CLINICAL INVESTIGATION

Comparing Three Methods for Reducing Psychotropic Use in

Older Demented Spanish Care Home Residents
William B. Weeks, MD, PhD, MBA,* Manish K. Mishra, MD, MPH,* David Curto, MD, MBA,†
Curtis L. Petersen, MPH,* Pedro Cano, MD,† Yulin Hswen, MPH,‡
Silvia Villamarín Serra, MSc,† Glyn Elwyn, MD, PhD, MSc,*
Marjorie M. Godfrey, PhD, MS, BSN,* Pedro Sánchez Soro, MSc, MBA,† and
José Francisco Tomás, MD†

and drug class–specific MEDDs. Compared to controls, patients

BACKGROUND/OBJECTIVE: In nursing homes across
the world, and particularly in Spain, there are concerns that
psychotropic medications are being overused. For older
Spanish nursing home residents who had dementia, we
sought to evaluate the association between applying inter-
ventions designed to reduce inappropriate psychotropic
medication use and subsequent psychotropic use.
DESIGN: Retrospective, propensity score–matched, con-
trolled, patient-level observational analysis.
SETTING: A total of 45 nursing homes in Spain.
PARTICIPANTS: A total of 1653 nursing home residents,
aged 70 to 99 years, who had dementia and were pre-
scribed an antipsychotic, anxiolytic, or antidepressant medi-
cation, 606 of whom received an intervention; the
remainder served as propensity score–matched controls.
INTERVENTION: Team Rounds, Screening Tool of Older
Persons’ Prescriptions (STOPP)/Screening Tool to Alert
Doctors to Right Treatment (START) criteria, or a Patient
Decision Aid.
MEASUREMENTS: At 2 and 4 weeks following interven-
tion: change from baseline drug class–specific milligram-
equivalent daily dose (MEDD); at 2 weeks: patient falls and
restraint use.
RESULTS: Within each intervention/drug-class cohort, inter-
vention patients and matched controls had similar baseline demo-
graphic characteristics, Charlson scores, lengths of admission,
exposed to Team Rounds experienced a 23.3% (95% confidence
interval [CI] = 13.9%-32.8%) reduction in antipsychotic and a
23.1% (95% CI = 18.3%-28.0%) reduction in anxiolytic
MEDDs; those exposed to Patient Decision Aids had a 24.8%
(95% CI = 15.6%-33.9%) reduction in antipsychotic and a
31.8% (95% CI = 25.5%-38.2%) reduction in anxiolytic
MEDDs; and those exposed to STOPP/START application had a
27.7% (95% CI = 22.4%-33.0%) reduction in antipsychotic and
a 39.5% (95% CI = 35.5%-43.5%) reduction in anxiolytic
MEDDs. Intervention-associated antidepressant MEDD reduc-
tions were statistically significant but less dramatic. Interventions
were associated with higher rates of medication discontinuation,
but not higher rates of deaths, patient falls, or physical restraints.
CONCLUSION: We found strong evidence that the interven-
tions we studied were associated with reduced psychotropic
use without commensurate harms, suggesting that such inter-
ventions should be incorporated into Spanish nursing home
care models. Public reporting of psychotropic medication use
in Spanish care homes may encourage care homes to regularly
monitor psychotropic medication use and implement such
instruments. J Am Geriatr Soc 00:1–10, 2019.
Key words: dementia; Spain; clinical care; psychotropic
medications; nursing homes

From the *Dartmouth Institute, The Geisel School of Medicine at

Dartmouth, Lebanon, New Hampshire; †Sanitas Mayores, Barcelona,
Spain; and the ‡Department of Social and Behavioral Sciences, Harvard
School of Public Health, Boston, Massachusetts.
Address correspondence to William B. Weeks, The Dartmouth Institute for
Health Policy and Clinical Practice, 307 River Rd, Lyme, NH 03768. E-
mail: wbw@dartmouth.edu.
DOI: 10.1111/jgs.15855
O veruse of psychotropic medications in older adults
has long been recognized,1 particularly in long-term
care settings.2 Such overuse in nursing home settings is con-
sidered a human rights issue in the United States,3 where
substantial variation in psychotropic medication prescribing
patterns exists.4 Those findings encouraged the Centers for
Medicare and Medicaid Services to enter a partnership with

JAGS 00:1–10, 2019

© 2019 The American Geriatrics Society 0002-8614/18/$15.00
2 WEEKS ET AL.
federal and state agencies that sought to reduce psychotro-
pic medication use among nursing home residents with
dementia,5 a successful endeavor.6
In Western Europe, despite numerous guidelines recom-
mending alternatives as first-line treatments for behavioral
symptoms of dementia,7 psychotropic drug prescribing for
patients with dementia living in nursing homes is common;
Spain has the highest rate of antipsychotic prescribing.8
Concerned about the high rate of physical9 and chemical10
restraint use in long-term settings in Spain, the Spanish
Geriatrics and Gerontology Society encourages reducing
both forms of restraint.11
Sanitas Mayores (Sanitas) is a large older people care
home network in Spain with over 5800 beds distributed over
46 care homes. A subsidiary of the global health and care
company, Bupa, Sanitas recently found that a multiyear effort
to reduce use of physical restraints decreased restraint use
from 18.1% to 1.6%.12 In a continued effort to advance its
residents’ care quality, Sanitas sought to evaluate the impact
of interventions designed to improve psychotropic prescribing
practice with an eye toward incorporating such interventions
into its care home treatment model.
METHODS
Aim, Design, and Setting
We sought to determine the association between short-term
changes in psychotropic medication prescribing and use of
interventions designed to optimize psychotropic prescribing
among older nursing home residents with dementia. There-
fore, we used a quasi-experimental design to conduct a retro-
spective, controlled, observational analysis of the association
between changes in psychotropic prescribing and the intro-
duction of three such interventions into the Sanitas system
between January 15 and June 3, 2018.
Tool Selection
To inform the quality improvement effort, we conducted a
literature search and worked with Sanitas leadership and
nursing home staff to identify three tools that warranted
evaluation.
Multidisciplinary Medication Reviews or
Interdisciplinary Ward Rounds (“Team Rounds”)
Such programs have been shown to decrease inappropriate
medication use among nursing home residents,13–24 particu-
larly for patients whose medication loads have increased
over time25 and those transitioning to nursing homes.26 We
developed a guideline for conducting Team Rounds, the
English version of which is available in Supplementary
Appendix S1.
Application of “STOPP/START” Criteria
Over the last decade, a consensus process established con-
tent validity of a tool designed to help physicians evaluate
medication appropriateness in older patients, known as the
Screening Tool of Older Persons’ Prescriptions (STOPP)/
Screening Tool to Alert Doctors to Right Treatment
(START) criteria.27 STOPP/START implementation reduces
MONTH 2019–VOL. 00, NO. 00 JAGS
inappropriate medication use, antipsychotic prescribing,
overall number of medications prescribed, delirium, visits to
the emergency department, and days in a hospital setting.28
STOPP/START criteria, as modified for Spanish use,29 were
made available in Sanitas’ electronic medical record (EMR).
Use of a Patient “Decision Aid” to Guide Treatment
Decisions Regarding Psychotropic Medications
Decision Aids have been recommended as a person-centered
approach to engage people with dementia (and their families)
about several issues pertinent to dementia, including use of
psychotropic medications.30–32 We modified the OptionGrid33
format, wherein risks and benefits of treatment options objec-
tively are presented to patients as comparison tables, to
develop a Decision Aid for psychotropic medications, the
English version of which is in Supplementary Appendix S2.
We developed recommendations for conducting STOPP/
START reviews or using the Decision Aid (Supplementary
Appendix S3); Team Rounds guidelines included such recom-
mendations (Supplementary Appendix S1).
Prior to the beginning of the quality improvement effort,
45 of Sanitas’ care homes were randomly assigned to one of
five groups. To ensure that adequate controls would be avail-
able for retrospective matching, one group was not assigned
any interventions; the other four groups were randomly
assigned to implement a particular intervention during the
study. Staff and administrators in care homes assigned to
implement interventions received video-conference training
on how to apply them; they were encouraged to use tools on
older patients who were on psychotropic agents and had
dementia. Staff selected patients for intervention without
interference and were encouraged to include family members
in the intervention process. If a patient received an interven-
tion, the date and type of intervention were recorded in that
patient’s EMR.
Data Sources, Sample Definition, and Variables
We collected all data from Sanitas’ EMR.
For care home residents, aged 70 to 99 years, who
lived in a Sanitas care home throughout the study’s dura-
tion, we retrospectively identified the care home in which
the patient lived, the date of admission, a list of Interna-
tional Classification of Diseases, Ninth Revision (ICD-9),
and International Classification of Diseases, Tenth Revision
(ICD-10), diagnoses, and patient age and sex. For each
study week, we determined whether the patient had fallen,
been physically restrained, or died.
The EMR contained one of two patient-specific mea-
sures of cognitive impairment that were generally updated
once a year: a Short Portable Mental Status Questionnaire
score (also known as a “Pfeiffer” score [range = 0-10])34 or
a Functional Assessment Staging score (“FAST” score
[range = 1-7]).35,36 For both, a score of 5 or higher indi-
cates moderate cognitive impairment consistent with
dementia. We used ICD-9 and ICD-10 codes to identify
patients who carried a dementia diagnosis (our inclusion
criteria are available in Supplementary Appendix S4). If a
patient had an ICD-9 or ICD-10 coded dementia diagnosis
or a Pfeiffer or FAST score equal to or greater than 5, we
classified that patient as having dementia. We also used
JAGS MONTH 2019–VOL. 00, NO. 00
ICD-9 and ICD-10 diagnostic codes to calculate a modified
Charlson score37,38 using an R-based algorithm.39,40
Data from the EMR also allowed us to determine
whether, during a particular week, a patient was prescribed
any of three psychotropic drug classes commonly used for
chemical restraint: antipsychotic drugs (antipsychotics), ben-
zodiazepines and “Z-medications” (anxiolytics), and sedative
antidepressants (antidepressants). While our main interests
were antipsychotic and anxiolytic use, we also evaluated the
antidepressant use because one study found increased sedative
antidepressant use concurrent with efforts to reduce antipsy-
chotics in the United States between 2009 and 2014.6
For patients who were prescribed any medications in any of
these three drug classes, we obtained the average daily dose of
each prescribed medication and calculated milligram-equivalent
daily doses (MEDDs) of chlorpromazine for antipsychotics,
diazepam for anxiolytics, or fluoxetine for antidepressants and
averaged MEDDs over each study week. The conversion ratios
that we used to calculate MEDDs and their sources are provided
in Supplementary Appendix S5. For analytic purposes, we used
the Tukey method to identify high MEDD outliers at the value
of the 75th percentile plus 1.5 times the interquartile range,41
and assigned that value to any MEDD above that value.
If a resident received any of the interventions (“inter-
vention patients”), the type of intervention and its applica-
tion date were recorded in the EMR. We limited our
analysis to residents who received only one intervention.
Using propensity scores can balance treatment groups on
observed confounders so that the outcomes of different treat-
ment groups can be directly compared. For a set of observed
covariates, a patient’s propensity score is the probability of a
patient with the same observed characteristics being in the
treatment group.42,43 For all patients who met inclusion cri-
teria, regardless of whether they received an intervention, we
used a multinomial logistic regression model44 that used the
following variables to generate propensity scores: age, sex,
length of time the patient had resided in the care home,
Charlson score, and MEDD of each drug class of interest.
Because we hypothesized that different interventions might
have different associations with each drug class, we analyzed nine
intervention/drug class combinations of intervention patients and
controls as separate cohorts. For each intervention patient, in
each of the nine intervention/drug class cohorts, we identified
two controls who had the closest propensity score to each inter-
vention patient; for each cohort, controls were used only once.
Matched controls were assigned an intervention date that was
defined by the median number of days of the controls’ collective
follow-up period so that the median follow-up period for inter-
vention patients and controls would be similar.
Our final sample then consisted of 1653 Sanitas nurs-
ing home residents, aged 70 to 99 years, who we classified
as having dementia, who were prescribed at least one of the
three drug classes, and for whom weekly data were avail-
able for the entire study period; 606 of those residents
received one of the interventions, and 1047 served as pro-
pensity score–matched controls (Figure 1, top).
Analytic Methods
For each study week, we identified the number and the
cumulative proportion of residents who met our inclusion
criteria who had each type of intervention.
REDUCING PSYCHOTROPICS IN SPANISH CARE HOMES 3
In unadjusted analysis, we used a Student’s t-test to
determine whether intervention patients and controls within
a cohort had demographic, Charlson score, length of stay,
or daily drug class daily equivalent dosing differences during
the baseline 2-week period prior to the intervention (T0;
identified by the date of intervention for intervention patients
and the assigned intervention date for controls). Further, we
used a Student’s t-test to determine the absolute change from
T0 in daily equivalent dosing of particular drug classes in
the first (T1) and second (T2) 2-week period following a
particular intervention (Figure 1, bottom). Finally, we identi-
fied the proportion of patients for whom the particular drug
class had been eliminated by T2 and used the χ2 test to com-
pare intervention patients to controls on that measure.
To adjust for any clustering effects at the care home
level or any individual demographic differences, for each
intervention/drug class cohort, we used a mixed-effects lin-
ear regression model (fitted using R’s “lme4” package) that
assigned a random effect for each care home and a random
effect for each individual and included each subject’s age,
sex, length of time in the care home, Charlson score, and
number of days since actual or assigned intervention date in
the model. As dependent variables, we used either the drug-
class–specific absolute MEDD or the MEDD as a percent-
age of the MEDD during T0. Using the models, we calcu-
lated a coefficient (and 95% confidence intervals [CIs]) that
represented the adjusted mean absolute or relative change
in drug-class–specific EDDs from T0 to T2 that were associ-
ated with a particular intervention.
To examine possible adverse consequences of medica-
tion changes, we used conditional logistic regression to cal-
culate odds ratios for patient falls and use of restraints in
the 2 weeks after, compared to the 2 weeks before, the
intervention period, controlling for age, sex, length of time
in the care home, and Charlson score. To determine
whether residents who received any intervention died fol-
lowing an intervention, we examined death incidence for
intervention patients and controls for all individuals, aged
70 to 99 years, who had dementia.
Human Subjects Approval
When they enter a Sanitas care home, residents sign a con-
tract that includes a clause that translates: “For the carrying
out of scientific, medical and/or historical analysis studies,
the user’s data will be anonymized, and otherwise the prior
written consent will be obtained directly from the resident
or of the person who assumes their representation.”
Data were anonymized as defined by Spanish law.
Because Dartmouth obtained anonymized data for analytic
purposes, Dartmouth College concluded that the study was
exempt from human studies review.
RESULTS
Uptake of each intervention was relatively consistent during
the study period; however, STOPP/START and Team
Rounds were about twice as commonly used as Decision
Aids (Figure 2).
At baseline, intervention patients and controls were
similar within each of the nine intervention/drug class
cohorts (Table 1). The mean preintervention antipsychotic
4 WEEKS ET AL.
Unique care home residents,
aged 70-99 y, who resided in
1 of 45 Sanitas care homes
throughout the period
December 18, 2017-July 2, 2018
(N=4029)
With dementia
(N=2861)
Received an intervention
(N=909)
INTERVENTION PATIENTS
Taking a medication in one of
the three drug classes
(N=606)
Actual or assigned
intervention date
T0 T1
2 wk 2 wk
Figure 1. Application of inclusion and exclusion criteria and analysis timeline. Sanitas, Sanitas Mayores; T0, baseline 2-week
period before the intervention; T1, first 2-week period following a particular intervention; T2, second 2-week period following a
particular intervention.
MEDD was higher for intervention patients and controls in
the Team Rounds/antipsychotic cohort than in the other
two antipsychotic cohorts (P < .001); mean preintervention
antidepressant MEDDs were higher in the Decision Aid/
antidepressant cohort than in the other two antidepressant
cohorts (P < .001).
All three interventions were associated with statistically
significant lower antipsychotic, anxiolytic, and antidepressant
MEDDs in T2 when compared to T0; exposure to Decision
Aids and STOPP/START criteria was associated with statisti-
cally significant decreases in MEDDs of all three drug classes
(Table 2). When compared to T0, controls in the STOPP/
START-antidepressant cohort experienced a statistically sig-
nificant, but modest, reduction in antidepressant MEDDs in
T1 and T2; otherwise, no controls demonstrated statistically
significant changes in daily equivalent drug class dosing.
MONTH 2019–VOL. 00, NO. 00 JAGS
Without dementia
(N=1168)
No intervention
(N=1952)
CONTROLS
Propensity score matched
and taking a medication in
one of the three drug classes
(N=1047)
T2
2 wk
Residents who received any of the interventions were statisti-
cally significantly more likely to have the drug class of interest
discontinued within 4 weeks after the intervention.
Results of the mixed-effects model indicated that,
across drug classes, exposure to Team Rounds was associ-
ated with a lesser reduction in MEDDs of all drug classes
than the other two interventions (Figure 3). Exposure to
Team Rounds was associated with an 11.9% (95% CI =
9.4%-14.5%) reduction in antidepressant MEDDs (or 0.8
[95% CI = 0.6-1.0] mg/d fewer), a 23.3% (95% CI =
13.9%-32.8%) reduction in antipsychotic MEDDs (or 52.0
[95% CI = 43.1-60.9] mg/d fewer), and a 23.1% (95% CI
= 18.3%-28.0%) reduction in anxiolytic MEDDs (or 3.8
[95% CI = 3.3-4.4] mg/d fewer). Exposure to Decision Aids
was associated with a 13.1% (95% CI = 8.8%-17.4%)
reduction in antidepressant MEDDs (or 2.4 [95% CI =
JAGS MONTH 2019–VOL. 00, NO. 00
Cumulative Uptake of Interventions as a Proportion of
Addition of Second
Intervention
the Eligible Population, %
Study Week Beginning
Run-In Period
Figure 2. Uptake of interventions: cumulative number of patients obtaining an intervention as a proportion of the eligible population,
by study week. START, Screening Tool to Alert Doctors to Right Treatment; STOPP, Screening Tool of Older Persons’ Prescriptions.
1.9-2.9] mg/d fewer), a 24.8% (95% CI = 15.6%-33.9%)
reduction in antipsychotic MEDDs (or 53.0 [95% CI =
45.5-60.6] mg/d fewer), and a 31.8% (95% CI = 25.5%-
38.2%) reduction in anxiolytic MEDDs (or 4.2 [95% CI =
3.7-4.7] mg/d fewer). Exposure to STOPP/START applica-
tion was associated with a 29.2% (95% CI = 26.7%-
31.8%) reduction in antidepressant MEDDs (or 2.5 [95%
CI = 2.3-2.7] mg/d fewer), a 27.7% (95% CI = 22.4%-
33.0%) reduction in antipsychotic MEDDs (or 43.9 [95%
CI = 38.3-49.6] mg/d fewer), and a 39.5% (95% CI =
35.5%-43.5%) reduction in anxiolytic MEDDs (or 5.4
[95% CI = 5.1-5.8] mg/d fewer).
Intervention patients were not more likely than controls
to have experienced a documented fall or to have been
physically restrained in the 2-week period following the
intervention when compared to the 2-week period preced-
ing the intervention (data not shown). None of the 372 resi-
dents who met our inclusion criteria and died during the
study period received any of the interventions we examined.
DISCUSSION
Among older Spanish nursing home residents with dementia,
we studied the association between exposure to one of three
interventions and MEDDs of three psychotropic drug classes
that are commonly overused. We found that for all three drug
classes, exposure to all three interventions was associated
with substantially reduced mean MEDDs—both before and
after adjustment—and with higher rates of medication dis-
continuation. We found no evidence of substitution of antide-
pressants for anxiolytics or antipsychotics and no evidence of
adverse consequences of medication reduction. Decision Aids
were less often used than the other two interventions.
Our findings suggest that using STOPP/START (ie,
based on clinical evidence and integrated into Sanitas’
EMR) was more consistently effective than the other tools.
This makes sense: providers more consistently provide
recommended care when tools that provide structure for
clinical decision making, like guidelines, are used.45
Our study has several limitations. First, we studied older,
Spanish nursing home residents who had dementia and who
lived within a single care home organization. Studying other
REDUCING PSYCHOTROPICS IN SPANISH CARE HOMES 5
Follow-Up Period
populations may generate different results. Second, our find-
ings relied on self-documentation of completion of the inter-
ventions in that EMR. While there may have been some
administrative errors in the data recording process, throughout
the project we confirmed consistent data collection. Third, we
studied only three drug classes; changes in prescribing patterns
in other drug classes—such as opioids or mood stabilizers—
might have occurred. Fourth, we did not randomly assign
patients to receive a particular intervention. Although interven-
tions were randomly assigned to care homes and we identified
propensity score–matched controls, our analysis was retrospec-
tive and observational; it is possible that providers selected
patients who were likely to have the best results following
intervention and that application of the intervention to remain-
ing eligible patients would not have the same effect. However,
our intent was to conduct a study that would inform actual
practice; it is reasonable to assume that the interventions that
we used are not appropriate for some patients and that practi-
tioners are likely to use interventions where they believe they
will be most effective. Fifth, while we examined some potential
adverse outcomes of medication reduction—falls, restraint use,
and death—our study did not evaluate other potential adverse
outcomes, such as increased disruptive behavior. Finally, we
studied only three interventions, and their impact only for a
short time period; other instruments studied for longer time
periods might generate different results.
Despite these limitations, our findings are promising in
that the use of simple, guided instruments was associated
with reduced use of psychotropic agents among older care
home residents with dementia. Given that the effects of Deci-
sion Aids and STOPP/START criteria application were simi-
lar, an organizational strategy might be to provide both tools
to providers and let them choose which they prefer to use.
Regardless, a new model of care for nursing home residents
should include efforts to use the minimal effective dose for
the shortest time period possible, regularly review medica-
tions, and involve families in care decisions. Each intervention
we studied is consistent with the recently published National
Institute for Health and Care Excellence (NICE) guidelines on
the assessment, management, and support of people living
with dementia.3 Sanitas might use its EMR to regularly
prompt providers to conduct such reviews.
 6
WEEKS ET AL.

Table 1. Characteristics of Controls and Intervention Patients by Intervention and Medicationa

Intervention

Baseline Patient Team Rounds Decision Aid STOPP/START

Characteristics and
Patients Daily Medication Intervention
Prescribed an … Dosing Controls (SD) Intervention Patients (SD) P Value Controls (SD) Intervention Patients (SD) P Value Controls (SD) Patients (SD) P Value

Antipsychotic agent No. 168 84 114 57 228 114

Age, y 86.7 (6.5) 86.7 (6.2) .97 87.2 (5.9) 86.2 (6.9) .35 86.4 (6.2) 85.7 (5.6) .34
Male sex, % 23.2 21.4 .87 21.1 22.8 .95 23.7 26.3 .69
Time admitted, y 2.9 (2.9) 2.95 (2.83) .82 2.2 (1.9) 2.2 (1.8) .80 1.8 (1.7) 1.7 (1.7) .58
Charlson score 0.97 (0.86) 1.04 (0.94) .53 0.49 (0.62) 0.46 (0.65) .80 0.42 (0.66) 0.40 (0.65) .82
CMEDD, mg/d 270 (283) 271 (255) .96 190 (234) 215 (198) .49 191 (217) 196 (207) .83
Anxiolytic No. 112 56 104 52 258 129
Age, y 87.6 (6.1) 87.7 (5.8) .88 87.3 (6.3) 87.7 (6.5) .77 86.8 (6.4) 86.8 (5.7) .92
Male sex, % 12.5 14.3 .94 18.3 15.4 .82 14.7 14.0 .96
Time admitted, y 3.3 (2.8) 3.2 (3.0) .85 2.6 (2.5) 2.9 (2.4) .45 2.7 (2.6) 2.7 (2.5) .92
Charlson score 0.95 (0.93) 1.04 (1.01) .57 0.36 (0.67) 0.35 (0.68) .90 0.44 (0.61) 0.40 (0.64) .58
DMEDD, mg/d 16.0 (9.2) 16.6 (9.6) .69 15.7 (9.3) 16.2 (8.6) .74 15.3 (9.3) 15.4 (9.6) .97
Antidepressant No. 194 97 152 76 448 224
Age, y 87.3 (5.8) 87.3 (6.0) .94 87.6 (5.7) 88.1 (5.8) .49 87.0 (6.1) 87.2 (5.5) .75
Male sex, % 11.3 14.4 .57 15.8 15.8 1.00 22.3 21.0 .77
Time admitted, y 3.5 (3.1) 3.5 (3.4) .96 2.2 (1.9) 2.5 (2.3) .46 2.3 (2.1) 2.3 (2.1) .87
Charlson score 0.89 (0.98) 0.97 (0.86) .46 0.64 (0.82) 0.53 (0.82) .32 0.52 (0.75) 0.50 (0.74) .64
SADMEDD, mg/d 11.9 (9.9) 11.6 (9.5) .84 16.1 (12.4) 16.6 (12.8) .78 12.5 (9.7) 12.7 (9.6) .84

Abbreviations: CMEDD, chlorpromazine milligram-equivalent daily dose; DMEDD, diazepam milligram-equivalent daily dose; SADMEDD, sedative antidepressant milligram-equivalent daily dose; START, Screen-
ing Tool to Alert Doctors to Right Treatment; STOPP, Screening Tool of Older Persons’ Prescriptions.
a
Controls were propensity score matched at a rate of 2:1, except for the STOPP/START intervention for sedative antidepressants, where there were not enough controls to achieve a 2:1 match. We found no statisti-
cally significant differences between controls and intervention patients within a particular intervention/medication combination.
MONTH 2019–VOL. 00, NO. 00
JAGS
 JAGS

Table 2. Unadjusted Comparison of Mean MEDDs of Three Medication Classes for Intervention Patients and Propensity Score–Matched Controls at T0, T1,
and T2a

Patients for Whom

Medication Was
Discontinued by
MONTH 2019–VOL. 00, NO. 00

T0 T1 T2 the End of T2

P Value: P Value: P Value:

Patients Mean Mean Compared Compared T1 Compared
Intervention Prescribed an … Group No. MEDD (SD) MEDD (SD) to T0 Mean MEDD (SD) to T0 to T2 % P Value

Team Rounds Antipsychotica Intervention 84 271 (255) 260 (255) .31 241 (252) .022 .003 23.8 .002
Control 168 270 (283) 267 (285) .60 263 (285) .33 .11 8.9
Anxiolytic Intervention 56 16.6 (9.6) 15.4 (11.0) .10 13.0 (11.0) <.001 <.001 32.1 <.001
Control 112 16.0 (9.2) 15.9 (9.5) .77 15.8 (9.4) .45 .49 8.9
Antidepressant Intervention 97 11.6 (9.5) 11.3 (9.8) .13 11.1 (9.8) .004 .06 21.6 <.001
Control 194 11.8 (9.9) 11.9 (9.9) .63 11.9 (10.0) .66 .90 4.1
Decision Aid Antipsychotic Intervention 57 215 (198) 180 (198) <.001 164 (207) <.001 .005 26.3 <.001
Control 114 190 (234) 186 (233) .45 184 (233) .30 .44 6.1
Anxiolytic Intervention 52 16.2 (8.6) 13.5 (9.3) <.001 11.8 (10.3) <.001 .002 26.9 .005
Control 104 15.7 (9.3) 15.6 (9.6) .83 15.5 (9.4) .54 .29 8.7
Antidepressant Intervention 76 16.6 (12.8) 15.2 (12.5) .033 12.9 (13.5) .002 .003 21.1 <.001
Control 152 16.1 (12.4) 15.9 (12.2) .092 15.5 (12.1) .057 .15 3.9
STOPP/START Antipsychotic Intervention 114 196 (207) 169 (198) .001 153 (201) <.001 .001 27.2 <.001
Control 228 191 (217) 195 (225) .098 190 (220) .82 .10 6.1
Anxiolytic Intervention 129 15.4 (9.6) 12.7 (10.1) <.001 10.7 (10.7) <.001 <.001 45.0 <.001
Control 258 15.3 (9.3) 15.2 (9.4) .51 15.0 (9.4) .18 .22 7.8
Antidepressant Intervention 224 12.7 (9.6) 11.2 (9.9) <.001 10.1 (10.2) <.001 <.001 26.3 <.001
Control 448 12.5 (9.7) 12.4 (9.6) .004 12.2 (9.5) .004 .07 5.8

Abbreviations: MEDD, milligram-equivalent daily dose; START, Screening Tool to Alert Doctors to Right Treatment; STOPP, Screening Tool of Older Persons’ Prescriptions; T0, baseline 2-week period before the
intervention; T1, first 2-week period following a particular intervention; T2, second 2-week period following a particular intervention.
a
P values comparing T1 to T2 are provided, as are the proportion of patients for whom the medication of interest was discontinued by the end of T2 and a P value comparing that value for intervention patients and
controls. Differences between controls and intervention patients that are statistically significant at P < .05 are bolded.
REDUCING PSYCHOTROPICS IN SPANISH CARE HOMES
7
8 WEEKS ET AL.
Antidepressant
Team Rounds
Antipsychotic
Anxiolytic
Antidepressant
Decision Aid
Antipsychotic
Anxiolytic
Antidepressant
STOPP/START
Antipsychotic
Anxiolytic
−45 −40
Mean Percentage Change in Milligram-Equivalent Daily Dose During the
Figure 3. Average adjusted impact of each intervention on relative change in mean milligram-equivalent daily doses of antidepres-
sants, anxiolytics, and antipsychotics in the 4 weeks following the intervention, with 95% confidence intervals. START, Screening
Tool to Alert Doctors to Right Treatment; STOPP, Screening Tool of Older Persons’ Prescriptions.
Our findings have policy implications. First, the world
is aging rapidly. Europe46 and Spain47 are anticipating
rapid growth in the proportion of older residents, demo-
graphic shifts expected to cause increased health demands
and costs.48 Optimal use of psychotropic medications in the
care home setting might not only improve older patient care
quality but also generate cost savings in care for patients
with dementia by reducing medication costs.
But more importantly, as other studies have found,8,10
our study confirms that there are opportunities to reduce
psychotropic medication use among older nursing home
residents in Spain who have dementia. An international
human rights group has identified a moral imperative for
national governments to address overuse of psychotropic
medications in older nursing home residents with demen-
tia.3 The United States has used public reporting of antipsy-
chotic utilization to facilitate reductions in psychotropic
prescribing rates,49,50 an effort that increased medication
reviews, reduced medication use, and increased use of non-
pharmacological interventions.51 Spanish regulators might
consider requiring public reporting of psychotropic medica-
tion use in care homes, an effort that might encourage
uptake of the types of tools we studied, reduce overuse of
such medications in a vulnerable population, and improve
the quality and dignity of care provided to older Spanish
citizens who are living with dementia.
ACKNOWLEDGMENTS
Financial Disclosure: This work was funded by Sanitas.
Weeks, Mishra, Elwyn, Godfrey, Petersen, and Hswen all
received funding to conduct the work described in the article.
Conflict of Interest: Curto, Cano, Tomás, Villamarín,
and Sanchez are all employees of Sanitas.
MONTH 2019–VOL. 00, NO. 00 JAGS
−35 −30 −25 −20 −15 −10 −5 0
4 wk Following Intervention, Relative to Controls
Elwyn has edited and published books that provide royal-
ties on sales by the publishers: the books include Shared Deci-
sion Making (Oxford University Press) and Groups (Radcliffe
Press). He has in the past provided consultancy for organiza-
tions, including: (1) Emmi Solutions LLC, which developed
patient decision support tools; (2) National Quality Forum on
the certification of decision support tools; (3) Washington State
Health Department on the certification of decision support
tools; and (4) SciMentum LLC, Amsterdam (workshops for
shared decision making). Elwyn is director of &think LLC,
which owns the registered trademark for Option Grids Patient
Decision Aids. He provides consultancy in the domain of
shared decision making and Patient Decision Aids to: (1) Access
Community Health Network, Chicago (Federally Qualified
Medical Centers); and (2) EBSCO Health Option Grids Patient
Decision Aids. Elwyn owns copyright in measures of shared
decision making and care integration (namely, collaboRATE,
integRATE, Observer OPTION-5, and Observer OPTION-
12); these measures are freely available for use.
Author Contributions: All authors who contributed sig-
nificantly to this work are listed and contributed adequately
to the conceptualization, planning, design, data collection,
analysis, data interpretation, writing, and revision of the
manuscript to warrant authorship.
Sponsor’s Role: Members of the Sanitas team contrib-
uted to the design, methods, data collection, analysis, and
preparation of the manuscript. Subjects were recruited by
Sanitas-employed providers at the care homes.
REFERENCES
1. Ackermann RJ, Meyer von Bremen GB. Reducing polypharmacy in the nurs-
ing home: an activist approach. J Am Board Fam Pract. 1995;8:195-205.
2. Brandt NJ, Pythtila J. Psychopharmacological medication use among older
adults with dementia in nursing homes. J Gerontol Nurs. 2013;39:8-14.
JAGS MONTH 2019–VOL. 00, NO. 00
3. "They Want Docile." How Nursing Homes in the United States Overmedicate
People With Dementia. United States of America: Human Rights Watch; 2018.
https://www.hrw.org/report/2018/02/05/they-want-docile/how-nursing-homes-
united-states-overmedicate-people-dementia#. Accessed February 27, 2018.
4. Cioltan H, Alshehri S, Howe C, et al. Variation in use of antipsychotic medi-
cations in nursing homes in the United States: a systematic review. BMC Ger-
iatr. 2017;17:32.
5. National Partnership to Improve Dementia Care in Nursing Homes. 2018.
https://http://www.cms.gov/Medicare/Provider-Enrollment-and-Certification/
SurveyCertificationGenInfo/National-Partnership-to-Improve-Dementia-Care-
in-Nursing-Homes.html. Accessed November 14, 2018.
6. Maust DT, Kim HM, Chiang C, Kales HC. Association of the Centers for
Medicare & Medicaid Services’ National Partnership to improve dementia care
with the use of antipsychotics and other psychotropics in long-term care in the
United States from 2009 to 2014. JAMA Intern Med. 2018;178(5):640-647.
7. Azermai M, Petrovic M, Elseviers MM, Bourgeois J, van Bortel LM, Vander
Stichele RH. Systematic appraisal of dementia guidelines for the management
of behavioural and psychological symptoms. Ageing Res Rev. 2012;11:78-86.
8. Janus SIM, van Manen JG, Ijzerman MJ, Zuidema SU. Psychotropic drug
prescriptions in Western European nursing homes. Int Psychogeriatr. 2016;
28:1775-1790.
9. Estevez-Guerra GJ, Farina-Lopez E, Nunez-Gonzalez E, et al. The use of
physical restraints in long-term care in Spain: a multi-center cross-sectional
study. BMC Geriatr. 2017;17:29.
10. Olazaran J, Valle D, Serra JA, Cano P, Muniz R. Psychotropic medications
and falls in nursing homes: a cross-sectional study. J Am Med Dir Assoc.
2013;14:213-217.
11. Ramos Cordero P, Lopez Trigo JA, Maillo Pedraz H, Paz Rubio JM. Physi-
cal and pharmacological restraints in geriatric and gerontology services and
centers [in Spanish]. Rev Esp Geriatr Gerontol. 2015;50:35-38.
12. Muniz R, Gomez S, Curto D, et al. Reducing physical restraints in nursing
homes: a report from Maria Wolff and Sanitas. J Am Med Dir Assoc. 2016;
17:633-639.
13. Crotty M, Halbert J, Rowett D, et al. An outreach geriatric medication advi-
sory service in residential aged care: a randomised controlled trial of case
conferencing. Age Ageing. 2004;33:612-617.
14. Davidsson M, Vibe OE, Ruths S, Blix HS. A multidisciplinary approach to
improve drug therapy in nursing homes. J Multidiscip Healthc. 2011;4:9-13.
15. Gallagher PF, O’Connor MN, O’Mahony D. Prevention of potentially inap-
propriate prescribing for elderly patients: a randomized controlled trial using
STOPP/START criteria. Clin Pharmacol Ther. 2011;89:845-854.
16. Maher RL, Hanlon J, Hajjar ER. Clinical consequences of polypharmacy in
elderly. Expert Opin Drug Saf. 2014;13:57-65.
17. Milos V, Rekman E, Bondesson A, et al. Improving the quality of pharmaco-
therapy in elderly primary care patients through medication reviews: a ran-
domised controlled study. Drugs Aging. 2013;30:235-246.
18. Monette J, Monette M, Sourial N, et al. Effect of an interdisciplinary educa-
tional program on antipsychotic prescribing among residents with dementia
in two long-term care centers. J Appl Gerontol. 2013;32:833-854.
19. Patterson SM, Hughes C, Kerse N, Cardwell CR, Bradley MC. Interventions
to improve the appropriate use of polypharmacy for older people. Cochrane
Database Syst Rev. 2012;(5):CD008165.
20. Pitkala KH, Juola AL, Kautiainen H, et al. Education to reduce potentially
harmful medication use among residents of assisted living facilities: a ran-
domized controlled trial. J Am Med Dir Assoc. 2014;15:892-898.
21. Prentice A, Wright D. Reducing antipsychotic drugs in care homes. Nurs
Times. 2014;110:12-15.
22. Richter T, Meyer G, Mohler R, Kopke S. Psychosocial interventions for
reducing antipsychotic medication in care home residents. Cochrane Data-
base Syst Rev. 2012;12:CD008634.
23. Stuijt CC, Franssen EJ, Egberts AC, Hudson SA. Appropriateness of pre-
scribing among elderly patients in a Dutch residential home: observational
study of outcomes after a pharmacist-led medication review. Drugs Aging.
2008;25:947-954.
24. Thompson Coon J, Abbott R, Rogers M, et al. Interventions to reduce inappro-
priate prescribing of antipsychotic medications in people with dementia resident
in care homes: a systematic review. J Am Med Dir Assoc. 2014;15:706-718.
25. Di Giorgio C, Provenzani A, Polidori P. Potentially inappropriate drug pre-
scribing in elderly hospitalized patients: an analysis and comparison of
explicit criteria. Int J Clin Pharmacol. 2016;38:462-468.
26. Crotty M, Rowett D, Spurling L, Giles LC, Phillips PA. Does the addition of
a pharmacist transition coordinator improve evidence-based medication
management and health outcomes in older adults moving from the hospital
to a long-term care facility? results of a randomized, controlled trial.
Am J Geriatr Pharmacother. 2004;2:257-264.
27. Gallagher P, Ryan C, Byrne S, Kennedy J, O’Mahony D. STOPP (Screening
Tool of Older Person’s Prescriptions) and START (Screening Tool to Alert
REDUCING PSYCHOTROPICS IN SPANISH CARE HOMES 9
doctors to Right Treatment): consensus validation. Int J Clin Pharmacol
Ther. 2008;46:72-83.
28. Garcia-Gollarte F, Baleriola-Julvez J, Ferrero-Lopez I, Cuenllas-Diaz A,
Cruz-Jentoft AJ. An educational intervention on drug use in nursing homes
improves health outcomes resource utilization and reduces inappropriate
drug prescription. J Am Med Dir Assoc. 2014;15:885-891.
29. Castillo-Paramo A, Pardo-Lopo R, Gomez-Serranillos IR, et al. Assessment
of the appropriateness of STOPP/START criteria in primary health care in
Spain by the RAND method [in Spanish]. Sem Ther. 2013;39:413-420.
30. Carmody J, Traynor V, Steele A. Dementia, decision aids and general prac-
tice. Aust Fam Physician. 2015;44:307-310.
31. Mikesell L, Bromley E, Young AS, Vona P, Zima B. Integrating client and
clinician perspectives on psychotropic medication decisions: developing a
communication-centered epistemic model of shared decision making for
mental health contexts. Health Commun. 2016;31:707-717.
32. Naarding P, van Grevenstein M, Beekman AT. Benefit-risk analysis for the clini-
cian: “primum non nocere” revisited--the case for antipsychotics in the treatment of
behavioural disturbances in dementia. Int J Geriatr Psychiatry. 2010;25:437-440.
33. Elwyn G, Lloyd A, Joseph-Williams N, et al. Option Grids: shared decision
making made easier. Patient Educ Couns. 2013;90:207-212.
34. Pfeiffer E. A short portable mental status questionnaire for the assessment of
organic brain deficit in elderly patients. J Am Geriatr Soc. 1975;23:433-441.
35. Reisberg B. Functional assessment staging (FAST). Psychopharmacol Bull.
1988;24:653-659.
36. Sclan SG, Reisberg B. Functional assessment staging (FAST) in Alzheimer’s disease:
reliability, validity, and ordinality. Int Psychogeriatr. 1992;4((suppl 1)):55-69.
37. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classify-
ing prognostic comorbidity in longitudinal studies: development and valida-
tion. J Chronic Dis. 1987;40:373-383.
38. Quan H, Li B, Couris CM, et al. Updating and validating the Charlson
comorbidity index and score for risk adjustment in hospital discharge
abstracts using data from 6 countries. Am J Epidemiol. 2011;173:676-682.
39. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining
comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care.
2005;43:1130-1139.
40. Gasparini A. Comorbidity: an R package for computing comorbidity scores.
J Open Source Softw. 2018;3:648.
41. Tukey JW. Exploratory Data Analysis. Reading, MA: Addison-Wesley Pub-
lishing Company; 1977.
42. D’Agostino RB Jr. Propensity score methods for bias reduction in the com-
parison of a treatment to a non-randomized control group. Stat Med. 1998;
17:2265-2281.
43. Weeks WB, Tosteson TD, Whedon JM, et al. Comparing propensity score
methods for creating comparable cohorts of chiropractic users and nonusers
in older, multiply comorbid Medicare patients with chronic low Back pain.
J Manipulative Physiol Ther. 2015;38:620-628.
44. Rassen JA, Glynn RJ, Brookhart MA, Schneeweiss S. Covariate selection in
high-dimensional propensity score analyses of treatment effects in small sam-
ples. Am J Epidemiol. 2011;173:1404-1413.
45. Reus VI, Fochtmann LJ, Eyler AE, et al. The American Psychiatric Associa-
tion practice guideline on the use of antipsychotics to treat agitation or psy-
chosis in patients with dementia. Am J Psychiatry. 2016;173:543-546.
46. World Population Prospects (2017 Revision). New York; 2017. https://esa.
un.org/unpd/wpp/Publications/Files/WPP2017_KeyFindings.pdf. Accessed Feb-
ruary 27, 2018.
47. Comas-Herrera A, Wittenberg R, Costa-Font J, et al. Future long-term care
expenditure in Germany, Spain, Italy and the United Kingdom. Ageing Soc.
2006;26:285-302.
48. World Report on Ageing and Health. Luxembourg: World Health Organiza-
tion; 2015. http://www.who.int/ageing/publications/world-report-2015/en/.
Accessed February 27, 2018.
49. Bowblis JR, Lucas JA, Brunt CS. The effects of antipsychotic quality report-
ing on antipsychotic and psychoactive medication use. Health Serv Res.
2015;50:1069-1087.
50. Hughes CM, Lapane KL. Administrative initiatives for reducing inappropri-
ate prescribing of psychotropic drugs in nursing homes: how successful have
they been? Drugs Aging. 2005;22:339-351.
51. Ellis ML, Molinari V, Dobbs D, Smith K, Hyer K. Assessing approaches and
barriers to reduce antipsychotic drug use in Florida nursing homes [pub-
lished correction appears in Aging Ment Health. 2015;19(6):i; PMID:
25751410]. Aging Ment Health. 2015;19:507-516.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this article.
10 WEEKS ET AL.
Supplementary Appendix S1: Guideline for conducting
Team Rounds.
Supplementary Appendix S2: Decision Aid used.
Supplementary Appendix S3: Documents providing
guidance to implementation for each tool.
MONTH 2019–VOL. 00, NO. 00 JAGS
Supplementary Appendix S4: ICD-9 and ICD-10 inclu-
sion criteria for making a code-based dementia diagnosis.
Supplementary Appendix S5: Conversion ratios used to
calculate milligram-equivalent daily doses of each drug
class.