Antipsychotic Deprescription for Older Adults in Long-term Care:
The HALT Study
Henry Brodaty MD, DSc
*, Liesbeth Aerts PhD
Fleur Harrison BA, GDipSc (Psych)
, Tiffany Jessop PhD
, Monica Cations
Lynn Chenoweth BA, MA Hons, MAdEd, PhD
, Allan Shell MB, BS
Gordana C. Popovic BSc, BEd, PhD
, Megan Heffernan PhD
Sarah Hilmer MBBS, PhD
, Perminder S. Sachdev MD, PhD
Brian Draper MBBS, MD
Dementia Centre for Research Collaboration, School of Psychiatry, University of New South Wales Sydney, New South Wales, Australia
Centre for Healthy Brain Ageing, School of Psychiatry, University of New South Wales Sydney, New South Wales, Australia
Academic Department for Old Age Psychiatry, Prince of Wales Hospital, Randwick, New South Wales, Australia
Stats Central, Mark Wainwright Analytical Centre, University of New South Wales Sydney, New South Wales, Australia
Kolling Institute, Royal North Shore Hospital and University of Sydney, St Leonards, New South Wales, Australia
Neuropsychiatric Institute, Prince of Wales Hospital, Randwick, New South Wales, Australia
behavioral and psychological symptoms of
Objectives: Despite limited efﬁcacy and signiﬁcant safety concerns, antipsychotic medications are
frequently used to treat behavioral and psychological symptoms of dementia (BPSD) in long-term resi-
dential care. This study evaluates the sustained reduction of antipsychotic use for BPSD through a
deprescribing intervention and education of health care professionals.
Design: Repeated-measures, longitudinal, single-arm study.
Setting: Long-term residential care of older adults.
Participants: Nursing staff from 23 nursing homes recruited 139 residents taking regular antipsychotic
medication for 3 months, without primary psychotic illness, such as schizophrenia or bipolar disorder, or
Intervention: An antipsychotic deprescribing protocol was established. Education of general practitioners,
pharmacists, and residential care nurses focused on nonpharmacological prevention and management of
Measurements: The primary outcome was antipsychotic use over 12-month follow-up; secondary out-
comes were BPSD (Neuropsychiatric Inventory, Cohen-Mansﬁeld Agitation Inventory, and social with-
drawal) and adverse outcomes (falls, hospitalizations, and cognitive decline).
Results: The number of older adults on regular antipsychotics over 12 months reduced by 81.7% (95% conﬁdence
interval: 72.4-89.0). Withdrawal was not accompaniedbydrugsubstitutionorasigniﬁcant increase in pro-re-
nata antipsychotic or benzodiazepine administration. There was no change in BPSD or in adverse outcomes.
Conclusion: In a selected sample of older adults living in long-term residential care, sustained reduction
in regular antipsychotic use is feasible without an increase of BPSD.
Ó2018 AMDA eThe Society for Post-Acute and Long-Term Care Medicine.
This study was funded by the Australian Department of Health under the Aged
Care Service Improvement and Healthy Ageing Grant Fund. The project is supported
by the Centre for Research Collaboration, UNSW Sydney.
Over the last 3 years, Henry Brodaty has been on the advisory board of Nutricia.
He has been recipient of grants for research by the National Health and Medical
Research Council and Australian Department of Health and Ageing. Monica Cations
is currently employed to assist with data collection for drug trials by Janssen,
Merck, and Paraxel. The other authors declare no competing interests reports and
no ﬁnancial relationships with commercial interests.
* Address correspondence to Henry Brodaty, MD, DSc, Dementia Centre for
Research Collaboration, AGSM Building, University of New South Wales Sydney,
NSW 2052, Australia.
E-mail address: email@example.com (H. Brodaty).
journal homepage: www.jamda.com
1525-8610/Ó2018 AMDA eThe Society for Post-Acute and Long-Term Care Medicine.
JAMDA 19 (2018) 592e600
Behavioral and psychological symptoms of dementia (BPSD)
are prevalent in people living with dementia especially as the
Although there is some evidence for the
efﬁcacy of antipsychotics in treating aggression and psychosis,
they have signiﬁcant side effects. Increased risks of cognitive
decline, cerebrovascular adverse effects, and death have triggered
warnings from international agencies against their use in people
Antipsychotics have been linked to socially
withdrawn behavior and increased risk of falls and
Despite some reports indicating a decline in antipsychotic pre-
their use remains common in people with dementia in
long-term care (LTC): 20% to 50% are prescribed antipsychotics.
Sydney, Australia, the setting for this study, more than a quarter are
Most have a long history of continuous
antipsychotic use, although current guidelines stipulate review and
withdrawal within 12 weeks.
Reduction in antipsychotic prescribing is feasible with re-
emergence of behavioral symptoms in only a minority of partic-
However, one key trial in people with Alzheimer’s
disease (AD) living in the community or assisted-living residences
found that BPSD increased after antipsychotic discontinuation.
Although a systematic review of interventions to reduce inappro-
priate prescribing of antipsychotics in people with dementia in LTC
concluded that 9 of 11 well-designed interventions were effective,
only 4 studies evaluated long-term effects, at up to 9 months post-
intervention, and conclusions were mixed.
In this trial, we aimed to achieve a reduction in inappropriate use
of antipsychotic medications for at least 12 months, without an in-
crease in BPSD or use of substitute psychotropic medications.
Because successful withdrawal depends on the implementation of
effective support strategies involving all relevant stakeholders, anti-
psychotic withdrawal was implemented by general practitioners
(GPs), following speciﬁc training for nursing staff, GPs, and pharma-
cists, and awareness raising with family members. This concerted
approach aimed to encourage a culture change in prescribing practices
and in the prevention and management of BSPD. LTC nursing homes
were approached and asked to recruit residents on antipsychotics for
study participation. As the Australian Government “funding of initia-
tives to improve aged care service delivery”proscribed randomized
controlled trials, we used repeated measures, including 2 pre-
intervention assessments, to assess changes in medication use, BPSD,
and adverse outcomes, including social withdrawal, cognition, falls,
We hypothesized that antipsychotic withdrawal would not be
associated with an increase in BPSD and would be beneﬁcial in
reducing adverse effects.
Study Design and Ethics
The Halting Antipsychotic use in Long Term care (HALT) study was
a single-arm longitudinal study in a convenience sample of 23 LTC
residences (12 private, 10 charitable, and 1 community-based, each
with >60 beds) in Sydney and surrounding areas conducted between
April 2014 and October 2016. In Australia, LTC is government subsi-
dized according to level of dependency, and residents can choose
which GP they consult. No nursing home in this study had dedicated
in-house doctors. Community pharmacists are rebated for undertak-
ing medication reviews.
The study was approved by the UNSW Ethics Committee
(HC13203) and registered with the Australian New Zealand Clinical
Trials Registry (no. 12614000309684). The study design has been
described in detail.
Brieﬂy, prebaseline and baseline assessments
were conducted approximately 1 month apart, before antipsychotic
withdrawal was initiated, to control for changes over time in the
absence of a control group. Follow-up assessments occurred after 3, 6,
and 12 months (Figure 1).
Adults aged 60 or older, living in LTC for 1 month, and on regular
antipsychotic medication for 3 months, without primary psychotic
illness such as schizophrenia or bipolar disorder, were identiﬁed via
an appointed registered nurse at each nursing home.
written consent from residents and/or their “person responsible”
(proxy) and consent from their general practitioner (GP; because the
deprescribing agreement was reliant on GP involvement), research
psychologists determined eligibility via staff interviews and ﬁle au-
dits. Residents with a terminal illness or very severe BPSD at baseline
were excluded [nursing home version of the Neuropsychiatric In-
ventory (NPI-NH) score 50; domain scores of 12 for at least 2 of
delusions, hallucinations, agitation/aggression, anxiety or disinhibi-
tion, and occupational disruptiveness score 4 for at least 2 of these
domains]. This exclusion criterion was an ethical consideration based
on previous ﬁndings of a signiﬁcant worsening of symptoms on
deprescribing among those with severe BPSD and based on contem-
poraneous Australian Department of Health rules for a severe
Dementia was not an inclusion criterion, as we relied on ﬁle audits
and a dementia diagnoses are not always recorded. However, 137 par-
ticipants (98.5%) were considered to have dementia based on docu-
mented diagnosis (n ¼130), or after review of all available data,
including cognitive assessment (Psychogeriatric AssessmenteCognitive
Impairment Scale [PAS-CIS]
) by an experienced old-age psychiatrist
(H.B.) (n ¼7).
The intervention comprised 2 components: (1) education/training
of health care staff, and (2) deprescribing antipsychotic medication.
The former included education and training opportunities for desig-
nated LTC staff (“Champions”), participants’GPs, and pharmacists
involved in the supply of medication services to participating nursing
Champions, usually 1 to 2 registered nurses from each home,
attended a 3-day workshop on dementia, BPSD, and person-centered,
nonpharmacologic approaches to BPSD prevention, reduction, and
management. Champions were encouraged to implement training
strategies at their respective nursing home to educate other staff
members. Participants’GPs were offered academic detailing involving
a 30- to 60-minute peer education session, additional reading material
on antipsychotic use, and an optional follow-up educational seminar
and evaluation. Pharmacists were offered a continuing professional
development module. Continuing education credits were awarded to
GPs and nursing staff.
The second component was an individualized deprescribing pro-
tocol, which followed Australian guidelines stipulating a dose reduc-
tion of 50% every 2 weeks and ceasing after 2 weeks on the minimum
dose, withdrawing one antipsychotic at a time, with risperidone (if
prescribed) to be withdrawn last.
Study pharmacists developed a
protocol for each participant that was provided to nursing staff at
baseline (Supplemental Figure 2). GPs were instructed to commence
the outlined medication change(s) within the next week. If GPs
considered emergency medication may be required, consistent with
current best practice guidelines,
they were recommended if con-
cerned about deprescribing antipsychotics to prescribe a short-acting
benzodiazepine (oxazepam or lorazepam) pro-re-nata (PRN). GPs
could restart antipsychotic therapy at any time if they determined it
H. Brodaty et al. / JAMDA 19 (2018) 592e600 593
was indicated. If extra support was needed, Champions, GPs, and
pharmacists could consult the study team.
Assessment Procedures and Outcome Measures
Antipsychotic and sedative prescription and administration
Research psychologists audited participant medication charts to
determine prescribed and administered dosage of regular and PRN
antipsychotics, sedatives, and antidepressants. Doses of antipsychotics
and benzodiazepines were converted to olanzapine and diazepam
equivalents, respectively, for analysis (Supplemental Table 1).
BPSD and adverse outcomes
Research psychologists interviewed participants (where possible)
and nursing staff. Staff assessments comprised the nursing home
version of the NPI (NPI-NH; range 0-144, higher scores indicating
more frequent and/or severe BPSD; weighted
: 0.93), the Cohen-
Mansﬁeld Agitation Inventory (CMAI; range 29-203, higher scores
indicating more frequent manifestations of agitation; weighted
0.98; the item “trying to get to a different place”was erroneously
excluded from the questionnaire package; therefore, the total score
was pro-rated for all participants), and the Multidimensional Obser-
vation Scale for Elderly Subjectsdwithdrawn behavior subscale
Fig. 1. Consort diagram. Appointed nurses at each of the 24 nursing homes that agreed to participate recruited residents aged 60 or older, living in current nursing home for
1 month and on regular antipsychotic medication for 3 months. After obtaining written consent from the resident and/or their “person responsible”(proxy), research psy-
chologists determined eligibility based on the absence of a primary psychotic illness and the severity of BPSD. Information on medication use, falls and hospitalizations was
determined by ﬁle audit. BPSD, socially withdrawn behavior and cognition required (complete) assessment by research psychologists.
H. Brodaty et al. / JAMDA 19 (2018) 592e600594
(range 8-32, higher scores indicating more withdrawal, weighted
Cognition was assessed with the PAS-CIS (range 0-21,
higher scores indicating worse performance;
: 0.84) or, for those
from noneEnglish-speaking backgrounds, the Rowland Universal
Dementia Assessment Scale (RUDAS; range 0-30, lower scores indi-
cating worse performance).
Both PAS-CIS and RUDAS evaluate
multiple cognitive domains and are recommended by the Australian
Department of Health to establish care needs.
Total scores were pro-rated when up to 20% of items were
missing. RUDAS scores were converted to the same scale as the PAS-
CIS, using a score of 21 on RUDAS and 5 on PAS-CIS as equivalent
For some participants, it was not possible to complete
resident or staff interviews at designated follow-up time points
(most often due to participant having deceased recently or lack of
an interpreter) or at (pre-)baseline because GPs enthusiastically but
erroneously commenced the deprescription protocol prior to study
assessment). File audit data were included where possible
Residents’records provided number of falls and hospitaliza-
tions prior to and during study follow-up. Information was gath-
ered from incident reports and all progress notes. Frequencies in
the 6 months prior to study commencement (or pro-rated for
those in the nursing home for less than 6 months) and the last
6 months of the study were compared. Data on falls were analyzed
taking into account mobility, deﬁned as able to ambulate with or
without assistance (ie, bedbound participants were not included
in the analysis of falls).
SPSS version 22 and GraphPad Prism version 6.07 were used for
analysis. Statistical signiﬁcance was set at 0.05 and the Holm-
Bonferroni or Bonferroni method were used to correct for multiple
testing. Differences between 2 groups were evaluated via Fisher exact
(categorical) or Student ttest (continuous data).
Baseline Characteristics of Completers Versus Noncompleters and of Successful vs Unsuccessful Deprescribing
Noncompleters Completers Deprescribed
referring to the notes
46 93 69 (74.2) 24 (25.8)
Gender, n females (%) 31 (67.4) 61 (65.6) 51 (73.9) 10 (41.7)
Age, y, mean SD (range) 88.2 ± 7.3 (71.2-101.8) 84.3 ± 7.3 (59.5-99.7) 83.8 7.4 (59.5-99.7) 83.3 6.9 (67.6-94.3)
P[.004; t[2.964; df [137
Time at facility, y, mean SD (range) 3.7 3.0 (0.5-15.3) 2.5 7.3 (0.1-7.6) 2.3 1.7 (0.1-7.5) 2.4 2.0 (0.5-7.6)
Daily dose in olanzapine equivalents, mg, median; IQR 1.88; 2.5 1.88; 2.5 1.88; 2.6 3.75; 3.75
Risperidone, n (%) 27 (58.7) 58 (62.4) 38 (55.1) 20 (83.3)
Olanzapine, n (%) 7 (15.2) 11 (11.8) 9 (13.0) 2 (8.3)
Quetiapine, n (%) 7 (15.2) 18 (19.4) 16 (23.2) 2 (8.3)
Haloperidol, n (%) 5 (10.9) 9 (9.7) 8 (11.6) 1 (4.2)
PRN prescribed, n (%) 17 (37.0) 14 (15.1) 10 (14.5) 6 (25.0)
PRN administered, n (%) 7 (15.2) 3 (3.2) 1 (1.4) 2 (8.3)
Time on antipsychotics, y, mean SD (range) 2.4 1.9 (0.4-8.1) 2.4 1.8 (0.1-8.1) 2.4 1.8 (0.1-8.1) 2.4 1.8 (0.4-7.0)
NPI-NH total score median; IQR (n [%]) 22.0; 22.0 (35 [76.1]) 26.6; 24.8 (70 [75.3]) 25.0; 22.0 (52 [75.4]) 31.6; 23.6 (19 [79.2])
Presence of BPSD, n (%)
Delusions 6/29 (20.7) 17/63 (27.0) 13/47 (27.7) 4/16 (25.0)
Hallucinations 5/34 (14.7) 9/74 (12.2) 6/53 (11.3) 3/21 (14.3)
Agitation/Aggression 25/35 (71.4) 65/73 (89.0) 47/53 (88.7) 18/20 (90.0)
Depression 12/35 (34.3) 32/72 (44.4) 30/51 (58.8) 2/21 (9.5)
Anxiety 13/35 (37.1) 39/73 (53.4) 31/53 (58.5) 8/20 (40.0)
Elation/Euphoria 4/34 (11.8) 14/74 (18.9) 8/53 (15.1) 6/21 (28.6)
Apathy 24/35 (68.6) 57/74 (77.0) 40/53 (75.5) 17/21 (81.0)
Disinhibition 10/35 (28.6) 23/73 (31.5) 13/52 (25.0) 10/21 (47.6)
Irritability/lability 21/35 (60.0) 54/73 (74.0) 38/53 (71.7) 16/20 (80.0)
Ab motor behavior 16/35 (45.7) 33/73 (45.2) 21/52 (40.4) 12/21 (57.1)
Sleep and nighttime behavior disorders 12/35 (34.3) 22/70 (31.4) 13/52 (25.0) 9/18 (50.0)
Appetite changes 9/32 (28.1) 13/66 (19.7) 9/46 (19.6) 4/20 (20.0)
CMAI, median; IQR (n [%]) 51.7; 28.8 (35 [76.1]) 51.6; 24.9 (72 [77.4]) 48.7; 22.5 (52 [75.4]) 58.0; 28.0 (19 [79.2])
MOSES, median; IQR (n [%]) 24.0; 9.0 (41 [89.1]) 24.0; 10.25 (78 [83.9]) 24.0; 9.5 (52 [75.4]) 26.0; 7.0 (19 [79.2])
None 27 (58.7) 43 (46.2) 37 (53.6) 6 (25.0)
1 fall 19 (41.3) 50 (53.8) 32 (46.4) 18 (75.0)
NPI-NH, Neuropsychiatric Inventory - Nursing Home
; CMAI, Cohen-Mansﬁeld Agitation Inventory
; MOSES, Multidimensional Observation Scale for Elderly Subjects
withdrawn behavior subscale.
For 3 individuals who dropped out between baseline and prebaseline, prebaseline data were used instead of baseline data. Differences were tested via Fisher exact test or
Student ttest. Bold values indicate signiﬁcant differences, adjusting for multiple testing within each column and set of characteristics (Demographics, Antipsychotic use, BPSD,
and Adverse outcomes).
There was no signiﬁcant difference in attrition in the group that has their regular antipsychotic medication deprescribed, compared to that with regular antipsychotic
medication never deprescribed or represcribed (Fisher exact test: P¼.83). Deprescribed and never represcribed; as indicated in Supplemental Figure 6.
*Falls in the 6-month period prior to prebaseline.
H. Brodaty et al. / JAMDA 19 (2018) 592e600 595
The primary outcome was reduction in antipsychotic use
comparing pre- and postintervention. For all secondary outcomes
except falls and hospitalizations, linear mixed models were used to
assess whether changes occurred on deprescribing. Variables included
in the models were age, sex, time, and regular antipsychotic medica-
tion use (ﬁxed) and subject and nursing home (random). As such, the
data of all participants could be included in the analysis, including that
of participants who were represcribed or never deprescribed, and
participants who dropped out during follow-up. Covariance between
repeated measures was speciﬁed according to an autoregressive
model of order 1. Where applicable, noninferiority for deprescription
was evaluated (Supplemental Figure 3).
Role of the Funding Source
The study sponsor(s) played no role in study design (apart from
funding rules precluding randomized controlled trials); in the
collection, analysis, and interpretation of data; in the writing of the
report; or in the decision to submit the paper for publication. The
corresponding author had full access to all data in the study and takes
ﬁnal responsibility for the decision to submit for publication.
Participant ﬂow is summarized in the Consort diagram (Figure 1).
Twenty-four LTC nursing homes with 3447 beds were recruited; 149
residents consented to participate (range: 2-22 per nursing home). Of
139 eligible participants from 23 different nursing homes, 93 (66.9%)
completed the study to 12-month follow-up. Reasons for attrition were
death (n ¼36), consent revocation [n ¼5; 3 of which occurred prior to
(completion of) intervention], relocation (n ¼3), terminal illness
(n ¼1), and nursing home withdrawal (n ¼1). Three participants
dropped out prior to the baseline assessment and 3 more shortly af-
terwards, meaning 133 participants provided at least some post-
intervention data and were included in analyses. Table 1 compares the
baseline characteristics, including baseline antipsychotic use for par-
ticipants who completed the study with those who dropped out at any
point (noncompleters). Among the 93 completers, participants whose
regular antipsychotics were sustainably deprescribed (ie, antipsychotics
were ceased and were not represcribed) are compared to those whose
antipsychotics were represcribed or never deprescribed (Table 1 ).
Of the 133 participants for whom postintervention data were
available, regular antipsychotic medication was ceased for 126 (94.7%).
Complete cessation of regular antipsychotics was achieved after
27 days on average (range: 0-78 days, plus 2 participants whose
medication was ceased after >5 months; Supplemental Figure 1). At
follow-ups, 86.2% participants were not on regular antipsychotics at
3 months, 79.1% at 6 months, and 81.7% at 12 months post interven-
tion (Figure 2). When assessing dose reduction, 93.5%, 87.3%, and
90.3% of all participants were administered none, or less than half of
Fig. 2. Regular daily dose of antipsychotic medication, expressed in terms of olanzapine equivalents.
H. Brodaty et al. / JAMDA 19 (2018) 592e600596
their original dose of regular antipsychotic, at the 3-, 6-, and12-month
follow-ups, respectively. Antipsychotic dose was initially reduced but
never ceased for 7 (6.3%) participants. Of these, 4 participants dropped
out shortly after intervention and 3 completed the study but remained
on regular antipsychotics throughout the entire 12 months (albeit
with dose halved for 2).
Of the 126 residents whose medication was ceased, 28 (22.2%)
were subsequently represcribed a regular antipsychotic
(Supplemental Figure 6), on average 2.7 months after original
cessation (range: 7 dayse8.4 months; Supplemental Figure 1);
20 (71.4%) were represcribed risperidone, 3 (10.7%) olanzapine,
1 quetiapine (3.6%), and 4 (14.3%) haloperidol (later supple-
mented with quetiapine for one participant). For 8 of these 28
participants, antipsychotics were deprescribed a second time, and
for 1 they were subsequently represcribed again. Accordingly, 105
of the total sample of 133 participants (78.9%) with post-
intervention data were no longer taking regular antipsychotics at
their last assessment.
Fig. 3. Scores for BPSD and adverse outcomes over time. (A) NPI-NH, (B) CMAI, (C) MOSES withdrawn behavior, and (D) PAS-CIS total scores for all participants at each assessment.
Proportions of participants who (E) fell or (F) were hospitalized in the 6 months prior to the intervention and at the end of the study. Start of deprescription is indicated with a red
dashed line. Blue lines and error bars indicate median and interquartile range. NPI-NH, Neuropsychiatric Inventory Nursing Home. CMAI, Cohen-Mansﬁeld Agitation Inventory.
MOSES, Multidimensional Observation Scale for Elderly Subjects.
H. Brodaty et al. / JAMDA 19 (2018) 592e600 597
Sustainability of deprescribing
The sustainability of deprescribing was assessed in the 93 partic-
ipants that completed the full 12-month follow-up period; antipsy-
chotics were deprescribed successfully for 69 of 93 completers (74.2%;
ie, regular antipsychotic ceased and not represcribed during the study
follow-up). They were not on regular antipsychotics for 11.5 months
on average (range: 9.4-14.4 months, with 1 outlier at 4.3 months).
Even when also considering the 29 noncompleters who were
deprescribed antipsychotics (69 þ29 ¼98 participants), the median
length of cessation remained high at 11.1 months. Among the group of
28 participants who were represcribed a regular antipsychotic, the
average total time postintervention not taking regular antipsychotics
was 4.1 months (range: 7 dayse11 months); 16 (57.1%) remained free
of regular antipsychotics for 3 months or more, and 12 (42.9%) were
represcribed a lower dose than their initial dose of antipsychotics.
Overall, there was a signiﬁcant reduction in antipsychotic use after
intervention: 81.7% (95% CI: 72.4-89.0) of participants (n ¼76/93) had
their regular antipsychotics deprescribed at 12 months.
Substitution: PRN antipsychotics and sedatives
Approximately 1 in 5 participants was prescribed a PRN antipsy-
chotic prior to baseline (Supplemental Table 2). PRN antipsychotic
administration was infrequent and occurred at a low dose. They were
administered to only 5 participants after ceasing regular medication, 3
of whom received less than 2.5 mg olanzapine (or equivalent) on
average monthly. Linear mixed model analysis of the data across the
entire sample (n ¼139) showed there was no signiﬁcant increase in
PRN antipsychotic prescribing (
¼0.3 mg/mo, P¼.33, 97.5% CI: 1.0
to 0.4, t ¼0.98, df ¼474) or administration (
¼0.9 mg/mo, P¼.31,
97.5% CI: 3. 0 to 1.1, t ¼1.03, df ¼110) on deprescribing of regular
antipsychotics (Supplemental Figure 3).
We checked whether the deprescribed regular antipsychotic
medication was substituted with other psychotropic medications such
as benzodiazepines. There was no increase in regular benzodiazepine
use, but more participants were administered a PRN benzodiazepine,
particularly lorazepam and oxazepam (Supplemental Table 3). The
proportion of participants administered PRN benzodiazepines
increased from 11.0% and 8.2% at prebaseline and baseline, respec-
tively, to 23.1%, 25.4%, and 29.8% at 3, 6, and 12 months, respectively.
However, PRN benzodiazepine administration occurred for less than a
third of participants at any time point, and most received 10 mg
diazepam equivalents on average per month or 0.3 mg/d (Supple-
mental Figure 4A). Moreover, this increase in benzodiazepine use
was independent of regular antipsychotic deprescribing (
mo, P¼.528, 95% CI: 0.99 to 1.92; t ¼0.63, df ¼396; Supplemental
Figure 3). In other words, although PRN use of benzodiazepine
increased throughout the study, this increase was very low and not
linked to deprescription of regular antipsychotic medication.
Antidepressant use remained stable throughout the entire study
period (Supplemental Figure 4B).
Participants in our sample showed no signiﬁcant behavioral
changes over the study period (Figure 3). There was no signiﬁcant
change in total NPI-NH score on deprescription (
¼1.0 point; ie,
total NPI score was 1 point lower when not on regular antipsychotics,
P¼.58, 97.5% CI: 5.03 to 3.02, t ¼0.56, df ¼450; Figure 3A and
Supplemental Figure 3). Agitation/aggression, as measured by total
CMAI score, did not increase with deprescribing (
¼1.7 point, ie,
total CMAI was 1.7 points lower when not on regular antipsychotics,
P¼.37, 97.5% CI: 5.77 to 2.47, t ¼0.90, df ¼435; Figure 3B and
Supplemental Figure 3). Speciﬁc behaviors were analyzed based on
individual NPI-NH domain scores (Supplemental Figure 5), but
antipsychotic deprescribing was not signiﬁcantly associated with
changes in any domain after correcting for multiple testing.
There was no signiﬁcant interaction effect between deprescribing
and the administration of PRN antipsychotic or benzodiazepine
medication on either NPI or CMAI total scores (data not shown).
We used the Multidimensional Observation Scale for Elderly Sub-
jects to verify whether our deprescribing intervention reduced social
withdrawal, but found no signiﬁcant change (Figure 3C;
the Multidimensional Observation Scale for Elderly Subjects score was
0.27 points higher when not regular antipsychotics, P¼.52, 97.5%
CI: 0.67 to 1.22, t ¼0.65, df ¼379; Supplemental Figure 3).
There were strong ﬂoor effects on cognitive scores (Figure 3C), but
there was no signiﬁcant improvement in cognition associated with
¼0.22, ie, (converted) PAS score was 0.22 points
higher when not on regular antipsychotics, P¼.56, 97.5% CI: 0.67 to
1.12 , t ¼0.56, df ¼245; Supplemental Figure 3].
Falls and hospitalizations were analyzed for the 6-month periods
prior to the intervention (when participants were on regular anti-
psychotics) and at the end of the study (when most were either
withdrawn or prescribed a lower dose of antipsychotics). When
considering only participants who were mobile, 54.2% fell at least once
in the 6 months prior to study enrolment (n ¼39/72 mobile partici-
pants; who fell 3.5 times on average, SD: 3.2, range: 1-12). This was
not signiﬁcantly different from the ﬁnal 6 months of the study, when
44.7% fell at least once (n ¼22/47 mobile participants; who fell 3.2
times on average, SD: 3.1, range: 1-13; Figure 3E).
There was no difference in attrition between mobile and immobile
participants (Fisher exact test: P¼.45). Accordingly, there were no
differences in falls pre- and postintervention for the mobile partici-
pants who completed the study and whose antipsychotics were
deprescribed (56.3% prior to intervention vs 42.4% after; P¼.32).
For hospitalizations, the proportion of participants hospitalized
was 18.0% at preintervention (n ¼25, mean: 1.1 hospitalization, SD:
0.4, range: 1-3) and 12.9% (n ¼12, mean: 1.4, SD: 0.7, range: 1-3) post
intervention, a nonsigniﬁcant decrease (Figure 3F).
Among participants who completed the study and were success-
fully deprescribed, 18.8% were hospitalized at least once in the period
before the intervention, compared to 8.8% at the end of the study; this
decrease was not statistically signiﬁcant (P¼.14). Baseline hospitali-
zation rate was not linked to attrition (P¼.16).
The Halting Antipsychotic use in Long-Term care (HALT) study
demonstrated successful and sustainable reduction of antipsychotic
use in a sample of older adults residing in LTC who were recruited by
nursing staff, without an associated rise in BPSD (Figure 3A, B).
Withdrawal was not accompanied by an increase in substitute regular
or PRN antipsychotics. Although PRN benzodiazepines were pre-
scribed for more participants, they were administered for a minority
of participants, infrequently, at very low doses (<0.3 mg/d in diaz-
epam equivalents), and independent of antipsychotic withdrawal
(Supplemental Figure 4). There was no signiﬁcant improvement in
adverse outcomes, such as withdrawn behaviors, falls, hospitaliza-
tions, and cognitive decline.
Four in 5 participants had their regular antipsychotic successfully
deprescribed compared to about 1 in 5 in other studies.
this high success rate may partially be driven by residence and resi-
dent selection (see limitations below), it is also in line with previous
ﬁndings that active engagement of all involved stakeholders, as in this
study, is linked to greater success.
In contrast to most other studies,
H. Brodaty et al. / JAMDA 19 (2018) 592e600598
our intervention involved multiple components. Providing additional
skills to nurses by nurses (via a train-the-trainer approach), to phar-
macists by pharmacists, and to GPs by an academic GP was a truly
multidisciplinary approach. Good leadership and focused training
providing clear practical directions on pharmacologic and non-
pharmacologic approaches for the prevention and management of
BPSD. We hypothesize that the application of best practice principles
of deprescribing in a clear, individualized way, which held both
nursing staff and GPs accountable to enact within a speciﬁc time
frame, was a key factor in the success of HALT.
The use of PRN benzodiazepines increased slightly (most likely as a
consequence of the HALT protocol allowing inclusion of a PRN pre-
scription of oxazepam or lorazepam, if deemed necessary), but this
increase was small in terms, <0.3 mg/d in diazepam equivalents, and
independent of antipsychotic withdrawal.
Participants who did or did not have their antipsychotics suc-
cessfully deprescribed showed few differences at baseline (Table 1),
and there were no signiﬁcant effects on any of the secondary outcome
measures on antipsychotic withdrawal. Interestingly, depression
scores at baseline were signiﬁcantly higher in the group that had their
antipsychotic medication successfully deprescribed. This suggests that
antipsychotics may have been administered, perhaps unsuccessfully,
as an add-on to antidepressant medication (antidepressant medica-
tion use remained stable throughout the entire study).
Overall, there was a large variation in BPSD over time and between
participants. We found no relationship between use of antipsychotics
and severity of BPSD over 12 months’follow-up (Supplemental
Figure 3). Devanand et al reported increased BPSD on antipsychotic
withdrawal for AD patients who had previously responded to risper-
idone for agitation or psychosis treatment.
could explain the differing outcomes, as we have no knowledge of the
appropriateness of the original prescription or the treatment response
for our sample.
There was no statistically signiﬁcant decrease in adverse events.
The lack of improvement could have reﬂected the natural history of
these outcomes over time (see limitations). Cognitive assessment
was limited by ﬂoor effects, reduced power due to frequent lack of
interpreters (31.7% of participants did not have English as a
preferred language), and lack of assessment of speciﬁc cognitive
domains. Our study was powered to measure decline in antipsy-
chotic prescription rates (primary outcome), but was insufﬁciently
powered for small effect sizes in secondary outcomes. Non-
inferiority analysis did indicate there was no increase in BPSD. A
larger sample and more sensitive assessment measures, ideally in a
randomized controlled trial, might conﬁrm the beneﬁcial effects
found in earlier studies.
The success of HALT should be interpreted with consideration of a
number of limitations. First, in a single-arm trial it is difﬁcult to tease
out whether any of the observed changes aredue to the intervention or
another time-dependent change in the outcome. For example, as de-
mentia severity increases, cognition and BPSD change over time (eg,
aggression declines and apathy increases), which may mask positive
effects of deprescribing. We attempted to control for time-dependent
changes by including 2 preintervention assessments 1 month apart
that did not demonstrate variability albeit over a limited time period
(see Figure 2). Also, we cannot ascertain whether theantipsychotic was
never or no longer indicated, or if other (nonspeciﬁc) aspects of the
intervention, such as greater care and vigilance, obviated their need.
Second, in common with most antipsychotic deprescribing in-
HALT had multiple components, including multidis-
ciplinary education and medication review. This precludes
identiﬁcation of component(s) or professional group(s) that were
instrumental in achieving such a high rate of deprescribing. On the
other hand, this multipronged strategy is a strength, and possibly
the essential factor for sustainable reduction of antipsychotic use.
A third limitation lies in nonrandom recruitment of nursing homes,
of which 24/58 agreed to join the study (and one subsequently
withdrew), and residents. Because of an ethical requirement for arms-
length recruitment, nurse champions were relied on to identify resi-
dents. Although they were instructed to approach families of every
resident who met eligibility criteria, selective recruitment of partici-
pants who were stable and perceived to no longer need the medica-
tion is possible. We expected that a much larger proportion of
residents from participating nursing homes would have been pre-
scribed antipsychotic medication. Because we do not have access to
this information, it is difﬁcult to assess the representativeness of our
sample. Based on the literature, we can estimate to have recruited
between 15% and 20% of all potentially eligible residents in the
participating nursing homes.
Fourth, a third of our sample dropped out, most commonly
because of death. Some outcomes may be skewed because of the loss
of primarily older participants at more advanced stages of dementia
(Table 1). However, our sensitivity analysis, conservatively allowing
that all noncompleters would have been represcribed regular anti-
psychotics, shows that we would still achieve a high success rate (50%
reduction in participants taking antipsychotics). We reiterate that
there was no signiﬁcant difference in attrition between participants
whose antipsychotic medication was deprescribed or not.
Fifth, neither staff nor assessors were blind to previous and current
medication administration. Because we did not ﬁnd signiﬁcant dif-
ferences in any of our secondary outcomes, reporting bias is unlikely
to have affected our conclusions.
Overall, our results highlight the feasibility and safety of with-
drawing antipsychotics for at least a subgroup of people with de-
mentia in LTC without (re)emergence of BPSD. Accepting the potential
for selection bias, the results of this study suggest that many people
with dementia in LTC who currently use antipsychotic medication
stand to beneﬁt from more active review practices.
Further analysis might provide more insight into differences be-
tween groups who were represcribed antipsychotics and those who
were not. In addition, qualitativedata collected from participants, care
staff and GPs will be used to explored the context of antipsychotic (re)
Even though health care policy and guidelines advise against
antipsychotic use in people with dementia, our ﬁndings demonstrate
that implementation of correct review and withdrawal practices is
currently lacking in LTC nursing homes. Anecdotal information from
the revocations for this study further reinforce that a conservative
approach to care, rather than actual symptom management, may
underlie the continued prescription of antipsychotic medication in
older nursing home residents.
Further research should focus on the speciﬁc strengths of the
different components of the multifaceted intervention implemented
in this study. A comprehensive strategy involving all stakeholders in
the care process likely improves its success rate, but cost and feasi-
bility are important limitations for its application in routine practice.
Previous studies have demonstrated the cost-effectiveness of imple-
menting training in person-centered dementia care.
The results of this trial should encourage antipsychotic withdrawal
in people with dementia living in LTC, following existing guidelines,
and the implemented measures could serve as a blueprint for wider
integration into dementia care practices.
We thank all participating residents, their families, GPs, pharma-
cists, and nursing staff. We thank the members of the steering com-
mittee, Jacqui Close, Juanita Westbury, Lee-Fay Low, Aine Heaney,
Marie Alford, Stacy Wake, Janet Mitchell, Millie Marinkovich, and
Jenny Blennerhassett. Linda Nattrass, Anne-NicoleCasey, Emily Trigge,
H. Brodaty et al. / JAMDA 19 (2018) 592e600 599
and Tiffany Chau are thanked for their assistance with administration
and data entry.
1. Brodaty H, Draper B, Saab D, et al. Psychosis, depression and behavioural dis-
turbances in Sydney nursing home residents: prevalence and predictors. Int J
Geriatr Psychiatry 2001;16:504e512.
2. Katz IR, Jeste DV, Mintzer JE, et al. Comparison of risperidone and placebo for
psychosis and behavioral disturbances associated with dementia: a randomized,
double-blind trial. Risperidone Study Group. J Clin Psychiatry 1999;60:107e115.
3. Ballard C, Corbett A, Chitramohan R, Aarsland D. Management of agitation and
aggression associated with Alzheimerʼs disease: controversies and possible
solutions. Curr Opin Psychiatry 2009;22:532e540.
4. Schneider LS, Dagerman KS, Insel P. Risk of death with atypical antipsychotic
drug treatment for dementia: meta-analysis of randomized placebo-controlled
trials. JAMA 2005;294:1934e1943.
5. Schneider LS, Dagerman K, Insel PS. Efﬁcacy and adverse effects of atypical
antipsychotics for dementia: Meta-analysis of randomized, placebo-controlled
trials. Am J Geriatr Psychiatry 2006;14:191e210.
6. Rochon PA, Normand S-L, Gomes T, et al. Antipsychotic therapy and short-term
serious events in older adults with dementia. Arch Intern Med 2008;168:1090e1096.
7. Vigen CLP, Mack WJ, Keefe RSE, et al. Cognitive effects of atypical antipsychotic
medications in patients with Alzheimer’s disease: Outcomes from CATIE-AD.
Am J Psychiatry 2011;168:831e839.
8. Declercq T, Petrovic M, Azermai M, et al. Withdrawal versus continuation of
chronic antipsychotic drugs for behavioural and psychological symptoms in
older people with dementia. In: Declercq T, editor. Cochrane Database of Sys-
tematic Reviews. Chichester, UK: John Wiley & Sons, Ltd; 2013. p. CD007726.
9. Gallini A, Andrieu S, Donohue JM, et al. Trends in use of antipsychotics in
elderly patients with dementia: Impact of national safety warnings. Eur Neu-
10. Briesacher BA, Tjia J, Field T, et al. Antipsychotic use among nursing home
residents. JAMA 2013;309:440.
11. Stock KJ, Amuah JE, Lapane KL, et al. Prevalence of, and resident and facility
characteristics associated with, antipsychotic use in assisted living vs. long-
term care facilities: A cross-sectional analysis from Alberta. Canada. Drugs
12. Janus SIM, van Manen JG, IJzerman MJ, Zuidema SU. Psychotropic drug pre-
scriptions in Western European nursing homes. Int Psychogeriatr 2016;28:
13. Snowdon J, Galanos D, Vaswani D. Patterns of psychotropic medication use in
nursing homes: Surveys in Sydney, allowing comparisons over time and be-
tween countries. Int Psychogeriatr 2011;23:1520e1525.
14. Ballard C, Hanney ML, Theodoulou M, et al. DART-AD Investigators. The De-
mentia Antipsychotic Withdrawal Trial (DART-AD): Long-term follow-up of a
randomised placebo-controlled trial. Lancet Neurol 2009;8:151e157.
15. Royal Australian and New Zealand College of Psychiatrists. The Use of Anti-
psychotics in Residential Aged Care. Clinical Recommendations; 2011.
16. Ballard C, Lana MM, Theodoulou M, et al. A randomised, blinded, placebo-
controlled trial in dementia patients continuing or stopping neuroleptics
(The DART-AD Trial). PLoS Med 2008;5:0587e0599.
17. Fossey J, Ballard C, Juszczak E, et al. Effect of enhanced psychosocial care on
antipsychotic use in nursing home residents with severe dementia: Cluster
randomised trial. BMJ 2006;332:756e761.
18. van Reekum R, Clarke D, Conn D, et al. A randomized, placebo-controlled trial
of the discontinuation of long-term antipsychotics in dementia. Int Psycho-
19. Ruths S, Straand J, Nygaard HA, et al. Effect of antipsychotic withdrawal on
behavior and sleep/wake activity in nursing home residents with dementia: A
randomized, placebo-controlled, double-blinded study the Bergen District
Nursing Home Study. J Am Geriatr Soc 2004;52:1737e1743.
20. Devanand DP, Mintzer J, Schultz SK, et al. Relapse risk after discontinuation of
risperidone in Alzheimer’s disease. N Engl J Med 2012;367:1497e1507.
21. Thompson Coon J, Abbott R, Rogers M, et al. Interventions to reduce inap-
propriate prescribing of antipsychotic medications in people with dementia
resident in care homes: A systematic review. J Am Med Dir Assoc 2014;15:
22. Jessop T, Harrison F, Cations M, et al. Halting Antipsychotic Use in Long-Term
care (HALT): A single-arm longitudinal study aiming to reduce inappropriate
antipsychotic use in long-term care residents with behavioral and psycholog-
ical symptoms of dementia. Int Psychogeriatr; 2017:1e13.
23. Australian Department of Health. Dementia and Severe Behaviours Supple-
ment. Canberra, AU: Australian Government; 2013.
24. Jorm A, Mackinnon AJ, Henderson AS, et al. The Psychogeriatric Assessment
Scales: A multi-dimensional alternative to categorical diagnoses of dementia
and depression in the elderly. Psychol Med 1995;25:447e460.
25. The Royal Australian & New Zealand College of Pyschiatrists. Assessment and
Management of People with Behavioural and Psychological Symptoms of De-
mentia (BPSD)dA Handbook for NSW Health Clinicians; 2013.
26. Mandrekar JN. Measures of interrater agreement. J Thorac Oncol 2011;6:6e7.
27. Wood S, Cummings JL, Hsu MA, et al. The use of the neuropsychiatric inventory
in nursing home residents. Characterization and measurement. Am J Geriatr
28. Cohen-Mansﬁeld J, Billig N. Agitated behaviors in the elderly. I. A conceptual
review. J Am Geriatr Soc 1986;34:711e721.
29. Pruchno RA, Kleban MH, Resch NL. Psychometric assessment of the Multidi-
mensional Observation Scale for Elderly Subjects (MOSES). J Gerontol 1988;43:
30. Storey JE, Rowland JTJ, Basic D, et al. The Rowland Universal Dementia
Assessment Scale (RUDAS): A multicultural cognitive assessment scale. Int
31. Patterson SM, Hughes CM, Crealey G, et al. An evaluation of an adapted U.S.
model of pharmaceutical care to improve psychoactive prescribing for nursing
home residents in Northern Ireland (Fleetwood Northern Ireland Study). J Am
Geriatr Soc 2010;58:44e53.
32. Lawrence V, Fossey J, Ballard C, et al. Improving quality of life for people with
dementia in care homes: Making psychosocial interventions work. Br J Psy-
33. Declercq T, Petrovic M, Azermai M, et al. Withdrawal versus continuation of
chronic antipsychotic drugs for behavioural and psychological symptoms in
older people with dementia. Cochrane Database Syst Rev 2013;3:1e80.
34. Ballard C, Orrell M, Psych FRC, et al. Impact of antipsychotic review and
nonpharmacological intervention on antipsychotic use, neuropsychiatric
symptoms, and mortality in people with dementia living in nursing homes:
A factorial cluster-randomized controlled trial by the Well-Being and Health
for People With Dementia (WHELD) Program. Am J Psychiatry 2016;173:
35. Chenoweth L, King MT, Jeon YH, et al. Caring for Aged Dementia Care Resident
Study (CADRES) of person-centred care, dementia-care mapping, and
usual care in dementia: A cluster-randomised trial. Lancet Neurol 2009;8:
H. Brodaty et al. / JAMDA 19 (2018) 592e600600
Supplemental Table 2
PRN Antipsychotic Prescriptions and Administrations
Olanzapine Equivalents Preintervention, n (%) Postintervention, n (%)
Prebaseline (n ¼139) Baseline (n ¼136) 3 mo (n ¼124) 6 mo (n ¼110) 12 mo (n ¼93)
Prescribed None 105 (75.5) 104 (76.5) 98 (79.0) 94 (85.5) 74 (79.6)
0 < x ≤ 2.5 mg
28 (20.1) 26 (19.1) 23 (18.5) 12 (10.9) 14 (15.1)
2.5 < x ≤ 5 mg
4 (2.9) 5 (3.7) 1 (0.8) 2 (1.8) 4 (4.3)
> 7.5 mg
2 (1.4) 1 (0.7) 2 (1.6) 2 (1.8) 1 (1.4)
Administered N/A 103 (74.1)*103 (75.7)
98 (79.0) 93 (84.5)
None 18 (12.9) 25 (18.4) 18 (14.5) 10 (9.1) 9 (9.7)
0 < x ≤ 2.5 mg
7 (5.0) 4 (2.9) 5 (4.0) [1 no reg AP] 0 8 (8.6) [3 no reg AP]
2.5 < x ≤ 5 mg
5 (3.6) 1 (0.7) 3 (2.4) [1 no reg AP] 2 (1.8) 3 (3.2) [1 no reg AP]
0 0 0 3 (2.7) [2 no reg AP] 0
> 7.5 mg
3 (2.2) 2 (1.5) 0 1 (0.9) 0
3 (2.2) 1 (0.7) 1 (0.9)
Note: Time frame for administration: amount of olanzapine (in milligrams) administered in one month [monthly average over last 3 months (for prebaseline) or in the time
period since the previous assessment (for all other assessments)].
[“X”no reg AP] ¼individuals who were not prescribed regular antipsychotic medication at the current time of assessment.
*Three participants were administered PRN antipsychotics even though not prescribed.
One person was administered PRN antipsychotics even though not prescribed.
Supplemental Table 1
Equivalents for Antipsychotic Medications and Benzodiazepines
Antipsychotics (Eq to 1 mg
Benzodiazepines (Eq to 10 mg
Haloperidol 0.533 mg Alprazolam 0.75 mg
Risperidone 0.267 mg Clobazam 15 mg
Quetiapine 20 mg Clonazepam 0.5 mg
Lorazepam 1.5 mg
Midazolam 8.75 mg
Nitrazepam 10 mg
Oxazepam 30 mg
Temazepam 20 mg
Triazolam 0.5 mg
H. Brodaty et al. / JAMDA 19 (2018) 592e600600.e1
Supplemental. Fig. 1. Time proﬁle of participants’de- and represcription. Time from baseline to complete cessation of all regular antipsychotic medication (solid line; % of total
n¼133, ie, all participants with postbaseline data; see text for details) and the time between baseline and (ﬁrst) represcription of a regular antipsychotic (dashed line, % of total
n¼126 participants who had ceased after intervention).
Supplemental Table 3
Benzodiazepine Use: Participants on Regular and PRN Benzodiazepines (and Frequency of PRN Administration)
Preintervention, n (%) Postintervention, n (%)
Prebaseline (n ¼139) Baseline (n ¼136) 3 mo (n ¼124) 6 mo (n ¼110) 12 mo (n ¼93)
Diazepam 1 (0.7) 1 (0.7) 1 (0.8) 0 1 (1.1)
Lorazepam 1 (0.7) 1 (0.7) 1 (0.8) 1 (0.9) 2 (2.2)
Oxazepam 3 (2.2) 3 (2.2) 6 (4.8) 2 (1.8) 3 (3.2)
Temazepam 21 (15.1) 19 (14.0) 17 (13.7) 17 (15.5) 16 (17.2)
Diazepam 3 (2.2)
Lorazepam 1 (0.7)
Oxazepam 1 (0.7)
Temazepam 9 (6.5)
Other sedatives such as midazolam, nitrazepam, triazolam, clobazam, clonazepam, and melatonin were administered to less than 2% of the sample population at any time
point (and there were no signiﬁcant differences across the different timepoints). These data are not included in the table.
H. Brodaty et al. / JAMDA 19 (2018) 592e600 600.e2
Supplemental Fig. 2. Example of participant deprescribing protocol.
H. Brodaty et al. / JAMDA 19 (2018) 592e600600.e3
Supplemental Fig. 3. Sample conﬁdence intervals and inference for trials assessing
superiority or noninferiority of regular antipsychotic prescription and deprescription.
0s for deprescription of regular antipsychotics are as fol-
lows: for PRN medication prescription (olanzapine or diazepam equivalents for anti-
psychotics and benzodiazepines, respectively), 5 mg/mo; for NPI-NH total score, 4
points; and for CMAI total score, 6 points. Conﬁdence intervals indicated are 97.5% to
correct for multiple testing, except for benzodiazepine PRN administration, which is
95%. Antipsychotic withdrawal was not inferior to regular antipsychotic administration
in terms of PRN administration of antipsychotics or benzodiazepines, or in terms of
NPI-NH and CMAI total scores. There was no signiﬁcant change in MOSES withdrawn
behavior or PAS-CIS total scores on withdrawal.
Supplemental Fig. 4. Substitution: (A) PRN benzodiazepine administration, expressed
in terms of diazepam equivalents; (B) antidepressant use.
H. Brodaty et al. / JAMDA 19 (2018) 592e600 600.e4
Supplemental Fig. 5. NPI domain scores. Domain scores (severity by frequency score) for the NPI-NH at 5 assessments. Blue lines and error bars indicate median and interquartile
H. Brodaty et al. / JAMDA 19 (2018) 592e600600.e5
Supplemental Fig. 5. continued
H. Brodaty et al. / JAMDA 19 (2018) 592e600 600.e6
Supplemental Fig. 6. Sankey diagram of regular antipsychotic prescription patterns and attrition. Participant ﬂow in terms of participant inclusion for intervention and at 12-
month follow-up is indicated and patterns of de- and represcription of regular antipsychotic medication are indicated (irrespective of at which time point during the study de-
or represcription occurred).
1. Leucht S, Samara M, Heres S, Patel MX, Woods SW, Davis JM. Dose equivalents for
second-generation antipsychotics: The minimum effective dose method. Schiz-
ophr Bull 2014;40:314e326.
2. CooperAJ. Benzodiazepines: towardsmore logicaluse. ScottMed J 1982;27:297e304.
3. Goodman LS, Gilman A, Brunton LL. Goodman & Gilman’s
Manual of Pharmacology and Therapeutics. New York: McGraw-Hill
4. Ruiz P, Strain EC, Lowinson JH. Lowinson and Ruiz’s Substance Abuse: A
Comprehensive Textbook. Philadelphia: Wolters Kluwer Health/Lippincott Wil-
liams & Wilkins; 2011.
H. Brodaty et al. / JAMDA 19 (2018) 592e600600.e7