The impact of nursing home residency on psychotropic drug use in major neurocognitive disorder: A nationwide comparison

Psychotropic drugs are utilized against neuropsychiatric symptoms among people with major neurocognitive disorder (NCD) despite well‐documented risks, and older people in nursing homes are expected to be more frequently exposed to those medicines. This study compared psychotropic drug use and associated factors between older people with major NCD and matched references.


| INTRODUCTION
In a previous study based on data from the Swedish registry for cognitive/dementia disorders (SveDem), 1 we found notably high proportions of psychotropic drug use (especially antidepressants and zopiclone) in this population, but also worrying associations between antipsychotics and certain major neurocognitive disorder (NCD) subtypes.Proportions of antipsychotic use in the range of 20%-40% have previously been observed among people with major NCD in Scandinavian nursing homes, 2,3 and living in a Swedish nursing home has been identified as a risk factor for overall psychotropic drug use. 4e reasons for utilizing antipsychotics and anxiolytics in that context seem partly due to caregiver distress and staff shortage. 5,68][9] Among anxiolytics and hypnotic drugs, benzodiazepines are associated with an increased risk for falls and fractures, as well as delirium and other types of cognitive impairment. 10,11The pharmacologically related Z-drugs (e.g.zopiclone or zolpidem) are associated with similar risks for falls and fractures. 12,13e relationship between and depression and age is complicated, 14,15 especially in the context of cognitive impairment, [16][17][18] but approximately 15% of older individuals who live in community housing present clinically relevant depressive symptoms, a percentage that is probably even higher in nursing homes and hospital settings. 19portantly, antidepressants are a treatment option in the management of many neuropsychiatric symptoms that often emerge during the progression of major NCD. 20,21Use of antidepressants is one of the strongest predictors of fall-related injuries, although causality is uncertain 22,23 .Moreover, use of selective serotonin reuptake inhibitors (SSRIs) is associated with hyponatremia, 24 a condition for which older individuals can be considered an especially high-risk group due to other age-related factors and concomitant use of diuretics. 25There are also some clinically relevant interactions to be considered regarding SSRIs, for example, increased risk of bleeding 26 and prolonged QT interval. 27In summary, psychotropic drug use is associated with considerable risks, especially in cases of major NCD, but uncertainties remain about whether and to what extent people with major NCD are particularly exposed, or whether our earlier findings simply reflect an extensive utilization of psychotropics among older Swedish people in general.Since the prevalence of cognitive impairment is expected to be higher in nursing home, it is also relevant to consider how type of residency affects the association between major NCD and psychotropic use.In the present study, we aimed to compare the use of psychotropic drugs between older people with major NCD and reference individuals from the Swedish Total Population Register.Our primary objective was to investigate whether the proportions of psychotropic drug users are different in the population of older people diagnosed with major NCD than in a reference population matched by age and sex but without records of the disorder.As a second objective, we investigated subgroup and interaction effects regarding major NCD, nursing home residency and psychotropic drug use.

| Study population
The study population comprised all individuals aged 65 years or older on 31 December 2019 who could be found in at least one of the following three databases from 2007 to 2019: (1) SveDem, including both basal registrations and the specific nursing home module; (2) the National Patient Register, registered with ICD-codes F00, F01, F02, F03, F09, F10, G30 or G31 (i.e.those associated with various subtypes of major NCD or unspecified psychiatric disorder; 3) the Swedish Prescribed Drug Register, based on any prescriptions filled for antidementia drugs, that is, subgroup N06D according to Anatomic Therapeutic Chemical (ATC) classification.References were randomly matched 1:1 by age and sex from the Total Population Register at the same year as their pair-wise index-date for major NCD (i.e.earliest date of either a major NCD diagnosis or filled prescription for antidementia drugs).Prior to the sampling, people with certain cognitive, personal, behavioural or neurodegenerative conditions covered by ICD-codes F05, F06, F07 and G32 had been excluded from the total population selection frame, to avoid the inclusion of uncertain cases with symptoms similar to manifestations of NCDs in the reference population.The initial dataset was filtered through the National Cause of Death Register to exclude any persons who had died before 1 January 2020.

| Drug use, comorbidities and nursing home residency
The study population dataset was inter-linked with the National Patient Register, the Swedish Prescribed Drug Register and the register regarding social service efforts for older and/or disabled people to obtain information about diagnoses, drug use and nursing home residency, respectively.All drug dispensations and social service efforts were reshaped from a row-format for each event into dichotomized variables; missing data indicated that no diagnostic coding, drug dispensation or nursing home stay had occurred.Drug dispensation data was sorted by individual substances according to the ATC system, and we dichotomized drug use based on at least one dispensation from 1 July 2019 through 31 December 2019.We applied the same principle and ATC classification to quantify proportions of psychotropic drug for the following subgroups: antipsychotics (N05A), anxiolytics (N05B), sedatives and hypnotics (N05C), and antidepressants (N06A).
We defined nursing home residents as persons with at least one record of nursing home stay from 1 July 2019 to 31 December 2019.

| Statistics
Psychotropic drug users and comorbidities were summarized and presented as proportions.Pearson's chi-squared test or independent sample t-test were used to evaluate differences in proportions and means, respectively.Generalized estimating equations (GEEs) using binary logistic models were used to assess the relation of age, sex, diagnosed major NCD and nursing home residency to psychotropic drug use.An exchangeable correlation matrix was used within each matched pair, and in the first analysis, we assessed interaction between the factors major NCD and nursing home residency.Interactions between those two variables were found, and we subsequently performed stratified analyses for people in nursing homes and ordinary homes.Results from the regression analyses are presented as odds ratios (ORs) with 95% confidence intervals and pvalues.The cleaning of the registry data was carried out in Stata 17.
During the subsequent data handling, IMB Statistics SPSS version 28 was used for descriptive and analytical statistics.

| Ethical considerations
The study was approved by the Swedish Ethical Review Authority (registration number 2020-04663).

| RESULTS
The mortality rate was higher for individuals with major NCD: in total, 119,717 persons who met the inclusion criteria in 2007-2019 had survived until 31 December 2019, while the number of surviving reference individuals was 210,509 (population characteristics available in eTable 1).A total of 102,419 complete matching pairs (Table 1) remained for the primary analysis.Individuals with major NCD had more comorbidities on average than did the reference group, but the most notable differences were an almost ten-fold higher occurrence in nursing home residents and five times as many with a dose-dispensing service among people with major NCD.
More than twice as many individuals with major NCD had filled prescriptions for any psychotropic drug than had the reference population.The proportions of psychotropic drug users (Table 2) were also higher among people with major NCD for all ATC subgroups and most individual drugs (the exceptions being amitriptyline and zolpidem).The most common psychotropic drugs overall were zopiclone followed by mirtazapine with total user proportions of 13.6% (n = 27,862) and 12.9% (n = 26,499), respectively.There were significantly higher proportions of psychotropic drug users in the female population for all ATC subgroups (eTable 2).The proportions of psychotropic drug users in different major NCD subtypes are available as supplementary material (eTable 3).
The initial GEE analysis showed that psychotropic drug use was positively associated with female sex, major NCD and nursing home residency, regardless of ATC subgroup (eTable 4).For all ATC subgroups, there were negative interaction terms in the range of 0.28-0.60 between major NCD and nursing home residency.When analysing residents of nursing home and ordinary homes separately (Table 3), the adjusted ORs concerning major NCD and psychotropic drug use were positive for all ATC subgroups (except for sedatives and hypnotics), but consistently lower among nursing home residents.In that comparison, the most apparent difference was between the ORs regarding major NCD and antipsychotic use when comparing residents of nursing homes and ordinary homes.Based on the underlying numbers of antipsychotic drug users, the stronger association in ordinary homes could be explained by a higher proportion of antipsychotic users among nursing home residents in the reference group compared with reference individuals using antipsychotics and living in ordinary homes (Figure 1).

| Proportions of psychotropic drug users
There was an obvious difference in psychotropic drug use between older people with and without major NCD.This observation is in line with the presumption that management of neuropsychiatric symptoms to a great extent relies on pharmacological treatment options.
As expected, the relative difference was also more prominent for antipsychotics, a drug class which has few other indications for older people apart from schizophrenia treatment.However, even after subtracting all cases of schizophrenia and schizoaffective syndrome (n = 299) from the reference group, more than 1100 antipsychotic users remained.As a plausible explanation, it should be noted that antipsychotics can, in certain situations, be used for other indications; suspect that the reference group included an undisclosed number of NCDs that had not been fully diagnosed, which on one hand could be seen as potential contamination of the reference group, but nevertheless is an important observation.
Mirtazapine was the most common individual antidepressant in both groups.Given its wide receptor affinity and favourable safety profile, mirtazapine can be a seemingly rational pharmacological alternative for managing both sleep disturbances and depressive symptoms in older people.However, for symptoms that are mainly specific to major NCD (e.g.agitation), a recent clinical trial of mirtazapine failed to demonstrate that it was more effective than placebo, 29 and the sparse evidence regarding antidepressants for agitation is mainly limited to SSRIs. 30,31Because individuals can have differential dose responses and the sedative effects of mirtazapine often appear to be temporary, 32 it is essential to have a clear indication for therapy and follow-up before initiating therapy with this drug.Of all the drugs listed in Table 2, only amitriptyline and zolpidem were more commonly used in the reference group than in the group with major NCD.Since both amitriptyline and zolpidem can trigger and worsen cognitive impairment, [33][34][35] it is likely that physicians pay more attention to such adverse effects in patients with confirmed cognitive impairment.The use of Z-drugs was extensive in both groups, which might signal a worrisome overuse among older people in general.
Regarding subtypes of major NCD, there was a large proportion of unspecified cases, but it is nonetheless evident that certain subtypes contributed more than others to the overall results.For example, the category that included Lewy body dementia and frontotemporal dementia was especially exposed to antipsychotics (eTable 3), which reflects the neuropsychiatric symptoms that might need to be managed in those subtypes, for example, visual hallucinations and wandering behaviour in Lewy body dementia and frontotemporal dementia, respectively. 36Similarly, depressive symptoms can be considered a clinical characteristic of vascular dementia, 37 and that specific subtype accounted for relatively more antidepressant users than did the other NCDs.Disease severity is, although not applicable for the reference population, another relevant factor that deserves to be mentioned.We have, for example, previously observed that lower baseline Mini Mental State Exam scores were positively associated with antipsychotic and anxiolytic drug use. 1 As a final comment on user proportions, the 'drug of choice' for each ATC subgroup was the same in both comparison groups, and also in line with national treatment guidelines for older people. 380][41] Our results confirm that psychotropic drug use is more common in the nursing home setting and, except for sedative and hypnotic drugs, positively associated with major NCD, regardless of residency.Moreover, as interpreted from Table 3 and partly visualized in Figure 1, the differences related to major NCD are even more pronounced among individuals who remain in ordinary homes, especially concerning antipsychotics.Here, it should be acknowledged that reference individuals in ordinary homes could differ considerably from those in nursing homes.
Reference individuals in ordinary homes could in theory have been highly functioning individuals, both socially and physically, free of any chronic diseases and associated medicines.Still, differences in comorbidities between people in nursing home and ordinary homes as an explanation for our observations on psychotropic drug use should be most pertinent for the results of antidepressants, anxiolytics, and sedatives and hypnotics-not antipsychotics.One worrisome interpretation of our findings could thus be that a substantial   number of older individuals in nursing homes have not formally been diagnosed with major NCD but are nonetheless being treated for various burdensome neuropsychiatric symptoms.Although the specific symptoms should reasonably guide the choice of treatment, it would be difficult to weigh the risks associated with a specific pharmacological agent if the underlying subtype of major NCD has not been fully examined.For example, people with Alzheimer's disease and Lewy body dementia may both present with grave psychotic symptoms (e.g.hallucinations) and experience symptom relief from antipsychotics; however, the risks associated with such drugs would theoretically be far worse for the latter patient group. 42garding the covariates in Table 3, higher age was negatively associated with psychotropic drug use among nursing home residents, a result that was consistent for all ATC subgroups.However, the association was partly in opposite direction among people in ordinary homes.One plausible explanation for this finding could be that neuropsychiatric symptoms do not emerge linearly with disease progression; rather, they increase and eventually decline in the endstage of the NCD. 43,44Those in nursing homes might had moved there because of severe cognitive deficits and burdening neuropsychiatric symptoms, hence in a later stage of cognitive decline, a point from which the occurrence of neuropsychiatric manifestations would on average decrease.

| Strengths
A major strength of the present study is the combination of three different data sources for inclusion, which together provide an almost complete coverage of individuals diagnosed with major NCD in Sweden.The only cases of clinically diagnosed major NCD not included in our dataset would be individuals who were not registered in SveDem, who did not have records of antidementia drugs or had been diagnosed by primary care units exclusively.Although we cannot know whether the medicines have ultimately been taken by each individual, using dispensing records was, compared to medication lists and prescribing data, probably as close as possible to measure drug use in this type of extensive dataset.

| Limitations and risk of bias
The references were matched based on sex, age and first date of either a major NCD diagnosis or prescription fill of antidementia drugs, and it should be noted that drug dispensing data for the reference group is not fully representative of older people in general.
A random sample from the total population of older people would likely have resulted in younger references on average, since age is generally considered to be the strongest known risk factor for developing major NCD. 45There are also uncertainties concerning our methodological need to dichotomize a concept as complex as drug use.First, some psychotropics are usually administrated according to a daily dosage regimen while others are used only when needed, and we had no information regarding the total drug supply used during the 6-month period.Second, the Swedish Prescribed Drug Register provide no information regarding indication for treatment and is therefore not optimal for assessing the appropriateness of psychotropic medication use, since these drugs can theoretically be utilized for variety conditions and symptoms.Furthermore, we had no information on non-pharmacological treatment during the time period of interest.When we analysed nursing home residents and people in ordinary homes separately, the matching pairs were inevitably disrupted; however, that should probably be of minor importance given the size of the data set and the fact that the study did not primarily aim to estimate causal relationships. 46
for example, quetiapine can be an alternative for treating depression in bipolar disorder and was used by 196 references.Other cases of antipsychotics in the reference group might be attributed to reminiscent psychiatric disorders (e.g.very late-onset schizophrenia-like psychosis28 ) that are not covered by the International Classification of Diseases and Health Related Problems (ICD-10), and are consequently not documented in the Swedish Patient Register.Still, we KINDSTEDT ET AL.
Despite well-documented risks in people with cognitive impairment, psychotropic drug use was overall high and positively associated with both major NCD and nursing home residency.Taken together, interventions to better target neuropsychiatric symptoms in older people are warranted.Hypnotic drug use among older people in general as well as antipsychotic drug exposure among older people in nursing homes appear to be two important focus areas.AUTHOR CONTRIBUTIONSJonas Kindstedt wrote this manuscript.Jonas Kindstedt and MariaGustafsson developed the original idea for the study and planned the principal study protocol.Jonas Kindstedt and Marie Lindkvist performed the descriptive and analytical statistics.All listed authors provided feedback regarding the final study design, contributed to the revision of the study protocol, were involved in the review process of the manuscript, approved the final version and will ensure that issues related to the accuracy or integrity of the work are investigated and resolved.
Study population characteristics.
Abbreviations: IQR, interquartile range; SD, standard deviation.a Age refers to number of birthdays.b The same individual could have more than one documented subtype.c ICD-codes F02 and/or G31 (e.g.dementia with Lewy bodies, frontotemporal dementia or Parkinson's disease dementia).d ICD-codes F03 and/or F09 (e.g.alcohol-related dementia).
Regression analyses according to type of residency of psychotropic drug use, including major neurocognitive disorder (NCD), age and sex as independent variables.
T A B L E 3