The global impact of COVID-19 control measures on people with dementia living at home and their carers during the two first waves: A systematic review of quantitative and qualitative research across 27 countries

Data source and searches

We identified all peer-reviewed studies, irrespective of the country or study setting, published from March 11, 2020, to the last day of July 2022 through Medline, PsycINFO, EMBASE, Web of Science, CINAHL, Latin American and Caribbean Health Literature (Lilacs), Scientific Electronic Library online (Scielo), and EM Premium. The medical subject headings (MeSH) for Medline were “dementia”, “mental health”, “COVID-19”, and “care partners”. We also searched grey literature using Google Scholar (the first 10 pages), and an inventory of existing national surveys and reports in each participant’s country for items published between the start of the pandemic through the end of July 2022. We cross-checked the references list of all included studies.

Eligibility criteria based on the PICOS approach

P (Participants): people with dementia and/or their carers living at home/community dwelling during COVID-19 pandemic;

I (Interventions): not applicable;

C (Comparisons): not applicable;

O (Outcomes): disease progression, physical/cognitive/daily functioning, mental health, wellbeing, access to care/health services, carer burden or social outcomes in the context of COVID-19 control measures;

S (Study design): quantitative observational studies of any design and qualitative studies of lived experiences of dementia and/or their carers living at home.

Inclusion and exclusion criteria

We considered original peer-reviewed observational or qualitative studies in any language of all designs that: (1) included people with dementia and/or their carers living at home/community dwelling during COVID-19 and (2) assessed their physical/cognitive/daily functioning, mental health, wellbeing, access to care/health services, or other health, social and/or functional outcomes. In addition, the studies were required to have (3) quantified the magnitude of the association between COVID-19 control measures and the outcome of interest or reported on the implications of COVID-19 control measures on PWD and/or their carers living at home.

We excluded non-human studies, letters to editors, reviews, conference abstracts, articles without primary data, single case studies, studies that focused on therapeutic interventions, studies that included people living in long-term care facilities or nursing homes, and whose samples were composed largely of people with a diagnosis of mild cognitive impairment or other disorders of neurodegenerative nature. We also excluded studies that focused on investigating the effect of the COVID-19 infection itself on PWD and/or their carers.

Data collection and analysis

Selection process. We extracted keyword-based searches into an online reference management tool (EndNote^®, version 20, Clarivate, Philadelphia, PA, USA). One researcher (YC) eliminated duplicates and inserted data into collaborative online tool (Covidence^®, Melbourne, Australia). Two researchers per title (DS, KS, ET, MB, ZA, YC, TL, RF) independently assessed the study titles and abstracts for compliance with the inclusion criteria. Differences in opinion were resolved through group discussion. Then, each researcher assessed the selected full texts to determine their compliance with inclusion/exclusion criteria. When information regarding any of the above was unclear, we contacted authors of the reports to provide further details.

Data extraction. Two researchers per study (DS, KS, ET, MB, YC, TL, GP, RF, FRF) independently extracted the data into Excel files, consisting of (1) study information (authors, publication year, country, objectives, design, method of recruitment, sample size, measures used, and effects of COVID-19 control measures investigated in the study) and (2) participant information (age, sex, dementia type and severity if PWD, relationship to the participant if care partners). Outcomes were categorized into following themes: physical/daily functioning; cognitive symptoms; mental health; behaviour; wellbeing/loneliness; social cost; access to care; and other. We further collapsed the categories into themes based on conceptual unity emerging from data synthesis: (1) physical health and daily routine; (2) cognitive functioning; (3) behavioural problems, mental health and wellbeing; (4) social and economic consequences; and (5) access to health services.

Risk of bias and quality assessment

Four researchers (YC, IL, RF, TM) independently assessed study quality using the National Institutes of Health study quality assessment tools²⁶ that suited each included study design. The researchers involved in the assessment of risk of bias initially met for a calibration review, in which they independently reviewed one study of each type and discussed each item on the list to clarify its meaning and interpretation. Following this, the same three reviewers independently rated the methodological quality of each study across a set of items. In cases of disagreement between the researchers, a team discussion took place, with the aim to reach consensus.

Quantitative studies. We appraised quantitative studies in two steps. The first step included rating items related to potential sources of bias according to the most critical criteria for external and internal validity within cohort, cross-sectional, or case-control studies. The second step consisted of summarizing the presence of potential sources of biases as “yes,” “no”, “not reported,” or “cannot determine.” The overall rating of potential bias for each study was summarized in accordance with the Scottish Intercollegiate Guidelines Network methodology²⁷, where summarized study quality into three groups: high-quality (“++”), when all or most of the quality were fulfilled (i.e. allowing one “CD” or “NR” based six potential sources of bias); moderate-quality (“+”), when half items of criteria were fulfilled; or low-quality (“-”) when less than half of items of criteria were fulfilled (Extended data 2).

Qualitative studies. We evaluated qualitative studies using the Critical Appraisal Skills Programme (CASP) qualitative studies checklist²⁸. Assessments were categorized as follows: "yes" indicated that the item was adequately addressed in the study; "no" indicated that the item was not clearly addressed; and "cannot tell" reflected uncertainty. A score was given for each item that was categorized as “yes”. Study quality was then summarized into three groups: high-quality (“++”), when all or most of the criteria were fulfilled; moderate-quality (“+”), when half items of the criteria were fulfilled; or low-quality (“-”), when less than half of the criteria were fulfilled (Extended data 3).

Data synthesis

We tabulated the study characteristics into a file to evaluate similarities between studies. We observed heterogeneity across all PICOS criteria, confirming that the assumptions for conducting a classical meta-analysis were not met. Further, variations in outcome definition and categorization, and differences in the study samples and characteristics prevented us from performing data conversions. We therefore used a best-evidence synthesis approach, synthesizing findings from included studies through tabulation, data visualization, and qualitative description^29,30. We summarized the evidence and presented an overview of the findings across the studied outcome categories according to quantitative and qualitative data by global region.

Sensitivity analysis

We carried out sensitivity analysis to test whether critical methodological concerns affected the results. This involved repeating an analysis using combined studies regardless of study quality and separately and informally comparing the findings. We grouped the main results by study design (quantitative and qualitative) and compared the results of studies that focused on the same outcome theme.

Publication bias

Due to high heterogeneity of study design, population, COVID-19 waves, and outcomes (Extended data 4 Table 1), and variability among studies in the bias domains affecting study quality (Extended data 2 and 3), we did not perform evaluation of the publication bias using statistical tests³⁰. To minimize the potential for such bias, we implemented several strategies, including conducting literature searches with attempts to locate grey literature and unpublished studies and assessed reporting bias in outcomes of included studies³¹.

Dealing with missing data

We contacted study authors to verify key study characteristics, aiming to obtain missing data where possible for inclusion and analysis of select study. The remaining missing data were considered in the risk of bias assessment.

Ethical review

We did not seek ethical approval, as this study did not involve primary data collection.

Study characteristics

Table 1 (Extended data 4) summarizes the study characteristics pertinent to our research questions, including country/region of study origin, sample characteristics, analysis methods, and study results.

Study origin

Out of a total of 69 studies, 41 were conducted in Europe (France, Germany, Greece, Hungary, Italy, Netherlands, Norway, Poland, Spain, Turkey, UK), ten in Asia (China, India, Iran, Israel, Japan, Kuwait) and North America (Canada, Cuba, Jamaica, USA) each, five in South America (Argentina, Brazil, Chile) and one in Oceania (Australia). Two studies were cross-country studies, and included participants from multiple regions (Europe, Oceania, North America). No studies were conducted in the African continent (Extended data 4 Table 1).

Sample characteristics

The studies included a total of 209,738 participants. Fifteen studies engaged strictly PWD, 30 studies carers of PWD, and three PWD-carer dyads. The remaining 21 studies included a mix of PWD, MCI and SCD, carers and healthy older adults (Extended data 4 Table 1).

The studies included a total of 194,532 PWD. The study samples consisted of female participant samples ranging between 37% and 80%, with an average of 56% across studies. Seventeen studies did not explicitly state the sex ratio. The mean age of PWD ranged between 59.8 and 83.2 years of age. The studies included a total of 12,883 carers of PWD. The study samples consisted of female participant samples ranging from 48% to 100%, with an average of 74% across studies. Twenty-four studies did not explicitly state the sex ratio. The mean age of carers ranged between 45.7 and 70.1 years of age (Extended data 4 Table 1).

Risk of bias and quality of the evidence

In this review, the quality of the studies was not part of the inclusion or exclusion criteria, and none of the studies were excluded because of their quality.

We rated the majority of the qualitative studies as high-quality (“++”, 11 studies), and the remaining studies as moderate-quality (“+”, 7 studies) (Extended data 3). We rated the majority of the quantitative studies as moderate-quality (“+”, 36 studies), and the remaining studies as low-quality (“-”, 15 studies) (Extended data 2). Figure 2 positions a global region comparison based on identified themes for PWD and carers, and quality assessment, for qualitative (left) and quantitative (right) studies. None of the outcomes was investigated by all the papers from five regions contributing evidence included in this review (Asia, Europe, North America, South America, and Oceania). There were no qualitative studies originating from South America nor studies including participants coming from multiple regions. There were no quantitative studies originating solely from Oceania (Figure 2).

Figure 2. Number of included studies investigating outcomes among people with dementia (top) and their carers (bottom).

Outcomes were grouped into seven categories: (1) well-being/loneliness; (2) social cost/social participation; (3) physical/daily functioning; (4) mental health; (5) cognitive symptoms; (6) behavioural disorders; and (7) access to care/health services. Studies were grouped by quality (“-”, low-quality; “+”, moderate-quality; “++”, high-quality) and study methodology (qualitative, left; quantitative, right). The colors indicate the region where the study was conducted (yellow, Asia; green, Europe; teal, North America; dark blue, South America; purple, Oceania; grey, multiple regions).

Physical health and daily routine

In total, 39 studies reported on topics related to physical health and daily routine. These included 15 qualitative studies and 24 quantitative studies (14 cross-sectional studies, nine cohort studies, and one analysis of administrative data).

Physical health and daily routine of PWD. The qualitative studies reported that the continuing changes in control measures meant ongoing adaptation in everyday activities for PWD³². Inability to go outdoors or meet with others led to changes in diet and (physical) daily activities^33,34. Lack of a regular schedule was perceived as distressing and believed to increase symptom progression^35,36. Daily activities such as grocery shopping became more difficult and a source of stress, as people were worried about bringing the virus home^37–39, and these activities were more often taken over by the carer^32,35. Food packages that were handed out by councils in the UK did not take into account ethnical or dietary preferences (e.g. halal or vegetarian). In contrast, some people reported having more time to make fresh meals or a focus on nutrition to boost the immune system, resulting in positive improvements in their eating habits^34,39. PWD who were able to continue their daily routine seemed to manage well⁴⁰.

The results of qualitative findings were complemented by quantitative studies. A Chinese study found that physical activity levels were lower during the pandemic than before the pandemic⁴¹. A survey among PWD in Turkey found that 9.3% could not carry out their hobbies⁴², and a Brazilian study found that half of PWD ceased performing usual tasks or activities⁴³. Several studies reported a decline in daily functioning, either measured as functional dependency/limitations, personal care, or mobility^5,42–48. One Chinese study looked at the prevalence of malnutrition and found that only 3.3% of 787 PWD were malnourished, but half were at risk of being malnourished⁴⁹. In the same study, 7.8% stated loss of appetite and 4.1% lost weight, while 2.1% showed an increase in appetite and 6.0% gained weight⁴⁹. Studies were also consistent in that 37–61% of carers reported a decline in the health of the PWD^50,51.

Sleep problems were not mentioned in the qualitative studies, but were reported in quantitative studies. Studies in Asia, South America, Europe and Oceania all found that the prevalence of sleep problems was higher during or after lockdown than before^47,52–54. In addition, one Chinese study found that sleep quality, sleep efficiency, and sleep duration were poorer after lifting of the lockdown than during the lockdown, while sleep latency, sleep disorder, hypnotic usage, and daytime function remained stable⁵³.

Physical health and daily routine of carers. In the qualitative studies, carers reported having to adapt their own lives (e.g. quit job or work remotely) to take care of the PWD^34,35,55. Reduced work hours led to financial worries^34,35,55. A new role for carers was to ensure the PWD was adhering to the control measures^32,34, which was often perceived as challenging^32,56. With the increased care duties and reduced formal home care services, the carers reported stress around coping with day-to-day activities⁵⁷, even limiting their ability to look after themselves (e.g. stopping exercise)⁵⁶. Through its impact on the PWD, lack of a regular schedule also affected the health of the carers (e.g. chest pain)^32,58,59.

Quantitative studies confirmed the findings of qualitative studies in that the lockdown had a substantial impact on the daily lives of care partners^51,60,61. For 52% of Italian carers and 31% of Hungarian carers, the employment circumstances changed during the lockdown (with the most common change being working from home), which had financial consequences in a third of the cases⁵⁰. Due to the closure of services and control measures, new care needs arose⁵¹. The time spent caring increased for most carers^47,50 and there were proportionally more carers spending >10 hours caring per day⁶². The caregiving required more physical effort, and carers felt more tired⁴⁷, consequently increasing the burden of care^{5,45,46,53,60,63,64}. Nearly a third of 4913 Italian carers and 12% of 165 German carers reported a reduction in time available for their own activities^60,61.

Physical health and daily routine: country differences. While studies conducted in different countries measured different outcomes, overall, the tendency of decline in physical health and changes in daily routine and increased care needs were consistent across countries and regions. No notable differences between countries were found across the studies, apart from some reflections in qualitative studies on services that were specific in a given country (e.g. food packages handed out in the UK).

Cognitive functioning

In total, 28 studies reported on topics related to cognitive functioning. These included 9 qualitative studies and 19 quantitative studies (12 cross-sectional studies and 7 prospective cohort studies).

Cognitive functioning of PWD. In qualitative interviews, care partners expressed that they felt the cognitive functioning of PWD declined more rapidly during the lockdown than before^{36,40,55–57,65}. Most studies attributed this accelerated decline to a reduction in cognitive, mental, or social activation^36,57,65, confusion due to not understanding the control measures^35,55,66, and lack of routine⁵⁵. In contrast, in one study, carers argued that the accelerated decline could be explained by the progressive nature of dementia and/or their increased proximity to the PWD, such that they noticed the decline more⁴⁰.

Findings from quantitative studies used a variety of study designs to examine (rate of) decline in cognition during lockdown. Most studies were cross-sectional surveys asking PWD or carers about subjective (decline in) cognitive functioning of the PWD^{5,42–45,47,61,67}. They reported subjective worsening of cognition in 35–73% of PWD. Two studies that measured cognitive functioning before and during lockdown, found a similar magnitude of decline in Mini Mental State Examination (MMSE) scores with a mean decline of 1.61 ±3.4 in 105 Chinese people with Alzheimer’s disease (AD)⁴¹ and 1.99 ±0.42 in 38 Italian people with AD⁶⁸. Only two studies measured the rate of decline. In 36 PWD from Kuwait, the mean monthly rate of decline was 0.53 ±0.3 points during lockdown, which was significantly higher than before lockdown (0.2 ±0.1, p<0.001)⁶⁹. In 40 PWD from Italy, the within-group annual rate of decline was greater from 2019–2020 than from 2017–2019 (p<0.05)⁷⁰. In summary, despite differences in study design, findings consistently point toward a decline in cognitive functioning during lockdown. There is sparse evidence suggesting that the rate of decline was more rapid during than before lockdown.

Cognitive functioning: country differences. The qualitative studies were all conducted in Europe and North America, with consistent findings across these two regions. Quantitative studies were conducted in all regions except Oceania and Africa. We observed consistency in findings across study designs and countries.

Behavioural problems, mental health and wellbeing

In total, 66 studies reported on topics related to behavioural problems, mental health and wellbeing. These included 20 qualitative studies and 46 quantitative studies (28 cross-sectional studies and 18 prospective studies).

Behavioural problems, mental health and wellbeing of PWD. In qualitative interviews, both PWD and carers indicated that the control measures negatively affected the mental health and wellbeing of PWD and led to increased behavioural problems^{33,36,65,66,71}. Behavioural problems included increased apathy, anxiety, restlessness, irritability, sleep disturbances, aggression, and agitation^{33,35,38–40,65}. This was mainly attributed to reduced social contacts, closure of day care services, others wearing masks causing confusion, and worries about their loved one’s health and wellbeing^{33,35,36,38,39,55,65}. Social isolation and lack of meaningful activities were often cited as the main causes for increased feelings of loneliness^36,58,65 and the inability to understand the need for control measures further exacerbated these problems^38,39,65. The impact on mental health was even stronger in PWD who were aware of their declining independence⁴⁰. Coping strategies such as using alternate communication methods to maintain social contacts (e.g. video calls), were often not available or feasible⁶⁵.

In contrast, some studies reported positive consequences of the pandemic, including time to return to past hobbies, learn new skills, and a welcome break from the busy and therefore stressful normal routine^36,39. Practical support from family in adapting to the situation provided a sense of security and helped maintain wellbeing^34,40.

Quantitative studies showed higher prevalence of neuropsychiatric symptoms during the pandemic than before the pandemic^{41,44,52,61,67,72–76} and lower prevalence after lifting of quarantine^53,73. Studies that compared prevalence of symptoms later on in the pandemic with pre- or early stages of the pandemic found a decrease in prevalence as the pandemic continued^6,77, although changes were not always statistically significant⁷⁸. In the early stages of the pandemic, 44–65% of PWD reported increases in symptoms, which was fairly consistent across countries given variations in study design^{5,41,44,61,64,67,79}. The most commonly reported symptoms included anxiety, depression, apathy, agitation, irritability and sleep problems/night time behaviour^{5,6,41–44,46,47,49,53,54,63,67,72,75,76,80}. During lockdown, a higher percentage of PWD were using benzodiazepines, neuroleptics, or combinations⁷³, and doses also increased⁶⁴. Factors associated with symptom prevalence included not living with a carer, awareness of COVID-19, limited COVID-19 understanding, and inability to access support services^54,61,80.

PWD seemed to experience less worry or stress related to the COVID-19 outbreak itself than healthy controls⁸¹. In 787 Chinese PWD, only 8.6% worried about the outbreak. However, among those who did worry, the prevalence of anxiety and nervousness was high (83.8%)⁴⁹. In a sample of 38 Italian PWD, stress levels increased from the start of the lockdown to 2 weeks later, then stabilized another 2 weeks later. (p<0.05)⁶⁸.

During the lockdown, social contact decreased relative to before the lockdown^41,67. Studies that reported on loneliness all showed increases in loneliness during lockdown relative to before lockdown, but with varying prevalence. In an Iranian study of 21 PWD, 81% had moderate and 14% had severe levels of loneliness⁸². In a Dutch sample of 204 PWD or MCI, 17% reported (more) loneliness⁶. In a sample of 321 carers from Argentina, Brazil, and Chile, the carers reported that 31% of PWD felt more lonely⁴⁷.

Two UK studies and one Spanish study found no significant changes in quality of life from pre-lockdown to during lockdown (measured as self-reported and/or proxy-reported Dementia Quality of Life (DEMQOL) or EuroQol Five-Dimensions (EQ-5D))^75,83,84. In contrast, another UK study found that mental wellbeing (measured with the Short Warwick-Edinburgh Mental Wellbeing Scale (SWEMWBS)) gradually improved as the lockdown progressed⁷⁸, and an Italian study found that the quality of life (measured with the Quality of Life in Late-stage Dementia (QUALID) scale) first declined during lockdown and then recovered after lockdown to pre-lockdown levels⁷³.

Behavioural problems, mental health and wellbeing of carers. Stress and concerns were common themes across qualitative studies, for which a number of causes and consequences were cited. Care partners were concerned about themselves or the PWD being infected with COVID-19 and being isolated, hospitalized, or even losing their life, and being unable to provide the required care^{33–35,39,56–58,65,66}. They were concerned about the future, specifically related to developments of the pandemic^32,35,55,65 and increased care burden^55,58. These concerns were exacerbated by lack of information from governments or health services^55,56. Carers experienced an increased care burden as support from family or professional care dropped away^{56,57,65,66,85}. Carers experienced extra stress if they: (a) had to take over tasks previously done by care professionals that they did not feel qualified for³⁷, (b) felt forced to move in with the PWD³⁷, or (c) feared or experienced unemployment or economic difficulties^57,66. Spending more time together also meant that carers were more aware of the symptoms of the PWD, which increased their concerns about disease progression³⁵. Adhering to the control measures, (repeatedly) explaining these to the PWD, and adjusting to the routine of the PWD added to the care burden and subsequently caused stress and fatigue^{32,38–40,58,66}. The pandemic situation also confronted carers with the need for care-planning conversations with the PWD, family, and medical team. Making such decisions in a time of heightened risk and reduced availability of services exacerbated their stress³⁹. Other sources of stress were the lack of social interactions and not having an outlet or time for hobbies^{32,37,39,65,66,71,85}.

For some carers, the stress led to panic attacks, anxiety, depression, fatigue or feelings of loneliness^{32,37,39,65,71}. The stress also negatively affected the care relationship, triggering behavioural problems in the PWD^35,65,66. This caused a vicious cycle, with more stressed carers leading to more behavioural problems in the PWD leading to more stressed carers, etcetera³⁸. Carers also expressed they felt unable to cope on their own^57,85. In contrast, two studies reported that spending more time with the PWD helped strengthen the care relationship^35,59.

Some studies discussed coping strategies. Spirituality, meaningful activities (e.g. gardening, exercise), enjoying good weather, and self-care provided relief and sense of purpose^34,65. Maintaining routine/daily structure, creatively adapting to the situation, and hiring paid care helped to cope^34,56,58,65. Also, other people’s willingness to help and (online) peer support networks aided in reducing negative emotions^56,58. Carers tried to assist the PWD with alternate forms of communication, mainly digital tools, to compensate for the lack of in-person social connections, but this was not satisfactory to the same degree^39,40,65,66 and also not accessible to all^34,85.

The quantitative studies confirmed the findings from qualitative studies in that the control measures negatively affected the care partners’ mental health. Studies reported increases in depressive symptoms, anxiety, irritability, and stress^{44,47,48,54,60,61,76,79,86,87}, as well as concerns related to the pandemic, health of the PWD and their own health^46,51,60. During the pandemic, the mean scores for depression, but not anxiety, were significantly higher in carers than in non-carers⁸⁷. A number of studies examined the prevalence of mental health problems as the pandemic progressed and reported mixed findings. During the first three months of the first lockdown, levels of anxiety and depression remained constant in 377 UK carers⁷⁸. After lifting of the quarantine, one study reported increases in anxiety and depression and a decrease in distress in Chinese carers⁵³, while another study reported decreases in anxiety, depression, and stress in Italian carers⁷³. Another Italian study reported that the prevalence of a range of mental health problems increased from 6–39% at the start of the first lockdown to 32–60% one year later⁸⁷. Many factors may explain these contrasting findings, including (but not limited to) differences in timing of measurements, measurement instruments, sample characteristics (e.g. differences in disease progression of the PWD), as well as actual country-differences. Factors associated with mental health problems in carers included disease progression, neuropsychiatric symptoms and health of the PWD, perceived care burden, informal support network, paid home care, day care, carer-PWD relationship, pandemic-related concerns, social isolation, and hobbies^{6,48,49,51,54,67,74,76,79,87–89}.

Several studies reported either an increase in levels of loneliness or high levels of loneliness during the pandemic^8,47,60,82. Factors associated with higher levels of loneliness included anxiety, having more formal activities during the day, relationship to the PWD, co-residing with the PWD, health of the PWD, cognitive functioning of the PWD and caregiving burden^8,90. At the same time, loneliness contributed to increased perceived caregiving burden⁸.

A series of UK studies found that approximately half of carers reported high levels of wellbeing^62,91, with no or modest decline in quality of life (measured with the Dementia Quality of Life of Carers (DEMQOL-C) and self-rated quality of life) during the pandemic^83,84,91. In contrast, a survey among 84 Italian carers found that, on average, participants scored very low on physical and mental wellbeing (measured with the 12-item Short Form Survey SF-12)⁴⁸. Factors associated with wellbeing included resilience, self-efficacy, coping strategies, employment status, type of dementia, dementia severity, and type of dwelling^84,92.

As stated earlier, many studies reported an increase in caregiver burden, which plays a key role in the mental health and wellbeing of carers. The negative impact of the care burden may be mitigated by coping strategies. In a UK study, approximately 70% carers were coping well ‘often’ or ‘always’. However, during the pandemic, more carers shifted from ‘sometimes’ to ‘often’ or from ‘often’ to ‘always’ than in a pre-pandemic control group, suggesting that carers used coping strategies to adjust to the new carer demands during the pandemic⁶². The approximately 30% who reported coping ‘sometimes’ or ‘never’, is in the same range as reported in other studies. In a Chinese sample, 34.5% of carers needed a break and 29.1% could no longer cope⁴⁹ and in a Dutch sample, 29% reported a need for more support⁶⁷. Finally, in a Turkish sample of 54 carers, 50.0% felt they could not find time for themselves, 14.8% wished to leave the patient and go away, 38.9% said they were irritable. Of note, an Italian study concluded that participants mostly used maladaptative coping strategies, such as avoidance strategies, but these strategies did not affect the stress level of caregivers⁴⁸.

Behavioural problems, mental health and wellbeing: country differences. While individual studies from different countries highlighted different elements, the collection of findings tells a coherent story that illustrates a uniform impact of the control measures on the mental health and wellbeing of both the PWD and their carers. These findings seem consistent across countries and regions. The long-term impact of the control measures on the mental health of the care partners remains unclear.

Social and economic consequences

In total, 28 studies reported on topics related to social and economic consequences of the control measures. These included 17 qualitative studies and 11 quantitative studies (seven cross-sectional and four cohort studies).

Findings from qualitative and quantitative studies consistently indicated that decreased social life and support was a key issue for both PWD and their carers^{33–35,38,54,64,66}. As described above, this had a negative impact on their mental health and wellbeing. For care partners, lack of informal support meant that they had to take on more caregiving tasks, particularly if paid care had also dropped away.

The pandemic had financial consequences for many people, but there were some consequences that were specific for PWD and their carers, namely: (1) loss of income due to the need to reduce work hours or even quit their job to care for the PWD^34,51,66. Carers, particularly those whose job situation did not allow for remote work, had to choose between care duties and income³⁴. And (2) increased costs due to having to hire paid care^34,66 and purchasing hygiene products to reduce the risk of infection (e.g. masks, sanitizer)³² or other products they were already using (e.g. wipes, medication) that went up in price due to scarcity^32,33. These financial consequences particularly affected people with low incomes, therefore increasing already existing inequalities^34,55.

Social and economic consequences: country differences. While no notable regional differences could be observed, financial consequences came up more frequently in studies conducted in low- and middle-income countries such as India and Jamaica. A few studies pointed towards differences in people in lower or higher socio-economic positions. The financial consequences were felt more by people with low incomes, while people in higher socio-economic positions had more access to social and other resources to cope with the situation^34,55.

Access to health services

In total, 39 studies reported on topics related to social and economic consequences of the control measures. These included 16 qualitative studies and 23 quantitative studies (17 cross-sectional studies, 4 prospective cohort studies, and 2 studies analyzing administrative data).

Qualitative studies reported that PWD and their carers were greatly affected by the reduced formal health and social care services, such as closure of day care centres, closure of respite care, cessation of physio- and occupational therapy, decrease in domestic and/or home care and peer support groups^{32,35,37,39,56,58,65,66}. As the entire health care system was focused on controlling the pandemic, even access to basic health care services was limited. For example, it was difficult to get doctors’ appointments or access to medication^{33,34,39,58,59,66,93}. Offered solutions such as telemedicine were not always accessible or feasible for PWD^{35,55–59,71}. As already alluded to above, this led to (1) disruption of routine and less activation of the PWD, possibly leading to accelerated disease progression and (2) increased care burden, with no prospect of a break in care duties.

When home or day care services were available, care partners sometimes chose to stop the care due to difficulties with ensuring care professionals were adhering to the control measures or due to fear of bringing the virus home^37,39,58,66. A few studies also noted that lack of information on available support was an issue^56,85, with respondents in one study suggesting that word of mouth was crucial in obtaining relevant information³⁵.

Quantitative studies confirmed the findings in qualitative studies, suggesting care use was lower during the pandemic and therapies stopped^{9,46,60,63,64,72,80,94}. The only exception is a USA study of 64 PWD, which reported no significant difference in the number of days receiving formal care during versus pre-pandemic⁹⁵. The proportion of respondents that reported discontinuation of care varied greatly, ranging from 10% in French carers⁷⁴ to 49% in UK carers⁹¹, 75% in Dutch PWD⁶⁷ and 90% in Argentinian carers⁶⁴. However, these differences are likely explained by variations in how the questions were asked. In time-series analyses of administrative care use data in Canada, the home care visits dropped by 16% at the start of the pandemic (mid-March 2020), but recovered to normal levels by September 2020⁹⁴. Similarly, use of therapies dropped by 50% at the start of the pandemic and exceeded pre-pandemic levels by September 2020⁹⁴. These findings were partly confirmed in a second Canadian study using administrative health services data, which also found a drop in visits per 100 PWD (2.41 in reference period versus 1.20 in pandemic period), but in that study, the number of visits did not return to pre-pandemic levels after the second wave⁹⁶.

Two quantitative studies reported contrasting socio-economic differences in access to care. In 73 Israeli carers, those with higher levels of education or higher income were less likely to report forgone care⁹⁷, while in a time-series analysis of administrative data of 131,466 Canadian PWD, the decline in personal care was greater in those living in more affluent areas⁹⁴. Hence the potential modifying effect of socio-economic factors on impact of the control measures on health services use remains unclear, or could potentially reflect differences across countries.

Access to health services: country differences. While there is consistent evidence that the control measures had a significant impact on access to health and care services for PWD in all countries (except the US), the magnitude of this impact is likely to differ across countries given the differences in availability of health services for PWD, as well as cultural differences in seeking formal support. However, direct or indirect country comparisons are hampered by the variation in study designs and measurement of services use.

Principal findings

Pre-pandemic studies have reported that PWD often experience social isolation and loneliness, physical and mental adversities, challenges with relationship and intimacy, and reduced quality of life^34,35. Therefore, we anticipated exacerbation of adverse outcomes related to the control measures during COVID-19 pandemic. Adverse outcomes were consistently reported in the scientific evidence included in this review, spanning many countries and regions. The impact of the pandemic on carers of PWD was also largely uniform, who often reported living with anxiety and fear, and experiencing difficulties balancing caregiving challenges with their own needs.

The experiences of PWD and carers that emerged in the qualitative research were of diverse and personal nature. Some PWD were able to continue regular daily activities at home, such as gardening, doing crosswords, or reading, and seemed to manage the control measures well, especially were not excessively complex. Some carers and PWD reported that their diets had improved or become healthier during the pandemic as they were able to make fresh meals. On the other hand, carers supporting a person with FTD than with AD reported greater burden in the included studies, likely because people with FTD are known to experience more behavioral and psychological symptoms⁶¹ and reduced access to formal care support⁷⁴. In some cases, there was a change in the person delivering the care, from a spouse to an adult child⁶⁸ or a lack of family support to the main carer⁶⁴. Synthesis of qualitative research raise important discussions about variability in the impact of control measures and the role of a person’s living environment and social influences (i.e., PROGRESS- Plus parameters)⁹⁸ in their experiences.

The majority of the studies included in this review came from high income countries. Among included studies, financial consequences came up more frequently in studies conducted in low- and middle-income countries, such as India and Jamaica^33,34. A few studies pointed towards differences in people in lower or higher socio-economic positions^32,94. It is probable that those with lower income were more affected due to increased likelihood of holding jobs that put them in a vulnerable position (unable to work from home; higher risk of losing job; front-line jobs with higher risk of infection), indicating that social capital and socioeconomic status play an important role on brain health. Furthermore, regarding the influence of race and ethnicity, the research conducted by West et al included black, Asian, and minority ethnic groups in the UK³⁹. Findings were similar to those in other studies, suggesting little ethnic or cultural differences, or a failure of studies to capture this in their analyses. Research sensitive to the person’s living environment and social capital can enhance our understanding of the needs of PWD and their carers within and between countries, and has the potential to foster complex discussions with research, policy, and practice implications. This is particularly relevant for countries in Africa, since no studies included in this review originated from this region.

Strengths and limitations

Our review has several strengths. First, we did not limit our searches to English language studies only, rather, we included studies published in multiple languages. This approach allowed for extraction of data collected in non-English-speaking countries with limited resources to disseminate their work with the larger scientific community. Second, we included both quantitative and qualitative research. Our approach took into account the advantages of the two types of research. We collected topics from open-ended questions and qualitative research to better understand the lived experience and the societal impacts of the pandemic, while also synthesizing results from quantitative studies, which used standardized measures to study multiple associations relevant to our research objectives. To the best of our knowledge, this is the first comprehensive systematic review on this topic aggregating data from both qualitative and quantitative studies. This is also the first systematic review that reviewed evidence through a global lens. Finally, our multiple subgroup analyses with a focus on relevant endpoint and visual data presentation allowed for nuanced comparisons across studies, countries, and global regions.

We acknowledge several limitations. First, the heterogeneity in included studies prevented us from conducting a formal assessment of publication bias. Heterogeneity was observed at multiple levels (i.e., sample characteristics, clinical settings, spectrum of studied comorbidities, reporting, statistical approaches, and outcome measurements); even within the same country and group of people the strength of association varied. Second, prior to data extraction, we decided on the confounding variables that were considered as important to studied outcomes. However, many of these variables were missing. The type of dementia and severity were not always reported in included studies, despite being important variables that could influence the outcomes for PWD and their carers. A concern inherent to all dementia research is that cognitive impairments of PWD may affect their ability to understand survey questions, and carers who provide care to PWD may have influenced responses of PWD in studies involving dyads, consequently impacting the data collected. Access to and proficiency with technology required to complete self-report surveys during the pandemic may have led to biased samples and limited generalizability to people who are living with more advanced stages of dementia, those facing communication barriers, and those with limited access to or proficiency with technology⁹⁹.

The quantitative studies included in this systematic review reported associations while considering various confounding parameters. It is possible that the associations between confounding parameters and the outcomes of interest can be attenuated due to “over-controlled” variables (Extended data 4 Table 1). For instance, since there is a possible causal relationship between age, gender, insomnia, dementia, and studied outcomes, including them in a model might have attenuated the association between them and the studied outcomes because they might be in the causal pathway to many adverse effects^100,101. However, we observed consistency in findings, which strengthens the robustness of results and mitigates this concern to some extent. Finally, we included studies up to July 2022; thus, the results are limited to the earlier stages of the pandemic (wave 1 and 2). The perspectives of PWD and carers may have shifted over time as the pandemic progressed and control measures changed. However, the control measures were most strict in the earlier stages of the pandemic, suggesting that our review captured the major consequences of such measures.

Implications of the results for practice, policy, and future research

Results from our systematic review have highlighted challenges with access to health services, (exacerbated) deterioration of physical, cognitive and mental function, as well as social and financial adversities faced by PWD and their carers living at home during the COVID-19 pandemic. It is essential that, globally, decision-makers understand the needs of PWD when implementing control measures. People with dementia and their carers represent a heterogeneous group of people across countries and communities; despite that, we did not identify many contradictory results in the included literature. Social equity parameters should be considered and implemented in future studies to identify variables within populations of interest that could be associated with a poorer overall experience during the pandemic, and lead to studies of higher quality to enhance certainty in the results.

The impact of COVID-19 should be assessed separately according to various living environments to identify more at-risk groups and regions, particularly those without evidence included in this review (i.e., Africa). Future studies should also analyze different protective and risk factors among PWD and their carers. For example, it would be important to compare the effect of living alone versus living with carer, of being younger in age versus older, or of having greater community support versus less. Moreover, future studies can delve deeper into resilience, regulation of emotions, and coping strategies of PWD and their carers. Micro-, meso-, and macro-level policies should be established to ensure that PWD and their carers are trained on how to use communication technologies to enable them to preserve social links, uphold family bonds, and maintain the ability to give or receive needed care when faced with isolation and protective measures during future pandemic or crises.