing the risk of identifying standardized prescriptions that were not administered to the patient.42
Statistical Analysis
We performed descriptive analyses to compare resident characteristics and outcomes across linguistic groups. The risk of PIP-AP based on linguistic factors was assessed using multivariable logistic regression analysis. We ran two separate models with the following exposures: 1) resident language and facility language, 2) resident language, facility language, and resident-facility language concordance/discordance. Adjusted analyses included the potential confounders of age, sex, neighbourhood income quintile of the resident prior to entry into LTC, urban or rural residence of the resident prior to entry into LTC, geographic region of the LTC home (Eastern Ontario, Northern Ontario, Southwestern Ontario), immigration status, number of chronic conditions, diagnosis of dementia, mental health-related conditions (anxiety, delusions, depression, dementia, hallucinations), medication use in the last 7 days. We performed complete case analysis by excluding all observations with missing data for any of the variables included in the regression analysis (n = 26). Statistical tests were 2-tailed and the significance threshold was α = 0.05.