Impact of COVID-19 restrictions on behavioural and psychological symptoms in home-dwelling people with dementia: a prospective cohort study (PAN.DEM).

OBJECTIVES: To investigate the impact of the COVID-19 restrictions on behavioural and psychological symptoms of dementia (BPSD).
DESIGN: Prospective cohort study (PAN.DEM) nested within the halted parent trial (LIVE@Home.Path).
SETTING: Households in Norway immediate before and 6-9 weeks into the COVID-19 restrictions. PARTICIPANTS: 104 dyads (persons with mild to moderate dementia aged ≥65 and their informal carers) completed both prepandemic and pandemic assessments, among 237 in the parent trial. Mini-Mental Status Examination score 15-26 or Functional Assessment Staging score 3-7 covered dementia severity. MAIN OUTCOME MEASURES: Neuropsychiatric Inventory (NPI-12) total (range 0-144), psychosis (range 0-24), hyperactive behaviour (range 0-60) and mood subsyndrome (range 0-48) scores; Cornell Scale for Depression in Dementia (CSDD) total score (range 0-38).
RESULTS: We found an overall increase in BPSD by NPI-12 total score comparing prepandemic to pandemic levels (median 16 IQR (4.5-29) to 20 (7-32.5), p=0.03) over a mean of 86 days (SD 19). NPI-12 total score worsened in 57 (55%) of people with dementia and was associated with postponed or averted contacts with healthcare professionals (logistic regression, OR 3.96, 95% CI 1.05 to 14.95). Psychosis subsyndrome levels increased (0 (0-3) to 0.5 (0-6), p=0.01) in 37 (36%) persons; this worsening was associated with partial insight (9.57, 1.14 to 80.71) and reduced informal carer contact (4.45, 1.01 to 19.71). Moreover, depressive symptoms increased as assessed by CSDD total score (5 (3-9) to 7 (4-12), p=0.01) and worsened for 56 (54%), which was inversely associated with psychotropic drugs on-demand (0.16, 0.03 to 0.75).
CONCLUSIONS: BPSD worsened during the first months of the COVID-19 restrictions, most pronounced for psychosis and depression. These BPSD exacerbations have implications for pandemic policies, emphasising that restrictions must balance COVID-19 morbidity and mortality against dementia deterioration. TRIAL REGISTRATION NUMBER: NCT04043364; Results. © Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

This is the first prospective cohort study investigating the impact of the COVID-19 restrictions on behavioural and psychological symptoms of dementia (BPSD).

The same informal carers reported BPSD for each home-dwelling person with dementia both before and during the pandemic scenario using validated, well-established instruments.

The COVID-19 restrictions left some informal carers with less basis of observation, as 28% reported reduced contact with the person with dementia.

Our study captures the impact of the initial phase of the outbreak in Norway and does not describe the long-term impact of the COVID-19 restrictions on BPSD.

Introduction

Dementia is among the most critical risk factors for COVID-19 mortality.1 In England and Wales alone, 12 869 people with dementia have died, accounting for 26% of the COVID-19 death toll.2 Until vaccination is widely available globally, hygiene and physical distancing interventions will remain cornerstones of protecting vulnerable populations.3 The subsequent restrictions have been disrupting for home-dwelling people with dementia as private homes were not accessible to family members and volunteers, day care centres closed and home nursing services were restricted to those most in need. As a result, people with dementia living in the community are not only at risk from COVID-19 morbidity and mortality; they are also threatened from unforeseen effects of the restrictions.4 5

Behavioural and psychological symptoms of dementia (BPSD) cover a wide range of clinical presentation including depression, anxiety, agitation and psychosis. Longitudinally, persistent BPSD may be found in up to 80% of people with dementia.6 BPSD are best managed with structured, non-pharmacological interventions, placing psychotropic drugs as secondary treatment options.7 Preliminary evidence indicates that BPSD may be exacerbated under the COVID-19 restrictions. Eight weeks into the Argentinian quarantine, informal carers reported worsening of anxiety, insomnia and depression among persons at different stages of Alzheimer’s and related dementias living at home (N=119).8 In another study, family carers stated worsening BPSD in 60% of Italian outpatients with various stages and aetiologies of dementia 1 month into the pandemic (N=4913).9 This study also found that 28% required changes in psychotropic medication to address irritability, apathy, agitation and depression. Further, nursing home patients separated from the outside world in France with mild Alzheimer’s disease reported increased anxiety and depression when asked to evaluate their own experience of the pandemic retrospectively (N=58).10

However, all these studies are cross-sectional and thus far, there is a dearth of longitudinal data tracking changes in BPSD during COVID-19 by comparing prepandemic to pandemic rates.11 In this study, we aim to address this significant gap in the literature using data from the prospective PAN.DEM study.12 This study is nested within the ongoing LIVE@Home.Path trial13 and was launched by our team to investigate the impact of the COVID-19 restrictions (implemented in Norway on 12 March 2020) on home-dwelling people with dementia. Here, we present comparisons of prepandemic and pandemic BPSD, and explore factors associated with worsening BPSD.

Methods

Study design

This is a prospective cohort study comparing the prepandemic assessment of BPSD of the parent trial, LIVE@Home.Path, to the PAN.DEM assessment.

Setting

The parent trial is a stepped-wedge randomised controlled trial.13 It compares the cost-effectiveness in resource utilisation of a 6-month multicomponent intervention comprising Learning, Innovation, Volunteers and Empowerment to usual conditions for dyads of home-dwelling people with dementia and their informal carers. Trained data collectors blindly assessed all dyads in direct conversation every 6 months for 2 years (2019–2021). The prepandemic 6-month assessment was close to complete when the COVID-19 restrictions replaced trial protocol (figure 1A). Physical distancing (ie, restrictions on gatherings, public transport closure, stay at home-regulations and limitations on movement) formed the basis for the restrictions,3 which implied that healthcare was limited to those most in need.12 In response, we developed the semistructured PANdemic in DEMentia (PAN.DEM) telephone interview for informal carers to capture if, and how, dyads were affected by the outbreak (online supplemental file). This assessment included selected instruments from the parent trial in addition to questions regarding the pandemic. We consecutively invited as many dyads as possible from the parent trial to complete the PAN.DEM assessment from week 6 of restrictions until eased the 9th week (20 April 2020 to 15 May 2020). Potential respondents were considered unreachable when no response was given to two calls and a text message.

Figure 1

The parent trial, LIVE@Home.Path, including PAN.DEM. The COVID-19 restrictions replaced trial protocol from 12 March to eased on 15 May 2020. None of the dyads (person with dementia and informal carer, n) received the intervention while the PAN.DEM interviews were conducted (20 April 2020 to 15 May 2020). (A) Timeline. Vertical lines indicate assessments. The shaded parts illustrate the COVID-19 restrictions, postponing the Learning, Innovation, Volunteers and Empowerment (LIVE-Intervention) for the dyads of group 2. (B) Flow chart. This study includes the dyads of PAN.DEM completing the prepandemic assessment before the COVID-19 restrictions was implemented on 12 March 2020. *Parent trial attrition: rate within assumptions of lost to follow-up.

Participants

Dyads were eligible for inclusion in the parent trial if the persons with dementia were: ≥65 years, diagnosed with dementia (with Mini-Mental Status Examination (MMSE) score 15–26 or Functional Assessment Staging (FAST) score 3–7)14 15; home-dwelling in one of three Norwegian municipalities; and had weekly face-to-face contact with the informal carer. Dyads gave informed spoken and written consent for participation in the parent trial as described in the protocol.13 Informal carers gave additional informed consent to PAN.DEM.12

Measurements

The primary outcome was change in BPSD between the prepandemic and pandemic assessments. We administered two informal carer-rated scales at both time points: (1) The Neuropsychiatric Inventory (NPI-12) assesses frequency and severity of delusions, hallucinations, agitation, depression, anxiety, euphoria, apathy, disinhibitions, irritability, aberrant motorial behaviour, sleep disturbances and appetite changes over the four preceding weeks.16 Each of these 12 domains is scored from 0 (no symptoms) to 12 (very severe symptoms), a score ≥4 is regarded a BPSD with symptom load of clinical relevance.6 These domains are further aggregated to generate subsyndrome scores for psychosis comprised of delusions and hallucinations (0–24), hyperactive behaviour comprised agitation, euphoria, irritation, disinhibition, aberrant motor behaviour (0–60), mood comprised depression, apathy, sleep disturbances and appetite changes (0–48), and finally, a total NPI-12 score (0–144);17 (2) The Cornell Scale for Depression in Dementia (CSDD) assesses nineteen items of depressive symptoms during the prior week, each rated from ‘absent’ to ‘severe’ (0–2) or ‘symptoms not possible to evaluate’ (missing).18 Adding item scores generate the CSDD total score (0–38).18 A CSDD total score ≥8 indicates depression of clinical relevance.19 The Norwegian versions of NPI-12 and CSDD have robust psychometric properties.16 18–20

In addition to BPSD, we collected the following data at the prepandemic assessment: the persons with dementia’s level of functioning in activities of daily living by Physical Self-Maintenance Scale (PSMS)21 and Instrumental Activities of Daily Living Scale (IADL),22 health by the General Medical Health Rating Scale (GMHR),23 possible dementia aetiology following the International Classification of Diseases-10th version,24 and use of healthcare services and medications as specified by the dyads. Drugs catalogued in the Anatomical Therapeutic Chemical Index (ATC) administered in a set schedule were regarded ‘regular’, whereas all others were documented as ‘on demand’.25 Psychotropic drugs included antipsychotic (N05A), anxiolytic (N05B), hypnotic and sedative (N05C), antidepressant (N06A) and antiementia drugs (N06D) by ATC. Demographical data (age, gender, residency, kinship) were self-reported. We evaluated dementia severity in terms of cognition with MMSE and level of functioning with FAST at inclusion.14 15

At the pandemic assessment, the informal carers were also asked to estimate the degree of insight presented by the person with dementia into the COVID-19 situation and change in (1) contact with the informal carer, (2) volunteering services and (3) municipal healthcare services (home nursing services, home help, day-care, and in-home and out-of-home respite care) due to the COVID-19 restrictions.12 Finally, informal carers stated if contacts with healthcare professionals were postponed or averted.

Study size

This study includes all dyads in PAN.DEM completing the prepandemic assessment before the COVID-19 restrictions were effectuated (figure 1B).

Statistical methods

Initially, we aggregated median and IQR, and calculated NPI-12 subsyndrome scores and total scores for NPI-12 and CSDD if >80% of the scales were answered. We used the Wilcoxon matched-pairs signed-rank test to assess change between the prepandemic and pandemic assessments. Next, we dichotomised those NPI-12 and CSDD sum scores that changed into worsening/not worsening and used multiple logistic regression analysis to explore factors associated. We included the following covariates for persons with dementia: age, gender, residency, dementia aetiology, MMSE, FAST, IADL, PSMS, GMHR, number of psychotropic drugs prescribed regularly and on-demand, and the COVID-19 specific outcomes. We also included age and gender of the informal carers. Covariates were selected based on our expertise in research and clinical dementia care. The Akaike information criterion guided model selection. Selected models were then checked for multicollinearity, robustness and goodness-of-fit by Pearson and Hosmer-Lemeshow test. FAST, IADL and PSMS showed moderate to strong positive correlation, but including all three covariates substantially improved the models. Missing data were handled with listwise deletion, with 14% missing any covariates. Calculations are expressed in OR with 95% CI, and p value. Reported p values are two tailed, and p<0.05 was considered statistically significant. Descriptive statistics are presented by n (%), mean (SD), or median (IQR). We used Stata/IC, release V.16 (StataCorp) for all analyses.

Public and Patient involvement

The conceptualisation, design, assessments and conduct of the parent trial as well as PAN.DEM included close patient/informal carer and public involvement.12 13 A user-representative participated in the research group’s weekly meetings. In PAN.DEM, he consulted with the study team on priorities, length and wording of the interview, and its revisions, with a special focus on the potential burden on informal carers.12

Results

Of the 280 dyads participating in the parent trial, 237 completed the prepandemic assessment from December 2019 to March 2020 (figure 1B). This study includes 104 dyads recruited to PAN.DEM completing the prepandemic assessment before the COVID-19 restrictions were effectuated 12 March 2020. Mean time between assessments was 86 days (SD 19).

Table 1 shows that the mean age for people with dementia was 82 years (SD 7), 61% were women, 44% lived alone, and 50% received daily home-nursing services prior to the COVID-19 restrictions. Alzheimer’s disease constituted the most common dementia aetiology, while 6% had vascular dementia and 10% reported Lewy-body dementia or Parkinson’s disease. Most people with dementia lacked insight into the COVID-19 situation (table 2). The informal carers reported to have less contact with the person with dementia in 28% under the restrictions, and that contacts with healthcare professionals had been postponed or averted in 31%.

Table 1

Prepandemic characteristics for the 104 dyads (persons with dementia and informal carers, n)

	N=104
Person with dementia
Age, mean (SD)	82 (7)
Female gender, n (%)	63 (61)
Residency
Living alone, n (%)	46 (44)
Coresiding with the reporting informal carer, n (%)	46 (44)
Coresiding with someone else than the informal carer, n (%)	12 (12)
Dementia aetiology
Alzheimer’s disease, n (%)	45 (43)
Vascular dementia, n (%)	6 (6)
Dementia in other diseases classified elsewhere, n (%)	10 (10)
Unspecified dementia, n (%)	43 (41)
MMSE, range 0–30, median (IQR)	21(18–24)
FAST, range 1–7, median (IQR)	4 (4–4)
GMHR, range 1–4, median (IQR)	3 (2–3)
PSMS, range 6–30, median (IQR)	11 (9–14)
IADL, range 8–31, median (IQR)	22 (18–27)
Drugs in general
Total number, median (IQR)	6 (4–8)
Regularly, median (IQR)	5 (3–7)
Psychotropic drugs
Total no, median (IQR)	1 (0–2)
Regularly, median (IQR)	1 (0–1)
Antipsychotic drugs (N05A), n (%)	6 (6)
Anxiolytic drugs (N05B), n (%)	3 (3)
Hypnotic/sedative drugs (N05C), n (%)	10 (10)
Antidepressant drugs (N06A), n (%)	19 (18)
Antidementia drugs (N06D), n (%)	52 (50)
On-demand, median (IQR)	0 (0–0)
Antipsychotic drugs (N05A), n (%)	0 (0)
Anxiolytic drugs (N05B), n (%)	5 (5)
Hypnotic/sedative drugs (N05C), n (%)	12 (12)
Antidepressant drugs (N06A), n (%)	0 (0)
Antidementia drugs (N06D), n (%)	0 (0)
Volunteering services, n (%)	8 (8)
Healthcare services
Daily home nursing, n (%)	52 (50)
Weekly day care, n (%)	29 (28)
Respite care (In-home and out-of-home), n (%)	2 (2)
Informal carer
Age, mean (SD)	65 (12)
Female gender, n (%)	68 (65)
Kinship to the person with dementia
Spouse, n (%)	44 (42)
Child, n (%)	58 (56)
Others, n (%)	2 (2)

Drugs were classified by the Anatomical Therapeutic Chemical Index; antipsychotics, anxiolytics, hypnotics/sedatives, antidepressants and antidementia drugs constituted psychotropic drugs.

FAST, Functional Assessment Staging; GMHR, General Medical Health Rating Scale; IADL, Instrumental Activities of Daily Living Scale; ICD-10, International Classification of Diseases10th version; MMSE, Mini-Mental Status Examination; Prepandemic, Six-month assessment of parent trial (12 December 2019 to 11 March 2020); PSMS, Physical Self-Maintenance Scale.

Table 2

Pandemic characteristics for the 104 persons with dementia (n) as perceived by their informal carers

	N= 104
Degree of insight
Sufficient, n (%)	34 (33)
Partial, n (%)	54 (52)
To no degree, n (%)	16 (15)
Change in contact with the informal carer*
Reduced, n (%)	29 (28)
No change, n (%)	49 (47)
Increased, n (%)	23 (22)
Ceased volunteering services*, n (%)	8 (8)
Change in healthcare services*, n (%)	42 (40)
Postponed or averted contacts with healthcare professionals*, n (%)	32 (31)

*Relative the prepandemic situation. Healthcare services provided by the municipality: home nursing services, home help, day-care and respite care (in-home and out-of-home).

Pandemic, PAN.DEM assessment (20 April 2020 to 15 May 2020).

From the prepandemic to the pandemic assessment, people with dementia experienced an increase in NPI-12 total score (16 (4.5–29) to 20 (7–32.5), p=0.03) and in numbers of BPSD with symptom load of clinical relevance (2 (0–4) to 3 (1 –5), p<0.001) (table 3). Also, the NPI-12 score worsened for 55% (figure 2). We found an increase in the psychosis subsyndrome (0 (0–3) to 0.5 (0–6), p=0.01), with 36% experiencing more severe symptoms (figure 2). We also found an increase in depressive symptoms measured both by the NPI-12 depression domain (0 (0–3) to 1 (0–6), p=0.04) and CSDD total score (5 (3–9) to 7 (4–12), p=0.01, table 3). Additionally, the CSDD total score worsened for 54% (figure 2).

Table 3

Prepandemic compared with pandemic behavioural and psychological symptoms for the 104 persons with dementia (n)

	Prepandemic			Pandemic			P value
	N (%) with symptom load of clinical relevance*	Median	IQR	N (%) with symptom load of clinical relevance*	Median	IQR
Neuropsychiatric inventory (NPI-12)
Total score, range 0–144		16	4.5–29		20	7–32.5	0.03†
Subsyndromes
Psychosis‡, range 0–24		0	0–3		0.5	0–6	0.01†
Hyperactive behaviour§, range 0–60		5.5	0–12		4	0–12	0.79
Mood¶, range 0–48		6	0–12		6.5	1–12	0.21
Domain scores, range 0–12
Delusions	20 (19)	0	0–2	31 (30)	0	0–6	0.04†
Hallucinations	8 (8)	0	0–0	16 (15)	0	0–0	0.23
Agitation	23 (22)	0	0–3	18 (17)	0	0–2	0.45
Depression	25 (24)	0	0–3	40 (38)	1	0–6	0.04†
Anxiety	18 (17)	0	0–2	31 (30)	0	0–4	0.07
Euphoria	8 (8)	0	0–0	4 (4)	0	0–0	0.19
Apathy	35 (34)	0	0–4	30 (29)	0	0–4	0.50
Disinhibitions	9 (9)	0	0–0	15 (14)	0	0–1.5	0.16
Irritability	28 (27)	0	0–4	29 (28)	0	0–4	0.78
Aberrant motor behaviour	23 (22)	0	0–1	24 (23)	0	0–2.5	0.66
Sleep disturbances	25 (24)	0	0–3	28 (27)	0	0–4	0.82
Appetite changes	14 (13)	0	0–1	17 (16)	0	0–1	0.84
No of BPSD with symptom load of clinical relevance*, range 0–12		2	0–4		3	1–5	<0.001†
Cornell Scale for Depression in Dementia (CSDD)
Total score, range 0–38	34 (33)	5	3–9	41 (39)	7	4–12	0.01†

*NPI domain scores ≥4 indicate BPSD with symptom load of clinical relevance. CSDD total score ≥8 indicates depression of clinical relevance.

†Indicates two-tailed p<0.05.

‡ Psychosis: delusions and hallucinations

§ Hyperactive behaviour: agitation, euphoria, irritation, disinhibition, aberrant motor behaviour

¶ Mood: depression, apathy, sleep disturbances and appetite changes

BPSD, behavioural and psychological symptoms of dementia; P, p value for difference in median between time points by the Wilcoxon matched-pairs signed-rank test; Pandemic, PAN.DEM assessment (20 April 2020 to 15 May 2020); Prepandemic, Six-month assessment of parent trial (12 December 2019 to 11 March 2020).

Figure 2

Change in behavioural and psychological symptoms in n (%) persons with dementia from the prepandemic to the pandemic assessment. n: 104. Prepandemic: Six-month assessment of parent trial (12 December 2019 to 11 March 2020). Pandemic: PAN.DEM assessment (20 April 2020 to 15 May 2020). Neuropsychiatric Inventory, subsyndrome score: psychosis (delusions and hallucinations), hyperactive behaviour (agitation, euphoria, irritation, disinhibition, aberrant motor behaviour), mood (depression, apathy, sleep disturbances and appetite changes). Cornell Scale for Depression in Dementia, total score.

Table 4 shows the results of the logistic regression models exploring factors associated with worsening BPSD under the restrictions. Worsening NPI-12 total score was associated with postponed or averted contacts with healthcare professionals (OR 3.96, 95% CI 1.05 to 14.95) and impaired cognition as indicated by MMSE (OR 1.19, 95% CI 1.01 to 1.40), while a diagnosis of Alzheimer’s disease relative to other dementia aetiologies was associated with lower OR of worsening NPI-12 (OR 0.18, 95% CI 0.05 to 0.63). Worsening psychosis subsyndrome score was associated with partial insight into the COVID-19 situation (OR 9.57, 95% CI 1.14 to 80.71), reduced contact with the informal carer (OR 4.45, 95% CI 1.01 to 19.71), and impaired function as indicated by FAST (OR 2.59, 95% CI 1.07 to 6.27). An inverse association occurred for higher dependency in activities of daily living by PSMS and worsening psychosis subsyndrome (OR 0.68, 95% CI 0.51 to 0.91). Worsening depressive symptoms was associated with impaired function by FAST (OR 4.96, 95% CI 1.57 to 15.65), in contrast to lower odds associated with Alzheimer’s disease (OR 0.21, 95% CI 0.05 to 0.85) and psychotropic drug use on-demand (OR 0.16, 95% CI 0.03 to 0.75).

Table 4

Factors associated with worsening in behavioural and psychological symptoms of dementia from the prepandemic to the pandemic assessment

Covariates	NPI-12 total score				NPI-12 psychosis subsyndrome				CSDD total score
	OR	95% CI		P value	OR	95% CI		P value	OR	95% CI		P value
		Lower	Upper			Lower	Upper			Lower	Upper
Prepandemic characteristics
Person with dementia
Age	1.01	0.92	1.11	0.79	0.91	0.82	1.01	0.16	1.09	0.97	1.22	0.16
Female gender	0.51	0.13	1.98	0.34	0.36	0.09	1.52	0.09	0.19	0.03	1.31	0.09
Living alone	0.20	0.04	1.01	0.05	2.69	0.41	17.80	0.31	0.55	0.07	4.18	0.57
Alzheimer’s disease*	0.18	0.05	0.63	0.01¶¶	0.84	0.23	3.08	0.79	0.21	0.05	0.85	0.03¶¶
MMSE†	1.19	1.01	1.40	0.04¶¶	0.97	0.82	1.14	0.68	0.96	0.80	1.15	0.65
FAST‡	0.98	0.45	2.16	0.97	2.59	1.07	6.27	0.04¶¶	4.96	1.57	15.65	0.01¶¶
IADL§	0.96	0.80	1.15	0.64	1.19	0.98	1.45	0.08	0.84	0.67	1.07	0.16
PSMS¶	1.00	0.79	1.28	0.99	0.68	0.51	0.91	0.01¶¶	0.99	0.76	1.29	0.96
GMHR**	0.91	0.36	2.32	0.84	2.06	0.72	5.88	0.18	0.84	0.28	2.50	0.76
Psychotropic drugs††
Regularly	1.16	0.54	2.48	0.71	0.67	0.31	1.47	0.32	1.11	0.49	2.53	0.80
On-demand	0.35	0.09	1.46	0.15	2.95	0.69	12.66	0.15	0.16	0.03	0.75	0.02¶¶
Informal carer
Age	0.97	0.92	1.03	0.40	1.04	0.98	1.12	0.21	0.99	0.93	1.06	0.87
Female gender	1.81	0.50	6.49	0.36	0.70	0.18	2.80	0.62	0.82	0.16	4.27	0.82
Pandemic characteristics, person with dementia
Insight to the COVID-19 situation‡‡
Partial	0.61	0.10	3.69	0.60	9.57	1.14	80.71	0.04¶¶	0.67	0.10	4.44	0.68
Sufficient	1.14	0.15	8.82	0.90	3.69	0.33	40.93	0.29	2.70	0.26	28.27	0.41
Contact with the informal carer§§
Reduced	1.88	0.48	7.44	0.37	4.45	1.01	19.71	0.049¶¶	1.40	0.27	7.27	0.69
Increased	2.41	0.61	9.49	0.21	3.21	0.71	14.55	0.13	0.30	0.07	1.23	0.10
Ceased volunteering services	0.30	0.04	2.24	0.24	0.20	0.02	2.11	0.18	0.59	0.04	7.91	0.69
Change in healthcare services	0.48	0.13	1.78	0.28	0.48	0.11	2.08	0.33	1.16	0.28	4.83	0.84
Postponed or averted contacts with healthcare professionals	3.96	1.05	14.95	0.04¶¶	1.55	0.45	5.42	0.49	3.37	0.70	16.08	0.13

Change dichotomised into worsening/not worsening. OR explored by multiple logistic regression, estimates adjusted for all other factors in the models.

*Alzheimer’s disease, reference: all other dementia aetiologies.

†MMSE, range 0–30, higher scores indicate better cognition, reference: 30.

‡FAST, range 1–7, lower scores indicate better functioning, reference: 1.

§IADL, range 8–31, lower scores indicate better functioning, reference: 8.

¶PSMS, range 6–30, lower scores indicate better functioning, reference 6.

**GMHR, range 1–4, lower score indicate higher comorbidity burden, reference 4.

††Number of psychotropic drugs according to the Anatomical Therapeutic Chemical Index: antipsychotics (N05A), anxiolytics (N05B), hypnotics/sedatives (N05C), antidepressants (N06A) and anti-dementia drugs (N06D), reference: 0.

‡‡Degree of insight into the COVID-19 situation as perceived by the informal carer, reference: no insight.

§§Change in contact with the informal carer, reference: no change.

¶¶P: two-tailed p<0.05

CSDD, Cornell Scale of Depression in Dementia; FAST, Functional Assessment Staging, at inclusion; GMHR, General Medical Health Rating Scale; IADL, Instrumental Activities of Daily Living Scale; MMSE, Mini-Mental Status Examination, at inclusion; n, 89 dyads (person with dementia and informal carer); NPI-12, Neuropsychiatric Inventory, twelve item version, with psychosis subsyndrome constituting delusions and hallucinations; Pandemic, PAN.DEM assessment (20 April 2020 to 15 May 2020); Prepandemic, Six-month assessment of parent trial (12 December 2019 to 11 March 2020); PSMS, Physical Self-Maintenance Scale.

Post hoc analysis did not show any association between use of antipsychotic drugs before the restrictions and worsening psychosis subsyndrome using unequal variances t-test (online supplemental table A). Similarly, we found no association between use of antidepressants and worsening depressive symptoms. Neither randomisation to the intervention vs control of the parent trial showed associations with worsening NP-12 total score, psychosis subsyndrome nor depressive symptoms (online supplemental table A). To explore if consecutive sampling introduced bias, we compared our study sample to those not included yet still in parent trial at the prepandemic assessment, revealing minimal differences (online supplemental table B).

Discussion

Our primary aim was to compare prepandemic and pandemic levels of BPSD in home-dwelling people with dementia during the two first months of COVID-19 restrictions in Norway. Even though BPSD fluctuates over the dementia course, our study indicates that the COVID-19 restrictions caused an overall increase in BPSD over a mean of 86 days, and that odds of worsening were four times higher with postponed or averted contacts with healthcare professionals. More specifically, the increase was most pronounced for symptoms of psychosis and depression. The odds for worsening psychosis increased 10-fold with partial insight into the COVID-19 situation and 4-fold with reduced contact with informal carers, while as-needed use of psychotropic drugs was associated with fewer depressive symptoms.

Strengths and weaknesses

Our study provides prospective data obtained shortly before and under the COVID-19 restrictions rated by the same informal carer for each subject and based on extensive assessor-blinded interviews with validated, well-established instruments.12 13 We used established cut-off scores when presenting BPSD with symptom load of clinical relevance.6 19 The parent trial population was recruited from different municipalities to be representative to the Norwegian demographic in terms of dementia aetiology, severity and symptomatology.13 As our study sample was fairly similar to those dyads not included from the parent trial, we argue that our study was not biased by selection.

There are weaknesses to address. Despite efforts, we were not able to invite all potential respondents through consecutive sampling before the restrictions were eased for the first time, explaining the limited sample size. CSDD is not validated for telephone interviews18 yet our findings using CSDD were consistent with the depression domain of NPI-12, which can be used as a telephone interview instrument.16 Previous work has shown that carer psychosocial factors such as sense of competence, guilt and relationship quality account for up to 56% of the variance in BPSD-related distress.26 In the case of the pandemic, stress-related symptoms were experienced by two-thirds of family carers soon after the outbreak hit Italy (N=4913) and were associated with incident or worsening BPSD.9 The authors conclude that they could not determine whether increased BPSD were the cause or consequence of carer distress, as both counterparts were exposed to similar conditions during quarantine. Even though we did not assess such domains, these considerations apply to our study. Another point is that 28% of the informal carers reported reduced contact with the person with dementia, leaving them with less clinical observation. As 44% of the dyads were not living together, we suggest that some violated the restrictions to visit their loved ones and keep their obligations as careers, possibly mitigating the impact on BPSD. These weaknesses should be considered when interpreting the results, along with the wide CIs of the covariates associated with worsening BPSD. Notably, our data capture the impact of the initial phase of the outbreak in Norway and can therefore not answer longer-term consequences from either reimposition or lengthening of invasive restrictions.

Comparison with other studies

This study provides data on the negative mental health consequences of the COVID-19 restrictions for people with dementia. Using a non-randomised, non-controlled design to evaluate causations may be reasonable in the pandemic scenario as no other way of assessing the impact of the COVID-19 restrictions exist. However, our results should be interpreted with caution. The deterioration in BPSD could in theory be caused by the progression of the dementia syndrome itself, rather than being exacerbated by the pandemic restrictions. Arguing against this, change in BPSD over 4 months was substantially lesser in an observational cohort of nursing home residents of which the majority had dementia than what we demonstrate comparing prepandemic and pandemic symptom levels.27

Our findings echo a small body of the existing literature on this topic. A study from Spain noted increases in levels of agitation, apathy, and aberrant motor behaviour 5 weeks into lockdown in outpatients with mild cognitive impairment and Alzheimer’s disease (N=40), but no increase in psychotic symptoms.28 A cross-sectional study from Italy (N=139) describes exacerbation of psychotic symptoms in a small percentage of subjects with subjective cognitive decline, mild cognitive impairment and dementia.29 This study, in part, used self-assessments, that may have led to underreporting of delusions and hallucinations. Even though other studies are equivocal on whether psychosis worsened,8 9 UK registry data indicate higher antipsychotic prescription rates to people with dementia during the pandemic, and the authors speculate that this increase may be the result of worsened agitation and psychosis.30 Meanwhile, our study revealed no associations between psychotropic drugs and psychosis, likely given that very few patients used antipsychotics before the pandemic, in addition to the lack of real-time prescription data throughout the outbreak. Because this is a nascent area of research, discrepancies may be attributed to heterogeneity in design, as well as dementia severity and aetiology.

Early findings suggest that older adults at group level are more resilient to the mental health effects of the pandemic than younger ones.11 Nonetheless, our study adds to the cross-sectional reports calling attention to deteriorating depressive symptoms among people with dementia.8–10 For better communication within and between dyads and their formal caregivers, digital devices may enhance individual support.12 Further, anxiolytics and hypnotics/sedatives were associated with fewer depressive symptoms when used as-needed in our sample. These drugs are known to temporarily alleviate some of the symptoms assessed by the CSDD, such as anxiety, irritability and agitation. However, in line with national guidelines, we rather recommend that antidepressants are considered if severe symptoms persist.31

Our study supports the WHO’s concerns that the pandemic would negatively impact the mental health of people with cognitive impairments.5 Even though way of life varies globally, the policies implemented in response to COVID-19 are likely equally disruptive to the environment of home-dwelling people with dementia across nations.3 We, therefore, argue that our findings are generalisable to other countries. Furthermore, they emphasise that non-pharmacological approaches still should be the first-line treatment to avoid BPSD deterioration regardless of context.

Unanswered questions and future research

Future research should explore the long-term impact of the COVID-19 restrictions on BPSD, and whether moderations or service innovations can mitigate worsening. Less than 5% of trials on COVID-19 involve behavioural and mental health interventions,32 emphasising the need for knowledge to adapt restrictions and navigate the unforeseeable consequences for persons with dementia and informal caregiver of the current, and future, pandemics.