This is a multicentric, nation-wide survey. Eighty-nine Italian CDCDs were initially recruited for the study. First of all, a semi-structured questionnaire (see Supplementary Material) was administered to the Director of each Centre in order to evaluate its qualitative and quantitative characteristics, and variations in clinical activities after quarantine declaration. Due to the quarantine rules, all the out-patients visits were stopped. Therefore, the clinical staff members of each CDCD consecutively contacted by phone the family caregivers of patients registered in the waiting list. After an oral consent, a semi-structured, self-made interview gathering demographic and clinical data on the patient and the caregiver was administered (see Supplementary Material). Inclusion criteria were a diagnosis of one of the most common forms of dementias including: (1) Alzheimer’s Disease (AD), (2) Dementia with Lewy Bodies (DLB), (3) Frontotemporal Dementias (FTD), and (4) Vascular Dementia (VD). Exclusion criteria included current or previous diagnosis of other forms of dementias, mild cognitive impairment, and subjective cognitive complaints.

We collected all the socio-demographic characteristics of patients. The stage of dementia was assessed using the Clinical Dementia Rating (CDR) (Morris et al., 1997) scale. Furthermore, we investigated the patient’s variation in his/her clinical status, analyzing cognitive, psychological, behavioral, and motor symptoms during the quarantine period. Briefly, the following cognitive features were investigated: changes in memory, spatial and temporal orientation, language, attention, and perception. Questions about patient’s awareness of the pandemic were administered. In addition, we acquired data regarding variations in behavioral and psychological symptoms of dementia (BPSD, i.e., irritability, apathy, agitation, anxiety, depression, sleep change, aggressiveness, wandering, appetite change, hallucinations, and delusions), reporting for these symptoms both a worsening and/or a new onset that occurred during the quarantine period. Further questions about the need for modifying therapy because of BPSD changes were administered. Finally, we investigated the changes in patients’ motor activity, evaluating through a 5-point ordinal scale specific question if the subject walked better from the beginning of the quarantine period, remained stable, walked slower, became wheelchair, or became confined to bed.

Caregivers’ were asked about the cohabitation with the patient, composition of the family, and own work activity. Other questions concerned the impact of the quarantine period on caring for a person with dementia on their caregivers’ lives, investigating both social and psychological effects (changes in caregiver’s life, change in the relationship with the patient, concern for the pandemic, changes in therapeutic assistance, need to seek help from the emergency department, use of telemedicine, need to support during quarantine, own feeling of depression, anxiety, irritability, distress, overwhelm, and abandonment). Further details are available at Supplementary Material.

We performed statistical analysis using SPSS software, version 21, and the R statistical computing environment, version 3.6.2 (IBM Corp, 2012; R Core Team, 2013). Due to the low rate of missing data (<1%) no imputation was made. Firstly, we performed a descriptive analysis of all the demographic and clinical data. Then, we performed univariable and multivariable logistic regression of the dependent variables (outcomes in clinical symptoms, all described by binary variables) on the collected independent variables, using mixed effects logistic regression (as implemented in the lme4 R package) with the center as a random effect and all other regressors as fixed effects. Regressors with significant p (<0.05) in univariable logistic regression were included in multivariable regression. Bonferroni correction was applied to all the p-values of multivariable analysis, considering all outcomes together, as multivariable analysis is considered as our final result. We controlled for caregiver stress by including it in the multivariable analysis as a confounder, as a numerical fixed effect. The level of statistical significance was set at p < 0.05.

The study was initially approved by the Ethics Committee of the Coordinating Centre (University of Torino on April 7, 2020, n.00150/2020) and then by the local ethics boards. Caregivers gave first oral and then written fully informed consent to the study.

The response rate of CDCDs to the proposed questionnaires was 98%. Only two CDCDs were excluded because they did not recruit patients. The final 87 recruiting Centers were homogeneously distributed throughout the Italian territory (Figure 1).

Thirty-two percent of patients with dementia were recruited in Northern, 32% in Central, and 36% in Southern Italy. Thirty percent of CDCDs was based at university hospitals, 34% to general hospitals, and 36% to territorial based health services. In the pre-COVID-19 pandemic period, each Center followed a mean of 160 dementia patients per month. After the quarantine declaration, 85% of the CDCD suspended medical and psychological appointments, and visits were restricted to emergencies. Different forms of telemedicine (from phone calls to videoconferencing) were activated in 78% of Centers. One out of two Center provided on-line psychological support for caregivers. All the randomized clinical trials were stopped. Finally, 94% of support activities for dementia patients (Alzheimer’s café, Day Center, etc.) were closed.

The Survey response rates of the family caregivers ranged from 91.3 to 99.1%, according to different dementia Centers. We interviewed 5321 caregivers of patients regularly followed at different CDCDs. After data cleaning, the information collected from 4913 family caregivers (women 2934, mean age ± SD = 58.2 ± 12.0 years; men 1979, mean age ± SD = 60.7 ± 13.9 years) were analyzed. Fifty-nine % were cohabitants with the patients, and 36% were spouses of the dementia care recipient. Approximately half of the caregivers reported that quarantine induced a significant change in their lifestyle, with 30.3% complaining a reduction in time devoted to their own activities, and 15.5% reporting an increase in intrafamilial psychological conflicts. Caregivers reported a significant increase in anxiety (45.9%), depression (18.6%), irritability (26.2%), and distress (28.9%). Finally, 80.8% of the caregivers reported that telemedicine has been of help during quarantine.

From April 14 to April 27, 2020, we collected data regarding 4913 patients with dementia (2934 females, mean age ± SD = 78.9 ± 8.2 years; 1979 males, mean age ± SD = 77.2 ± 8.0 years) regularly followed at CDCDs. Demographic and clinical characteristics of dementia patients are shown in Table 1.

Family caregivers reported that, after a quarantine period of approximately 47 days, dementia patients showed a worsening in cognitive symptoms (+55.1%), and behavioral symptoms (+51.9%), the onset of new behavioral symptoms (+25.9%), and an increase in motor symptoms (+36.7%). According to caregivers, 40.1% of patients were totally aware of quarantine declaration, 36.0% were only partially aware, while 23.9% were totally unaware. Awareness of the quarantine varied significantly among the dementia subgroups, with DLB patients showing the highest degree (69.8%) of partial of total unawareness of pandemic.

Caregivers reported a worsening in cognitive symptoms in approximately 60% of dementia participants. About 37% of dementia participants showed worsening in two or more cognitive domains. Deficits more often reported as increased were: forgetfulness (68%), confusion (67.9%), and temporal disorientation (37%). Figure 2 displays the frequency distribution of cognitive symptoms worsened during quarantine.

When we examined diagnostic subgroups, we found a worsening of cognitive symptoms mainly in patients with DLB (59.6%) and AD (55.7%), followed by VD (54.2%), and FTD (48.3%) patients. Thirty-eight % of patients with AD showed a worsening in memory functions, 37.5% presented an increase in confusion, and 21.1% showed a worsening in temporal disorientation. The same effects were observed in patients with DLB, showing a worsening in memory functions in 26.1%, an increase in confusion in 40.3%, and in temporal disorientation in 20.3%. Patients with FTD showed mainly a worsening in language functions (25.5%). After multivariable analysis, the increase of cognitive symptoms in the overall group of patients with dementia was not associated with age, gender, duration of quarantine, disease severity, and changes in caregivers’ psychological features. Contrariwise, an inverse relationship between the aggravation of cognitive symptoms and the duration of the dementia was found (p < 0.001).

Interestingly, the pre-existing total or partial independence in motor function represents a risk factor for the worsening of cognitive functions (OR 1.85 [95% CI 1.42–2.39], p < 0.001; and OR 2.08 [95% CI 1.68–2.57], p < 0.001, respectively). Contrariwise, total awareness of quarantine was a protective factor against cognitive worsening (OR 0.68 [95% CI 0.57–0.81], p < 0.001) (Table 2).

During the quarantine period, BPSD worsened in approximately half of the patients with dementia. The most frequently worsened BPSD were irritability (40%), followed by apathy (35%), and agitation (31%) (Figure 3).

Examining subgroups, we found an increase in the number of symptoms mainly in patients with DLB (63.8%), followed by FTD (55.3%), AD (50.5%), and VD (50.3%). The increase of neuropsychiatric symptoms in patients with dementia was associated with moderate disease severity (CDR 2, p = 0.009), but not remained significant after applying Bonferroni correction. The worsening of BPSD was associated with pre-existing total or partial autonomy in motor function (totally independent OR 1.84 [95% CI 1.43–2.38], p < 0.001; partially dependent OR 1.88 [95% CI 1.53–2.31], p < 0.001). After multivariate analysis, the worsening of BPSD was not associated with age, gender, duration of quarantine, disease severity, duration of disease, and changes in caregivers’ psychological features. On the contrary, total awareness of the pandemic was inversely related to worsening of BPSD (OR 0.75 [95% CI 0.63–0.89], p = 0.03) (Table 3).

New BPSD occurred in approximately a quarter of the dementia patients. There was a different increase in the frequency of BPSD across groups. The most frequently reported new symptom was irritability (21.3%), followed by sleep change (21%), and agitation (18%) (Figure 2), which required therapy adjustments. The highest increase was observed in AD (26.7%), intermediate in VD and DLB (25.6 and 23.3%, respectively), and the lowest increase in FTD (21.9%). No suicide attempts were reported. The onset of new BPSD was associated with the total or partial physical independence (p < 0.001 and p = 0.01, respectively). Total awareness of the pandemic was inversely related to onset of new BPSD (OR 0.74, [95% CI 0.62–0.88], p = 0.03).

Increased motor dysfunction, mainly characterized by walking difficulties, was reported by caregivers in 36.7% of the patients. There was an increase mainly in patients with DLB (52.8%), followed by VD (42.3%), FTD (40.2%), and AD (33.2%). Age and disease severity were associated with an increase of motor dysfunction (p < 0.001) during quarantine, whereas full physical autonomy before quarantine was a positive predictor against worsening in motor functions (OR 0.49 [95% CI 0.37–0.64], p < 0.001).

The increased prevalence of cognitive, behavioral and motor symptoms was not influenced either by regional distribution of different CDCD or by the regional prevalence of COVID-19 at the time of the study.