Behavioural and psychological symptoms of dementia (BPSD) include affective symptoms, psychotic symptoms, non-aggressive agitation, irritability, wandering, elation and sleep problems. Reference Finkel, Costa e Silva, Cohen, Miller and Sartorius1 They have a high prevalence in dementia and nearly all people with dementia have at least one of these symptoms during the course of the disease. Reference Savva, Zaccai, Matthews, Davidson, McKeith and Brayne2 Such symptoms have negative effects on the quality of life of both patients and caregivers and are associated with increased costs of care. Reference Black and Almeida3,Reference Beeri, Werner, Davidson and Noy4 Better treatment and management of the symptoms are important, particularly as there is no effective treatment to alter the course of the underlying cognitive and functional decline. In order to design and conduct clinical trials for the treatment of BPSD, more information about the pattern of these symptoms in the different stages of dementia is needed to identify the best stage to intervene. In addition, insights into the extent to which BPSD occur over the course of dementia will help patients and care providers to plan for the future. Cross-sectional studies have shown that BPSD can occur at any time during the development of dementia. Their prevalence may increase from mild to severe dementia, whereas other studies suggest a non-linear course with the highest prevalence seen in the intermediate stages of disease. Reference Purandare, Allen and Burns5,Reference Lovheim, Sandman, Karlsson and Gustafson6 Symptoms may persist or be episodic over time, and this may differ between symptoms. Evidence from longitudinal studies is limited and has not been brought together systematically. Two reviews on the course of BPSD specifically in care-home residents have been published recently. Reference Wetzels, Zuidema, Jansen, Verhey and Koopmans7,Reference Selbaek, Engedal and Bergh8 They included a small number of studies (28 and 18) and concluded that the course of BPSD varied considerably between studies and between individual symptoms.
Our aim was to determine the longitudinal course of BPSD in individuals with dementia or in cohorts studied for dementia onset. We also investigated the persistence and incidence of symptoms and how persistence of BPSD over time relates to cognitive function. This review builds on five previous reviews by some of the same authors. Reference Van der Linde, Stephan, Savva, Dening and Brayne9–Reference Van der Linde, Stephan, Dening and Brayne13
Method
Studies were eligible for inclusion if they reported the persistence, incidence or association with cognitive function of one or more BPSD in older adults (i.e. majority of participants aged at least 65 years) with dementia or cognitive impairment, and measured symptoms at three or more time points. Observational studies or intervention studies where there was a control group were included. The symptoms included were apathy, depressive symptoms, anxiety, irritability or aggression, non-aggressive agitation, hallucination, delusion, misidentification, sleep problems, wandering and elation. No language restriction was applied. Studies with inadequate descriptions of the sampling of the population or measurement of BPSD were excluded. The review protocol was not registered.
Search method
Electronic searches of PubMed, EMBASE, Cinahl and PsycINFO databases were undertaken to identify potentially relevant articles published before March 2013. Search terms included text and MeSH terms for BPSD, dementia and longitudinal study (see online Fig. DS1). Two authors (R.v.d.L. and B.S.) independently searched titles and abstracts for potentially relevant articles. Following this, full text selection was completed by two authors: R.v.d.L. and A.M.P. (or B.S.). References of included studies were searched backwards and forwards, using the literature database Scopus.
Data synthesis
Data were extracted independently and in duplicate (R.v.d.L. and B.S. or E.E.). Details extracted from each paper included setting, participant recruitment method, number of participants, follow-up time, BPSD and their measurement, number of BPSD measurements, population age (mean and range), baseline Mini-Mental State Examination (MMSE) score, Reference Folstein, Folstein and McHugh14 baseline BPSD prevalence, statistical methods used, covariates taken into account and findings on the persistence, incidence and association of BPSD with cognitive function. Risk of bias was not formally assessed in a quality assessment. Findings were divided by dementia severity and BPSD. Dementia severity was defined using MMSE categories based on clinical practice guidelines from the National Institute for Health and Care Excellence (NICE): mild dementia (MMSE score 21–26), moderate dementia (MMSE 15–20), moderately severe dementia (MMSE 10–14) and severe dementia (MMSE <10). 15 When no MMSE score was reported, equivalent cut-off scores from the Cambridge Cognitive Examination (CAMCOG), modified MMSE, Clinical Dementia Rating (CDR) scale and the Alzheimer's Disease Assessment Scale (ADAS) were used. Reference Roth, Tym, Mountjoy, Huppert, Hendrie and Verma16–Reference Cano, Posner, Moline, Hurt, Swartz and Hsu22 By use of results from factor analyses and cluster analysis reported in the literature, Reference Van der Linde, Dening, Matthews and Brayne10 symptoms were grouped into affective symptoms (comprising depression, anxiety and apathy), psychosis (comprising delusions and hallucinations), hyperactivity (comprising irritability, agitation and wandering), elation and sleep problems.
Where possible the persistence of symptoms was reported as the percentage of people with a certain symptom at baseline who also had the symptom at the next measurement or for whom the symptom persisted during the entire follow-up period. Incidence was reported as the percentage of people without symptoms at baseline who had developed new symptoms at the next measurement or during the entire follow-up period. Results from a multistate model were reported when available. Studies investigating the association between BPSD and cognitive function were summarised by reporting the analysis methods (e.g. Cox proportional hazards model, latent class linear mixed model or logistic regression model), covariates taken into account, BPSD score and results, including hazard ratios, β coefficients and F or P values.
Baseline BPSD prevalence, persistence, incidence and association with cognitive impairment were compared for each of the symptoms in the studies that included several BPSD. Prevalence, persistence and incidence were summarised as ‘low’ if the majority of studies found that the results were lower than that of most of the other symptoms included, ‘high’ if the majority found that results were higher than for most of the other symptoms and ‘moderate’ if the results were intermediate or mixed. The range of the results was reported for each symptom.
Results
Owing to considerable heterogeneity in study objectives and methods, inclusion criteria were revised post hoc. The following exclusion criteria were added: a follow-up period of less than 3 months; reporting only the prevalence at different time points; symptom measurement through retrospective caregiver report (retrospective studies using medical records were included); and measuring pure major depression or clinical depression only. Studies reporting on minor depression or depressive symptoms only or depression as part of BPSD were included (as discussed in two previous publications). Reference Van der Linde, Stephan, Savva, Dening and Brayne9,Reference Van der Linde, Brayne and Dening11 From 5923 identified articles 48 were selected for inclusion after the abstract and full-text selection stages. Cross-referencing resulted in an additional 11 studies. In total 59 studies were included.
Study design
Characteristics of the studies included are shown in Table 1 and online Table DS1. The majority of studies recruited participants from psychiatric services including memory and dementia clinics (31 studies out of 59). Participants in these studies typically had moderate dementia (16 studies) and a younger mean age (in 20 studies participants had a mean age below 75 years). Other studies recruited participants from the population (8 studies), primary care (7 studies) or institutional care (8 studies), or recruited volunteers from other settings (4 studies). One study retrospectively reviewed medical records. The 8 studies that reported on care-home residents mostly included older participants (mean age 75 years or over) with moderately severe or severe dementia. Studies recruiting participants without dementia (10 studies) were found for depression only and most of these recruited from the general population (6 studies). Overall, most studies were from the USA or Canada (27 studies) and Europe (29 studies). Follow-up times ranged from 3 months to 14 years: 8 studies had a follow-up period of 1 year or less, 24 studies a follow-up period of 1–5 years and 26 had a follow-up period of 5 years or more.
Sample characteristics of included studiesa
Symptoms
Included symptoms are shown in Table 1 and their baseline prevalence is summarised in Fig. 1. Full details of each symptom, its definition, the instrument used for its measurement and the baseline prevalence can be found in online Table DS2. Affective symptoms were the most frequently studied (37 studies), with 24 studies reporting on depression only. Anxiety was studied in 11 studies, apathy in 4, and 2 studies reported on a factor of affective symptoms. Psychotic symptoms were studied in 26 studies (delusions in 20, hallucinations in 21, misidentifications in 1, psychosis symptoms combined in 5). Hyperactivity symptoms, including irritability (16 studies), non-aggressive agitation (often including pacing or wandering) (16 studies), wandering (4 studies) or a factor of hyperactivity symptoms (6 studies), were studied in 30 studies. Elation was measured in only 5 studies and sleep problems in 9. Many different instruments exist to measure and define BPSD, Reference Van der Linde, Stephan, Dening and Brayne12 and 28 different instruments were used across the included studies. The Neuropsychiatric Inventory (NPI) was used in 8 studies. Reference Cummings, Mega, Gray, Rosenberg-Thompson, Carusi and Gornbein23 Five studies used the total score of a BPSD instrument, rather than presenting individual symptom profiles.
Baseline prevalence of behavioural and psychological symptoms; see online Table DS2 for more details. Numbers are the reference numbers of the included studies. ‘Excluded’ indicates that the study excluded participants with a particular symptom at baseline (i.e. the prevalence was 0%). Twenty-six studies that did not report baseline prevalence or reported on a population already included in the figure are omitted. Dep, depression; Anx, anxiety; Apa, apathy; Del, delusions; Hal, hallucinations; Psy, psychosis; Irr, irritability; Agi, agitation; Wan, wandering; Ela, elation; Sle, sleep problems. *Subsymptom reported separately.
Prevalence
The baseline prevalence varied widely across the studies (see Fig. 1). Generally, higher baseline prevalence was reported by studies that included a population with moderate or moderately severe dementia than by studies that included those with severe dementia only. A higher prevalence of symptoms was also seen in studies that recruited participants from psychiatric settings rather than from the population or institutional care settings (e.g. for depression: psychiatric settings 20–57%, institutional care 8–20%, population 22%). Studies with a younger mean age typically showed a higher symptom prevalence (e.g. for delusion: <75 years 24–40%, 75+ years 9–22%). There may also be differences by BPSD instrument. For example, studies that measured symptoms using the BEHAVE-AD typically showed a higher prevalence than studies using the NPI.
Symptom persistence and remission
The persistence and stability of symptoms or the change in symptom scores over time were considered for each of the five symptom domains separately. Figure 2 summarises the results of studies investigating the persistence of depression, hallucination and irritability (see online Fig. DS2 for the persistence of all symptoms). Detailed findings are available in online Table DS3
Persistence of (a) depression, (b) irritability and (c) hallucinations.Squares indicate the reported percentage where the symptom persisted over the measurement period and the lines indicate 95% confidence intervals. The name of the first author is given next to the corresponding findings. If the study reported the persistence over several intervals, it is included in the figure more than once. Next to the name of the author the total follow-up time (in months unless specified) and the number of visits are reported. For example, Aalten et al measured symptoms at 5 visits over 24 months and reported on the percentage of participants with depression present at any consecutive period of 6 months (depression present at 2 visits), 12 months (present at 3 visits), 18 months (present at 4 visits) or 24 months (present at 5 visits).
Depression, anxiety and apathy
A large variation in persistence of affective symptoms was seen, with great intra-individual variability. Reference Fauth, Zarit, Femia, Hofer and Stephens24,Reference Neundorfer, McClendon, Smyth, Stuckey, Strauss and Patterson25 Aalten et al reported a relatively low persistence of depression, anxiety and apathy, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26 whereas Haupt et al reported a persistence of depression of up to 59% over a 2-year period. Reference Haupt, Kurz and Janner27 Anxiety and apathy may be more persistent over time than depressive symptoms, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28,Reference Berger, Bernhardt, Weimer, Peters, Kratzsch and Frolich29 although two studies reported that anxiety was less persistent than depression. Reference Haupt, Kurz and Janner27,Reference Hope, Keene, Fairburn, Jacoby and McShane30 Wetzels et al (study not included in the figures because it described only the persistence over each observation) found that resolution of anxiety was consistently higher than persistence of symptoms, whereas apathy showed a variable course. Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31 Where change was modelled statistically, affective symptoms were generally found to be stable without significant change over time. Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32–Reference Clare, Nelis, Martyr, Whitaker, Markova and Roth34
Delusions, hallucinations and misidentifications
The persistence of psychotic symptoms was mostly below 30% (5 studies), although one study reported that the 6-month persistence of delusions was 59% and hallucinations 52%. Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18 Further, multistate models by Eustace et al showed delusions were persistent over 12 months in 65%. Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28 Generally, hallucinations were less persistent than symptoms of delusion, Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28,Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35 although one study found similar results for delusions and hallucinations, Reference Haupt, Kurz and Janner27 and two studies found hallucinations were more persistent than delusions. Reference Hope, Keene, Fairburn, Jacoby and McShane30,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31
Irritability, agitation and wandering
Hyperactivity symptoms were mostly persistent, with one study showing that up to 76% of individuals had persistent symptoms of agitation over 2 years. Reference Haupt, Kurz and Janner27 Studies that investigated several hyperactivity symptoms found that agitation was more persistent than irritability. Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Haupt, Kurz and Janner27 A study investigating several symptoms of irritability found that verbal aggression was the most common and longest-lasting form of aggressive behaviour, whereas aggressive resistance and physical aggression were most likely to persist until death. Reference Keene, Hope, Fairburn, Jacoby, Gedling and Ware36 King-Kallimanis et al found that wandering status was more likely to change from wandering to non-wandering rather than the reverse and that wandering was a temporary phase for approximately half of care-home residents who were admitted as wanderers. Reference King-Kallimanis, Schonfeld, Molinari, Algase, Brown and Kearns37 Several authors analysed the course of hyperactivity over time using repeated measures analysis or a latent class linear mixed model. Garre-Olmo et al reported that over a 2-year period hyperactivity symptoms were mostly low and smoothly increasing (this pattern was found in two-thirds of participants). Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32 Cohen-Mansfield et al found that aggressive behaviours increased over time whereas physically non-aggressive behaviours did not change significantly. Reference Cohen-Mansfield and Werner38
Elation
The persistence of elation was investigated in only two studies. Wetzels et al found in severe dementia that, for each two consecutive assessments at 0–6 months, 6–12 months, 12–18 months and 18–24 months, symptoms were stable in 39%, 18%, 3% and 3% respectively. Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31 Over a total follow-up period of 2 years Aalten et al found in moderate dementia that symptoms were stable over a 6-month period in 2%, whereas for none of the participants did symptoms persist over 12 months, 18 months or 24 months. Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26 Therefore, these results suggest that elation is not persistent.
Sleep problems
Most studies that investigated sleep problems (4 studies) reported low persistence, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Berger, Bernhardt, Weimer, Peters, Kratzsch and Frolich29,Reference Hope, Keene, Fairburn, Jacoby and McShane30 or a fluctuating course. Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31 In only one study were sleep symptoms reported to be persistent. Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28
Incidence and absence of symptoms
Online Table DS4 and Fig. DS3 show the incidence of symptoms and the percentage of participants who did not have symptoms during the follow-up period. A summary is shown in Fig. 3.
Incidence of (a) depression, (b) irritability and (c) hallucinations. See Fig. 2 for an explanation of the symbols. NR, not reported.
Depression, anxiety and apathy
Affective symptoms commonly develop in people with dementia. Over a 1-year period a high or moderate depression incidence of up to 37% was reported by eight studies, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26-Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31,Reference Levy, Cummings, Fairbanks, Bravi, Calvani and Carta39–Reference Li, Meyer and Thornby42 whereas in others the onset of depression was low compared with other symptoms. Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18 The incidence of apathy has been reported to be particularly high: 64% over 2 years, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26 and 14–27% over a 6-month period. Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31
Delusions, hallucinations and misidentifications
In four studies the probability of new-onset hallucinations was reported to be low, Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Haupt, Kurz and Janner27,Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31 whereas in another four incidence was reported to be moderate or high. Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Ballard, Patel, Solis, Lowe and Wilcock41,Reference Scarmeas, Brandt, Blacker, Albert, Hadjigeorgiou and Dubois43,Reference Chang, Wang, Chen, Liu, Hong and Lin44 Other psychotic symptoms including delusions showed a consistently moderate incidence (11 studies). Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Haupt, Kurz and Janner27,Reference Levy, Cummings, Fairbanks, Bravi, Calvani and Carta39,Reference Scarmeas, Brandt, Albert, Devanand, Marder and Bell40,Reference Chang, Wang, Chen, Liu, Hong and Lin44–Reference Caligiuri, Peavy, Salmon, Galasko and Thal49
Irritability, agitation and wandering
All included studies that compared the incidence of hyperactivity with other symptoms (9 studies) concluded that the incidence of hyperactivity was high or moderate. Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26–Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31,Reference Levy, Cummings, Fairbanks, Bravi, Calvani and Carta39,Reference Scarmeas, Brandt, Albert, Devanand, Marder and Bell40,Reference McShane, Keene, Fairburn, Jacoby and Hope50,Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51 Although the incidence of agitation might be particularly high, Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Haupt, Kurz and Janner27,Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35 wandering might develop less often. Reference King-Kallimanis, Schonfeld, Molinari, Algase, Brown and Kearns37
Elation
The incidence of elation was investigated by three studies that used the NPI. Aalten et al reported a cumulative incidence over a 2-year period in 5%, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26 Wetzels et al reported that for each 6 months of observation new symptoms were seen in 3–4%, Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31 and Gillette-Guyonnet et al reported that new symptoms developed during a maximum follow-up of 4 years in 8%. Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51 These results suggest the incidence of elation is low.
Sleep problems
The probability of the onset of sleep problems was reported in four studies. No consistent findings were reported: at each 6-month period the incidence in one study was 15%, Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28 and in another 2–8%, Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31 whereas over a total follow-up of 2 years symptoms developed in 31%, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26 and over 4 years in 11%. Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51
Association with cognitive function
The results of studies investigating the association between the course of BPSD and cognitive function (25 studies) are summarised in online Table DS5.
BPSD and subsequent cognitive function
Eight studies investigated the association between depression and subsequent cognitive decline or development of dementia in those without dementia at baseline. Reference Kohler, van Boxtel, van Os, Thomas, O'Brien and Jolles52–Reference Wilson, Hoganson, Rajan, Barnes, Mendes de Leon and Evans59 Those with persistent depression showed significant decline over time in global cognitive function, memory, processing speed, recall and attention. Reference Kohler, van Boxtel, van Os, Thomas, O'Brien and Jolles52,Reference Dotson, Resnick and Zonderman53,Reference Bielak, Gerstorf, Kiely, Anstey and Luszcz58 Some found a slight increase in depression score before dementia diagnosis compared with those who did not develop dementia, Reference Amieva, Le Goff, Millet, Orgogozo, Peres and Barberger-Gateau55,Reference Wilson, Hoganson, Rajan, Barnes, Mendes de Leon and Evans59 whereas others did not find a significant change in depression before dementia diagnosis. Reference Wilson, Arnold, Beck, Bienias and Bennett56,Reference Becker, Chang, Lopez, Dew, Sweet and Barnes57 In those with dementia, associations with progression of cognitive function were found for psychosis, Reference Rosen and Zubenko45,Reference Haupt, Romero and Kurz60 hyperactivity, Reference Scarmeas, Brandt, Blacker, Albert, Hadjigeorgiou and Dubois43 and depression. Reference Zahodne, Devanand and Stern33 Two studies investigated the link between BPSD and mild cognitive impairment. In one study persistence of depression was associated with progression to dementia, Reference Houde, Bergman, Whitehead and Chertkow61 whereas another reported no difference in persistence between those who were cognitively stable and those who progressed to dementia. Reference Mackin, Insel, Aisen, Geda and Weiner62
Cognitive function and BPSD development
In individuals with dementia, psychosis, hyperactivity, agitation and physical aggression were associated with greater cognitive impairment. Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35,Reference Cohen-Mansfield and Werner38,Reference Asada, Kinoshita, Morikawa, Motonago and Kakuma63–Reference Blansi, Brubacher, Zehnder, Monsch, Berres and Spiegel65 In contrast, Marin et al found no association in dementia between cognitive impairment and depression, delusion, agitation and irritability. Reference Marin, Green, Schmeidler, Harvey, Lawlor and Ryan66 Four studies found that symptoms increased with cognitive decline in the early stages of dementia and were most commonly seen in moderate dementia, followed by a declining or stable course in the final stages of dementia. Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35,Reference Asada, Kinoshita, Morikawa, Motonago and Kakuma63,Reference Burgio, Park, Hardin and Sun64,Reference McCarty, Roth, Goode, Owen, Harrell and Donovan67 Cognitive function at onset of wandering was found to differ by type of wandering behaviour; for example, results suggested that excessive walking was more common in mild dementia, whereas in severe dementia getting lost was more likely. Reference Hope, Keene, McShane, Fairburn, Gedling and Jacoby68 No association was found between cognitive function and depressive symptoms in those with dementia, Reference Holtzer, Scarmeas, Wegesin, Albert, Brandt and Dubois69,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey70 whereas in those without dementia and without depression at baseline, cognitive impairment at baseline was associated with an increase of depressive symptoms over time. Reference Vinkers, Gussekloo, Stek, Westendorp and Van Der Mast54 In those aged 70 years and over cognitive function was found to be associated with initial scores for depression and anxiety, but not with symptom change over time. Reference Bunce, Batterham, Mackinnon and Christensen71 Baseline dementia diagnosis was not significantly associated with severity of depression at follow-up. Reference Janzing, Teunisse, Bouwens, Van 't Hof and Zitman72
Comparison of symptoms
We summarised the studies investigating several BPSD to compare baseline prevalence, stability, incidence and association with cognitive function for each of the symptoms (Table 2). Some symptoms were studied more often than others, and evidence is lacking for infrequently studied symptoms such as wandering (included in only one study investigating several BPSD), Reference Hope, Keene, Fairburn, Jacoby and McShane30 and apathy and elation (included in only four studies). Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31,Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51 Depressive symptoms were most often studied (included in 12 of the 13 studies investigating several BPSD). Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26–Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Levy, Cummings, Fairbanks, Bravi, Calvani and Carta39,Reference Scarmeas, Brandt, Albert, Devanand, Marder and Bell40,Reference McShane, Keene, Fairburn, Jacoby and Hope50,Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51 Compared with other symptoms, the results suggest that the persistence and incidence of depressive symptoms are moderate. Anxiety seems to be less prevalent and was reported to have a moderate persistence and incidence. Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26–Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31,Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51 The few studies investigating apathy suggest a high prevalence, persistence and incidence of symptoms. Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31,Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51 The prevalence, persistence and incidence of psychotic symptoms were suggested to be low to moderate, and may be particularly low for hallucinations. Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26–Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35,Reference Levy, Cummings, Fairbanks, Bravi, Calvani and Carta39,Reference Scarmeas, Brandt, Albert, Devanand, Marder and Bell40,Reference McShane, Keene, Fairburn, Jacoby and Hope50,Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51 Symptoms of hyperactivity were most frequently seen and the majority of studies reported a higher persistence and incidence compared with other symptoms. Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26–Reference Berger, Bernhardt, Weimer, Peters, Kratzsch and Frolich29,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31,Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35,Reference Levy, Cummings, Fairbanks, Bravi, Calvani and Carta39,Reference Scarmeas, Brandt, Albert, Devanand, Marder and Bell40,Reference McShane, Keene, Fairburn, Jacoby and Hope50,Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51
Results of 13 studies reporting at least two behavioural and psychotic symptoms of dementia
| Symptoms | Number of studies |
Baseline prevalence (%) 11 studies |
Persistence (%)
a
10 studies |
Incidence (%)
b
9 studies |
|---|---|---|---|---|
| Affective | 12 | High | Moderate | Moderate |
| Depression | 12 | High (8–57%) | Moderate (16–70) | Moderate (10–73) |
| Anxiety | 8 | High (17–52%) | Moderate (17–52) | Moderate (12–38) |
| Apathy | 4 | High (19–51) | High (20–55) | High (27–64) |
| Psychosis | 13 | Low | Moderate | Moderate |
| Delusions | 10 | Moderate (9–40) | Low (0–82) | Moderate (5–84) |
| Hallucinations | 11 | Low (0–18) | Low (0–52) | Low (4–45) |
| Hyperactivity | 12 | High | High | High |
| Irritability | 9 | High (6–57) | Moderate (12–80) | High (10–69) |
| Agitation | 7 | High (18–87) | Moderate (21–77) | High (19–80) |
| Wandering | 1 | NR | High (60) | NR |
| Elation | 4 | Low (3–9) | Low (2–39) | Low (4–5) |
| Sleep problems | 7 | Moderate (6–11) | Low (10–57) | Low (8–31) |
NR, not reported.
a. Percentage of symptoms persistent over 3 months or more.
b. Percentage incidence over 3 months or more.
Discussion
This systematic review confirms that BPSD are common and relatively persistent in individuals with dementia. The results suggest there are differences between symptoms: hyperactivity and apathy showed high persistence and incidence; depression and anxiety low or moderate persistence and moderate incidence; and psychotic symptoms low persistence and a moderate or low incidence. Studies of the association between BPSD and cognitive function suggest that in those without dementia the presence of depression is associated with subsequent cognitive decline. In those with dementia, psychosis, hyperactivity, agitation and physical aggression were associated with greater cognitive impairment.
Strengths and limitations
Standardised procedures were used for the literature search and data extraction, including double reading to ensure quality. However, no established search term for BPSD exists and therefore relevant studies may have been missed. The reference lists of included articles and reviews were searched to minimise the number of missed articles. The review included studies with a high degree of heterogeneity in study design and population characteristics, including large differences in the period over which the persistence and incidence was reported (1 month to 4 years), the total follow-up time (3 months to 14 years), the instrument and cut-off score used to measure symptoms, the number of symptoms measured, dementia severity, recruitment setting and mean age. This made cross-study comparisons difficult and a meta-analysis was not possible. We were not able to investigate whether the course of BPSD differs between types of dementia as only five of the 59 studies reported findings by dementia type. We adhered to most of the items of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (see online Table DS6). Reference Moher, Liberati, Tetzlaff and Altman73 Although we have reported on a range of factors that might influence the quality of the study, risk of bias was not formally assessed in a quality assessment. Our review therefore does not meet items 12, 15, 19 and 22 of the guidelines. Bias in the included studies may have led to an overestimate of persistence (e.g. participants remained under medical attention) or to an underestimate of persistence (e.g. gaps in the follow-up period or attrition through death or care-home admission). In addition, the review protocol was not registered.
Study differences
Many different instruments exist to measure BPSD, Reference Van der Linde, Stephan, Dening and Brayne74 and 28 different instruments were used by the studies included in this review. However, the increasing use of the NPI might improve comparability of future studies. Reference Van der Linde, Stephan, Dening and Brayne13 The NPI was used by eight studies. In these studies the baseline prevalence seemed lower than that reported by studies using other instruments (e.g. for irritability, NPI rates were 19–37%, Present Behavioural Examination (PBE) Reference Hope and Fairburn93 25–89% and BEHAVE-AD Reference Reisberb, Borenstein, Salob, Ferris, Franssen and Georgotas94 42–57%). The total score on the NPI significantly increased over time, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Clare, Nelis, Martyr, Whitaker, Markova and Roth34,Reference Tschanz, Corcoran, Schwartz, Treiber, Green and Norton75 and symptoms were mostly shown to be persistent, Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31 stable or increasing. Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Deudon, Maubourguet, Gervais, Leone, Brocker and Carcaillon76 The incidence reported by the studies using the NPI was low or moderate compared with studies using other instruments. Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31,Reference Gillette-Guyonnet, Andrieu, Nourhashemi, Gardette, Coley and Cantet51
Loss to follow-up is a challenge in longitudinal studies, and we have reported the number of participants at the end of the follow-up period for the included studies (online Table DS1). There was large variation in follow-up completion (24–100%, although often not reported) and reasons for leaving the study were often not reported. Persistence of BPSD may be associated with mortality and with refusal to participate in follow-up interviews, and differences in follow-up completion may have influenced the results. Furthermore, we have used study baseline as a proxy for disease and symptom onset and this may have affected the findings on the persistence of symptoms. Symptom course may be influenced by pharmacological or non-pharmacological interventions. Although there was large variation in medication use between study populations, in the majority of studies that included a sensitivity analysis the results were not altered when taking into account medication use. As we are not aware of a formal definition of high prevalence, persistence or incidence, we summarised the findings as ‘low’ if the majority of studies found that the results were lower than that of most of the other symptoms included, ‘high’ if the majority found that results were higher than for most of the other symptoms and ‘moderate’ if the results were intermediate or mixed.
Interpretation of findings
Prevalence
The prevalence of symptoms varied across the studies and between symptoms. Depression, apathy, irritability, agitation and wandering showed a high prevalence, whereas the prevalence of anxiety, hallucination and elation was low. Some studies consistently reported a relatively low prevalence of symptoms, Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26,Reference Wetzels, Zuidema, de Jonghe, Verhey and Koopmans31,Reference Marin, Green, Schmeidler, Harvey, Lawlor and Ryan66 whereas others consistently reported a relatively high prevalence. Reference Haupt, Kurz and Janner27,Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28,Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32 Differences might be due to variability in study design, population characteristics or measurement of symptoms. Indeed, a higher prevalence of symptoms was generally seen in studies that recruited participants with less severe dementia, in studies that recruited from psychiatric settings rather than from the population or institutional care settings, and in studies with a younger mean age. There may also be differences due to the BPSD instrument used.
Persistence
Large differences in persistence were seen across symptom groups and individual symptoms. Affective symptoms (including depression, anxiety and apathy) generally showed a moderate persistence, although a limited number of studies reported persistence of apathy to be high and in one study it was reported to be higher than for any other symptom. Reference Aalten, de Vugt, Jaspers, Jolles and Verhey26 Persistence of psychosis was low to moderate. In contrast, hyperactivity symptoms showed a high persistence. This is an issue of concern as these symptoms are among those most problematic for caregivers. Reference Matsumoto, Ikeda, Fukuhara, Shinagawa, Ishikawa and Mori77,Reference Nagaratnam, Lewis-Jones, Scott and Palazzi78 A low persistence was seen for elation and sleep problems. Differences in symptom persistence may reflect the nature of the symptom or might be explained by factors such as more widely available treatment options for depression and anxiety. Differences in dementia severity and baseline BPSD prevalence are likely to have affected results on persistence of symptoms. Persistence may be higher in those with more severe cognitive impairment at baseline, Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35,Reference Cohen-Mansfield and Werner38,Reference Wilkosz, Miyahara, Lopez, DeKosky and Sweet46 and a higher BPSD prevalence. Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Levy, Cummings, Fairbanks, Bravi, Calvani and Carta39 However, associations between study characteristics and results could not be tested because of the large degree of heterogeneity in study design and population characteristics.
Incidence
The results suggest that affective symptoms and hyperactivity symptoms commonly develop in people with dementia. Large differences in reported incidence were seen between studies. For example, the reported incidence of depression ranged from 12% over a mean follow-up period of 52 months, Reference Chang, Wang, Chen, Liu, Hong and Lin44 to 73% over a maximum follow-up period of 9.3 years. Reference Scarmeas, Brandt, Albert, Devanand, Marder and Bell40 Differences in study design and differences in baseline prevalence of symptoms are likely to have influenced the results. For example, the reported incidence might be higher in studies that reported a high baseline prevalence, Reference Haupt, Kurz and Janner27 compared with studies with a low baseline prevalence, Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Eustace, Coen, Walsh, Cunningham, Walsh and Coakley28 although no formal analysis of the association between study characteristics and incidence was possible.
Role of cognition
The presence of depression before the onset of dementia was associated with subsequent cognitive decline. Reference Kohler, van Boxtel, van Os, Thomas, O'Brien and Jolles52,Reference Dotson, Resnick and Zonderman53,Reference Bielak, Gerstorf, Kiely, Anstey and Luszcz58 In dementia, psychosis, hyperactivity, agitation and physical aggression were associated with greater cognitive impairment. Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35,Reference Cohen-Mansfield and Werner38,Reference Asada, Kinoshita, Morikawa, Motonago and Kakuma63–Reference Blansi, Brubacher, Zehnder, Monsch, Berres and Spiegel65,Reference Aalten, de Vugt, Jaspers, Jolles and Verhey70 Symptoms may be most common in moderate dementia, followed by a declining or stable course in the final stages of dementia. Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35,Reference Asada, Kinoshita, Morikawa, Motonago and Kakuma63,Reference Burgio, Park, Hardin and Sun64 However, heterogeneity in the pattern of findings across studies investigating the associations between BPSD and cognitive function prevented us from drawing more specific conclusions. The heterogeneity of results does, however, suggest that BPSD do not solely arise secondary to cognitive impairment.
Study implications
The results from this systematic review suggest that some symptoms such as hyperactivity are more persistent than others such as elation and sleep problems. In particular apathy, irritability, agitation and wandering showed a high persistence. These symptoms should be targeted in clinical trials to improve management and intervention. Clinical trials typically follow participants with more severe dementia over a short period. Reference Verkaik, van Weert and Francke79,Reference Lee, Gill, Freedman, Bronskill, Hillmer and Rochon80 However, results presented here show that symptoms may persist over long periods until death, Reference Devanand, Jacobs, Tang, Del Castillo-Castaneda, Sano and Marder18,Reference Hope, Keene, Fairburn, Jacoby and McShane30,Reference Li, Meyer and Thornby42 and may be most common in moderate dementia. Reference Holtzer, Tang, Devanand, Albert, Wegesin and Marder35,Reference Asada, Kinoshita, Morikawa, Motonago and Kakuma63,Reference Burgio, Park, Hardin and Sun64,Reference McCarty, Roth, Goode, Owen, Harrell and Donovan67 Clinical trials focusing on the earlier stages of dementia with a long follow-up time might therefore be particularly informative. Results could also inform patients and care providers about which symptoms are most likely to recur, so that measures can be put in place to reduce their impact. Recommendations for monitoring of patients and symptom management interventions are outlined in guidance by the Alzheimer's Society. 81
Future research
The heterogeneity in methods and results emphasises the importance of clearly reporting the study design, population characteristics and symptom definitions. Table 1 shows that studies typically included younger populations with moderate dementia, whereas studies recruiting those with mild or moderate dementia from the population or from primary care settings were lacking. As BPSD patterns may differ in these populations, they should be the focus of future studies. In addition, all included studies were conducted in high-income countries and the findings may therefore not be applicable outside these settings. Apathy was infrequently studied, and as the limited results suggest that it may have a high persistence and incidence, we recommend that this symptom should be the focus of future studies on symptom course.
These methodological issues reiterate the findings from several of our previous reviews. A review of reviews showed a focus on individual symptoms (particularly depression), raised the question how best to define and measure BPSD within and across populations, and recommended reporting more clearly the characteristics of the population, the inclusion and exclusion criteria and how BPSD were defined and measured. Reference Van der Linde, Stephan, Savva, Dening and Brayne9 Two reviews concluded that there were many instruments to measure BPSD, Reference Van der Linde, Stephan, Dening and Brayne12,Reference Van der Linde, Stephan, Dening and Brayne13 of which the NPI – a short, informant-based questionnaire measuring ten symptoms – has been cited most frequently and should form the core of any battery, although researchers choosing instruments should carefully address any gaps in its content with regard to their research question. In a guest editorial we discussed that the populations used in studies of depression and BPSD are often not quite comparable and that the results therefore cannot be readily extrapolated. Reference Van der Linde, Brayne and Dening11 Finally, we showed that studies investigating symptom groups show relatively consistent results, although there remains a large amount of individual variability. Reference Van der Linde, Dening, Matthews and Brayne10
Studying covariates that may be associated with higher persistence of BPSD, including impairment in activities of daily living, Reference Haupt, Kurz and Janner27,Reference Garre-Olmo, Lopez-Pousa, Vilalta-Franch, De Gracia Blanco and Vilarrasa32,Reference Burgio, Park, Hardin and Sun64 as well as medication use, Reference Cohen-Mansfield and Werner38,Reference Wilkosz, Miyahara, Lopez, DeKosky and Sweet46 could improve understanding of potential mechanisms involved in the presence and persistence of BPSD. Environmental factors such as overstimulation and a person's surroundings, as well as physical factors such as pain and dehydration, are recognised as important triggers for BPSD. Reference Purandare, Allen and Burns5,Reference Eriksson82–Reference Ballard, Gauthier, Cummings, Brodaty, Grossberg and Robert85 These factors are often difficult to capture and have not been investigated in the studies included here.
Clinical implications
Our findings underscore the existing evidence that BPSD are common in dementia and that they are also relatively persistent. Different symptoms have a variable course over time: for example, psychotic symptoms have relatively low persistence – that is, they may resolve during the course of the dementia. In contrast, apathy emerged as the only individual symptom with high baseline prevalence, high persistence and also a high incidence during the course of the dementia. Thus, increased interest in apathy as a possible early sign of dementia, as a marker for underlying brain changes and as a sign of progression of dementia seems entirely warranted. Reference Grool, Geerlings, Sigurdsson, Eiriksdottir, Jonsson and Garcia86 Although hyperactivity as a whole also had high baseline prevalence, high persistence and high incidence over time, the various symptoms subsumed under hyperactivity mean that it is not a unitary phenomenon. These findings are relevant to clinicians as they indicate which symptoms may be expected to persist or to occur anew, and therefore give a better understanding of the natural history of BPSD which, in turn, can influence approaches to management and treatment.
Funding
R.v.d.L. received a studentship from the National Institute for Health Research Collaborations for Leadership in Applied Health Research and Care for Cambridgeshire & Peterborough. A.M.P. was supported by the Medical Research Council (MRC RG56433 and MR/K021907/1).
Open access
eLetters
No eLetters have been published for this article.