Non-cognitive symptoms as early indicators of Dementia: a preliminary retrospective study from a Psychiatric outpatient clinic

Abstract

Introduction

Neuropsychiatric symptoms (NPS) may signal dementia risk or early pathology, yet clinical evidence regarding the association between pre-diagnostic NPS and outcomes remains limited. This preliminary study explored NPS features and their potential associations with dementia progression.

Methods

The study was conducted on 201 dementia patients from Shanghai Mental Health Center (ICD-10 diagnosed). NPS history was extracted from medical records. Statistical analyses included chi-square tests (demographics/subtypes), ANOVA (preclinical intervals), and logistic regression (AD vs. non-AD factors).

Results

69.2% of patients had pre-dementia NPS, which was associated with a shorter interval to dementia (23.4 vs. 42.2 months, p < 0.001) and a higher likelihood of non-AD dementia risk. Psychiatric disorders (74.1%) were the most prevalent NPS domain, followed by sleep disturbances (46.8%) and affective dysregulation (39.8%), with psychiatric symptoms showing the shortest dementia transition (34.3 ± 37.8 months). Females exhibited higher hallucination/delusion rates than males (30.2% vs. 15.4%; 26.2% vs. 9.6%). AD predominated (68.7%), followed by DLB (12.4%) and others. Psychiatric-to-dementia intervals differed significantly between AD (52.3 ± 47.6months) and DLB (26.5 ± 19.9months). Female sex, hallucinations, and behavioral dysregulation were identified as independent factors associated with AD (p < 0.05).

Conclusion

Pre-dementia NPS, particularly psychiatric manifestations, may link to early dementia. However, given the exploratory nature of the results, which is needed to confirm these findings. Clinicians should consider prodromal dementia in older adults with late-onset psychiatric symptoms to enable early intervention.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12888-026-08046-5.

Introduction

Over the past several decades, the aging population has led to a rapid increase in the prevalence of dementia. Dementia is a chronic, progressive neurodegenerative condition that leads to severe disability and eventually end-stage terminal illness, making early diagnosis and intervention crucial for improving prognosis. Mild Cognitive Impairment (MCI) is considered a transitional stage between normal cognitive function and dementia on the continuum of cognitive decline. Additionally, several biomarkers, such as β-amyloid and tau pathologies, have been identified as early indicators of dementia.

However, prior to the onset of dementia, even in cases of mild cognitive impairment (MCI), various neuropsychiatric syndromes (NPS) may manifest. NPS encompasses a constellation of abnormal clinical manifestations typically observed in the geriatric population, the clinical manifestations primarily include emotional/affective symptoms (e.g., anxiety, depression, apathy), psychiatric symptoms (e.g., agitation, irritability, hallucinations, delusions), and behavioral alterations (e.g., sleep disturbances, disinhibition, personality changes, aggressive behaviors).NPS are prevalent among the elderly population, with their incidence escalating as cognitive function deteriorates. The prevalence rates range from 30.8% in cognitively normal individuals to 80% in those diagnosed with dementia [1]. The incidence of affective disorders is notably high. A 2022 meta-analysis analyzed data from 10 studies involving 9,758 cognitively normal individuals, 1,057 with subjective cognitive decline (SCD), and 1,252 with mild cognitive impairment (MCI). The association between neuropsychiatric symptoms (NPS) and dementia risk highlights distinct patterns, with affective disorders (e.g., depression, anxiety) showing the highest incidence (32.84%), while abnormal perceptions or thoughts (e.g., delusions) are the least common (2.81%) [2]. However, these neuropsychiatric symptoms may manifest as either transient episodes or sustained late-life behavioral abnormalities. To rigorously distinguish sustained late-onset neuropsychiatric symptoms from transient occurrences and pre-existing psychiatric disorders, Taragano et al. (2009) first proposed the concept of ‘Mild Behavioral Impairment’ (MBI). MBI is defined as a cluster of abnormal behavioral symptoms emerging in later life (≥ 50 years of age), persisting for at least six months, and not attributable to other formal psychiatric diagnoses (such as major depressive disorder or schizophrenia) [3]. A large-scale retrospective study conducted in 2022 identified MBI as a significant predictor of both clinically diagnosed Alzheimer’s disease (AD) [hazard ratio (HR) = 1.75] and neuropathologically confirmed AD (HR = 1.59) [4]. Furthermore, this study revealed a gradient in the predictive utility of specific MBI domains for AD. Notably, psychotic symptoms exhibited the strongest predictive capacity (HR = 6.49), followed by social inappropriateness, impulse dyscontrol, decreased motivation, and affective dysregulation [4]. Moreover, a growing body of evidence from both longitudinal and cross-sectional studies consistently supports the association between MBI and subsequent cognitive decline or dementia [3, 5, 6]. Cognitive decline severity strongly correlates with mild behavioral impairment (MBI) prevalence and symptom severity, suggesting a bidirectional link between neurodegeneration and behavioral dysregulation [6]. Affective disorders and hallucinations/delusions are more prevalent in women, potentially reflecting sex-specific biological or psychosocial factors in Alzheimer’s disease progression [7]. Sleep disturbances, such as insomnia, are independently linked to increased dementia risk [8, 9]. Notably, these symptoms are often misdiagnosed as isolated psychiatric conditions in later life rather than early dementia biomarkers, delaying intervention. Cognitive impairment and NPS may synergistically be associated with accelerated neurodegeneration, underscoring the need for early screening of sleep disorders and sex-specific NPS patterns.

There has been ongoing exploration into the mechanisms underlying the occurrence of NPS. A substantial body of research has focused on the relationship between cerebrospinal fluid (CSF) Aβ levels and MBI symptoms. Cross-sectional studies consistently demonstrate that low Aβ levels in the CSF are linked to the onset of NPS, including aggression [10], agitation, irritability, anxiety [11], and depression [12]. Additionally, several large longitudinal studies have examined the relationship between amyloid deposition (PET-CT) and NPS symptoms. These studies indicate that NPS symptoms are often associated with greater Aβ deposition, and individuals with more significant Aβ accumulation are at an increased risk of developing NPS in the future [13, 14]. Sleep disturbances may also be associated with impaired amyloid clearance, subsequently potentially linking to the development of Alzheimer’s disease (AD) [15, 16].

Clinically, there is a growing awareness that NPS symptoms emerging in late life may be linked to neurodegeneration. However, current research continues to view NPS as a distinct domain. There is still no authoritative consensus on the comorbidity of different NPS symptoms, their sequence of onset, or the specific progression of NPS to dementia. Therefore, we conducted this preliminary study to explore: (1) determine the incidence of dementia among patients utilizing outpatient services and those admitted to wards in Shanghai, (2) potential differences between groups with and without NPS before dementia, (3) investigate the occurrence and sequence of non-cognitive symptoms preceding the onset of dementia, and (4) factors associated with clinical outcomes in dementia patients.

Method

Subjects

This was a preliminary retrospective observational study, which conducted a retrospective chart review of 201 patients (52 men and 149 women; mean age ± standard deviation, 76.01 ± 8.86 years) who had been diagnosed as dementia at the Department of Shanghai Mental Health Center, Shanghai Jiaotong University, between January 2018 and September 2024. Utilizing previous case records, we documented the time and exact type of dementia diagnosis for each participant and retrospectively reviewed their past medical records to identify the incidence of NPS from direct patient interviews, caregiver reports, and clinical observations recorded during outpatient visits. Patients with alcoholic dementia and syphilitic dementia were excluded. Patients without any history of NPS prior to dementia diagnosis were assigned to the NPS-negative group (NPS-). Patients were categorized into the NPS-positive group (NPS+) if they exhibited documented NPS preceding the dementia diagnosis that persisted for at least six months. For these participants, the time of first NPS onset and the history of psychotropic medication use were recorded. Cases where NPS were attributable to previously diagnosed psychiatric disorders (e.g., depression, schizophrenia) were excluded. In our study, we presented NPS as the overarching domain, with subsyndromal “affective dysregulation” (including depression, anxiety and apathy), “sleep trouble” (including early awake, easy awake, hard sleep and excessive dreaming) and “psychiatric disorders” (including disinhibition, agitation, hallucination, delusion and aggression) as components. This symptom grouping follows the structure commonly used in NPI-based NPS research, separating affective symptoms from psychosis/behavioral dysregulation domains. Additionally, we recorded the self-reported or caregiver-reported onset of memory decline(SCD) and calculated the time interval from memory decline to dementia diagnosis.

Diagnostic assessment for dementia

Diagnoses were established by senior professors of psychiatry based on a comprehensive integration of clinical presentation, neuropsychological assessments, and neuroimaging findings, with the neuropsychological assessments administered by independent, trained evaluators. Diagnoses were made in accordance with the ICD-10 criteria for research. Dementia with DLB and MCI were diagnosed using the DLB consensus criteria [17] and Petersen criteria [18], respectively. According to the previous studies [4, 19], SCD was defined as having subjective persistent cognitive complaints but not meeting the criteria of MCI or dementia. To ensure evaluator consistency and minimize potential bias from the initial clinical impression, a rigorous two-step diagnostic confirmation process was implemented.

The clinical evaluation included a comprehensive review of medical and psychiatric history, physical and neurological examinations, and neuropsychological assessments. While the Mini-Mental State Examination (MMSE) served as the primary screening tool for global cognitive staging in all participants, a broader battery of neuropsychological tests—including the Montreal Cognitive Assessment (MoCA), Clock Drawing Test (CDT), and Verbal Fluency Tests—was administered when clinically indicated to further characterize specific cognitive domains and differentiate between dementia subtypes.

Structural neuroimaging (MRI or CT) was reviewed for all included patients to support the clinical diagnosis. This was primarily used to assess patterns of atrophy (e.g., medial temporal lobe atrophy for AD, frontotemporal atrophy for FTD) and to rigorously exclude non-neurodegenerative causes of cognitive decline, such as vascular pathology (e.g., significant infarcts or white matter lesions), subdural hematoma, tumors, or normal pressure hydrocephalus. CSF biomarkers or amyloid PET were not routinely available due to the real-world retrospective nature of this cohort. Inter-rater discrepancies were resolved during consensus meetings. This procedure ensured maximal diagnostic rigor despite biomarker limitations.

The study protocol was approved by the Ethical Committee of the Shanghai Mental Health Center(2023-59). All data were coded and registered anonymously (Fig. 1).

Fig. 1.

Study design. This study selected elderly individuals who eventually developed dementia and retrospectively examined the presence of NPS prior to their dementia diagnosis. Furthermore, we conducted an in-depth exploration of the specific clinical features characterizing each symptom cluster among those elderly who had exhibited NPS manifestations prior to diagnosis. AD represents for Alzheimer’s Disease, DLB represents for Lewy bodies dementia, MD represents for mixed dementia, FTD represents for Frontotemporal dementia, VD represents for vascular dementia. In analyse 1, we compared the differences between participants with and without NPS before dementia. In analyse 2, we discovered characters of NPS before dementia, such as prevalence and order of appearance

Statistical analyses

We used the Chi square test to exam the differences between genders. Moreover, we used the Chi square test to exam the differences of non-cognitive symptoms between different types of dementia. The comparison of duration both between first non-cognitive symptom and memory decline as well as first non-cognitive symptom and dementia onset was performed using ANOVA-test. The subsequent forward conditional multivariate logistic regression analysis included variables with p < 0.05 in our cohort. The independent variables included age, gender, hallucination, difficulty falling asleep, sleep problems, psychiatric problems and disinhibition In the diagnosis, AD were compared with non-AD. Data were analyzed using SPSS 19.0 (IBM Corp., Armonk, NY, USA), and p < 0.05 was considered statistically significant in these tests. Post-hoc power analyses were conducted using G*Power 3.1 to determine the statistical power of the Chi-square tests given the sample size (N = 201) and α level of 0.05.

Results

Characteristics of the study participants

Between September 2023 and July 2024, a total of 383 patients participated in this survey, of whom 182 were subsequently excluded. Among the remaining participants, 201 individuals were diagnosed with dementia, including 138 with Alzheimer’s disease, 25 with Lewy body dementia, 17 with MD, 13 with FTD, and 8 with VD. The mean age of all participants was 76.0 years. Of the participants, 74.13% were women, 79.10% had attained secondary education or higher (i.e., more than 8 years of schooling), and 73.63% were married. It was observed that, compared to men, women had lower levels of educational attainment. The average MMSE score was 13.35 points, and the average educational attainment was 10.0 years (Table 1).

Table 1.

Characteristics of study participants in the total sample and by sex

Characteristics	Total sample (n = 201)	Men (n = 52)	Women (n = 149)	P-value
Age(years), means(SD)	76.01(8.86)	74.50(8.47)	76.54(8.96)	0.153
Age groups(years), n(%)
< 65	18(8.96)	4(7.69)	14(9.40)
65–69	24(11.94)	11(5.47)	13(8.72)
70–74	45(22.39)	16(7.96)	29(19.46)
75–79	42(20.90)	6(11.54)	36(24.16)
≥ 80	72(35.82)	15(28.85)	57(38.26)
Educational attainment, n(%)				0.069
Primary school and lower(6 years)	42(20.90)	7(13.46)	35(23.49)
Middle school(9 years)	80(39.80)	22(42.30)	58(38.92)
High school and above	79(39.30)	23(44.23)	56(37.58)
Marriage, n(%)				< 0.001**
Single	2(1.00)	0(0.00)	23(15.44)
Married	148(73.63)	47(90.38)	101(67.79)
Divorced or separated	4(2.00)	3(5.77)	1(0.67)
Bereft	47(23.38)	2(3.85)	47(31.54)
MMSE score	13.35(7.09)	13.12(6.89)	13.43(7.17)	0.782

Data are n (%) or mean (SD) unless specified otherwise. *p value for difference between men and women

Incidence of dementia and neuropsychiatric problems before dementia

Incidence of dementia

In all patients, AD was the most common diagnosis(68.66%), followed by DLB(12.44%), MD(8.46%), FTD(6.47%), and VD(3.99%)0.139 out of 201 psychiatric patients actually had NPS before diagnosis, these were presented as affective dysregulation(39.80%), sleep problems(46.77%) and psychiatric disorders(74.10%). By documenting the specific dementia subtype and the presence of pre-diagnostic neuropsychiatric problems for each patient, we found that 60.14% of AD patients, 100.00% of DLB patients, 75% of VD patients, 70.59% of MD patients and 100% of FTD patients had neuropsychiatric problems before they were diagnosed as dementia.

Prevalence of pre-existing NPS and comparative analysis between dementia patients with and without NPS

Among all the participants, 139 of them had neuropsychiatric problems before dementia (30 men, 109 women), and the mean age was 75.80 ± 8.91 years old. In analyse 1, we compared time interval between memory decline and dementia between participants with neuropsychiatric problems and participants without neuropsychiatric problems, and investigated the type of dementia they were ultimately diagnosed with, to compare the cognitive and outcome differences between the two groups, and further explore the relationship between NPS and dementia. Neuropsychiatric problems were more common in woman than in men (109/149 (73.15%) versus 30/52 (57.69%), p = 0.038). Patients exhibiting neuropsychiatric symptoms (NPS) experience accelerated dementia conversion, with a median interval of 23.41(± 25.01)months from memory decline to dementia diagnosis, while patients without NPS had a median interval of 42.21 (± 27.14) months from memory decline to dementia diagnosis. Moreover, patients with NPS are more likely to develop non-AD dementia.(40.29% vs. 11.29%, p < 0.001)(56/139 vs. 7/62) (Fig. 2).

Fig. 2.

Comparison of dementia types (%) between NPS + and NPS- groups at final outcome. *p value for difference in dementia types (AD versus non-AD) between NPS + and NPS-. *** is represented as p < 0.001. AD represents for Alzheimer’s Disease, DLB represents for Lewy bodies dementia, MD represents for mixed dementia, FTD represents for Frontotemporal dementia, VD represents for vascular dementia

Characters of NPS before dementia

Prevalence of non-cognitive problems

In analyse 2, we further discovered details of neuropsychiatric problems. We divided all the recorded problems into 3 domains(psychiatric disorders, sleep trouble and affective dysregulation). Among these domains, psychiatric problems was the most common (74.10%), followed by sleep trouble (46.77%), and affective dysregulation (39.80%). In each domain, there were significant differences in occurrence and time interval. In total problems, there were significant differences between AD and FTD, AD and DLB, DLB and MD(Effect size w = 0.33, Power = 0.98). In affective dysregulation, there was no significant difference in occurrence, while time interval between affective dysregulation and dementia had significant difference between AD and DLB (54 ± 39.13(months) versus 38.44 ± 28.39(months)). In sleep disorder, there were significant differences between AD and DLB, DLB and MD(Effect size w = 0.30, Power = 0.94). In psychiatric disorders, there were significant differences between AD and DLB, AD and FTD, DLB and VD, DLB and MD(Effect size w = 0.47, Power = 1.0), and time interval between psychiatric disorders and dementia had significant difference between AD and DLB (52.25 ± 47.64(months) versus 26.50 ± 19.94(months)) (Fig. 3).

Fig. 3.

Estimated prevalence of neuropsychiatric symptoms among people with dementia, by different kinds of dementia. AD represents for Alzheimer’s Disease, DLB represents for Lewy bodies dementia, MD represents for mixed dementia, FTD represents for Frontotemporal dementia, VD represents forvascular dementia. (A) Estimated prevalence of people with previous neuropsychiatric symptoms before they were diagnosed as dementia, by different kinds of dementia. (B) Estimated prevalence of people with psychiatric problems before they were diagnosed as dementia, by different kinds of dementia. (C) Estimated prevalence of people with sleep trouble before they were diagnosed as dementia, by different kinds of dementia. (D) Estimated prevalence of people with affective dysregulation before they were diagnosed as dementia, by different kinds of dementia. *p value for difference between AD, DLB, VD, MD and FTD. * is represented as p < 0.05, ** is represented as p < 0.01

Dementia clinical outcome factors

In our cohort, sex, neuropsychiatric problems before dementia, sleep disorder, hard sleep, BPSD, hallucination, and out control symptoms were all related to dementia outcome(p < 0.05). Subsequently, the multivariable logistic regression analysis identified sex, hallucination and disinhibition as independent predictive factors (p < 0.05) (Table 2).

Table 2.

Multivariate analysis of AD versus non-AD

Variables	Multivariate analysis
	β	OR	p-value
Age(years)	0.041	1.041(0.973-1.114)	0.240
Sex	1.652	5.217(1.504–18.103)	0.009**
Hallucination	1.726	5.618(1.623–19.444)	0.006**
Difficulty falling asleep	0.728	2.071(0.566-7.587)	0.272
Disinhibition	2.218	9.186(1.664–50.704)	0.011*

Note: The multivariable logistic regression analysis identified sex, hallucination and disinhibition count as independent cirrhosis predictive factors (p < 0.05) between AD and non-AD

According to the multivariate analysis comparing Alzheimer’s disease (AD) and non-Alzheimer’s disease (non-AD) conditions, sex, hallucinations, and disinhibition were identified as significant factors associated with the outcome.

Order of neuropsychiatric problems before dementia

According to the various types of neuropsychiatric symptoms (NPS), the time interval between the onset of NPS and subsequent cognitive decline varies among participants. Notably, psychiatric disorders exhibit the shortest interval between the onset of NPS and the development of dementia, suggesting that these types of NPS may manifest later in the disease progression. Similar trends were observed in the relationship between the onset of NPS and both memory decline and dementia. Conversely, sleep disturbances were associated with the longest interval between the onset of NPS and memory decline, averaging 47.5 months. It was observed that the time span between memory decline and the onset of dementia was relatively consistent across the three groups, ranging from 22.9 to 24 months. These findings highlight distinct temporal patterns in the progression from NPS onset to cognitive decline and dementia (Fig. 3).

Between different types of dementia, there were significant differences between AD and DLB regarding the time intervals from affective dysregulation to dementia and from psychiatric problems to dementia. Additionally, a notable difference was observed in the time interval from psychiatric disorders to memory decline between AD and DLB. However, once memory decline occurred, there was no significant difference in the time interval to dementia (Figs. 4 and 5).

Fig. 4.

A: Exclaims time interval between non-cognitive symptom onset to dementia, by different types of symptoms and dementia. B: Exclaims time interval between non-cognitive symptom onset to memory decline, by different types of symptoms and dementia. AD is Alzheimer’s Disease, DLB is Lewy bodies dementia, MD is mixed dementia, FTD is Frontotemporal dementia, VD is vascular dementia

Fig. 5.

Exclaims time interval from neuropsychiatric symptoms onset to cognitive decline(months), by different types of symptoms. There are three knots: symptom onset、memory decline and dementia. We tracked back from diagnosis and recorded the time from symptom onset to memory decline and dementia, and calculated the time interval between knots

Sex-related characters in dementia and non-cognitive problems before diagnosis

As was discovered before, we noticed several findings in sex. In each type of dementia, there was a significant difference only in proportion of sex between AD and MD(30/108 (0.28) versus 8/9 (0.89), p = 0.025). Totally, non-cognitive problems were more common in woman than in men (109/149 (73.15%) versus 30/52 (57.69%), p = 0.038). Moreover, in each domain of non-cognitive symptoms, there was no significant difference between men and women. In specific problem, there were significant differences in hallucination and delusion between men and women, respectively.(p = 0.037,0.013,respectively) (Table 3).

Table 3.

Prevalence of previous non-cognitive problems, by sex

Domains	Male	Female	p-value
Previous non-cognitive problems(total)			0.038*
Positive	30	109
Negative	22	40
Affective dysregulation			0.635
Positive	19	60
Negative	33	89
Sleep problems			0.152
Positive	20	74
Negative	32	74
Psychiatric problems			0.117
Positive	21	79
Negative	31	70
Hallucination			0.037*
Positive	8	45
Negative	44	104
Delusion			0.013*
Positive	5	39
Negative	47	110

Data are n unless specified otherwise. *p value for difference between men and women

Discussion

Neuropsychiatric symptoms (NPS) in preclinical dementia exhibit stratified predictive utility, with MBI, demonstrating superior specificity for neurodegenerative pathology compared to sleep disorders. While sleep disturbances in elderly populations, their diagnostic specificity for dementia subtypes remains limited. This preliminary study aimed to investigate the longitudinal associations between dementia and neuropsychiatric symptoms, including MBI and SD, and explore more characters of NPS. As hypothesized, this preliminary study demonstrated a longitudinal relationship between dementia and NPS: 69.2% participants had NPS before dementia, across various types of dementia. Furthermore, components of NPS were also associated with the incidence of dementia, psychiatric disorders has the shortest time interval between occurrence and dementia diagnosis. Additionally, the occurrence of NPS has differences in sex, delusions and hallucinations were observed more frequently in women.

In our study, the prevalence of neuropsychiatric symptoms (NPS) preceding dementia is observed to be 50.0% across various types of dementia. This is a high prevalence, worth more attention in clinical work. Past research also reveals that point of view, although there was considerable variability [20–22]. In a cohort study, they found that decreased motivation was associated with an increased risk of all-caused dementia [23]. This evidence underscores the importance of identifying early behavioral markers specific to dementia, rather than attributing these symptoms solely to isolated psychiatric conditions [24]. Patients with NPS before preceding dementia exhibited a shorter interval from symptom onset to dementia diagnosis compared to those without. This shares significant parallels with MBI in terms of underlying mechanisms. Compared to MCI patients without MBI, those with MBI exhibit more severe episodic memory decline [25] and carry a higher risk of progressing to dementia [26]. A study employing the MBI-C questionnaire revealed that cognitively normal subjects with MBI showed a more rapid decline in attention and working memory compared to healthy controls. Notably, the accelerated decline in working memory was associated with preclinical Alzheimer’s disease (AD) [6]. This suggests that older adults with MBI, regardless of their cognitive status, demonstrate accelerated memory decline, which may correlate with cerebral Aβ deposition and regional brain atrophy. A study of cognitively normal older adults found that MBI scores correlated with global and striatal Aβ burden. Among participants with MBI-C scores exceeding 8, 71% were amyloid-positive, compared to only 25% positivity in those scoring below 8 [7]. Another clinical cohort study of non-demented elders identified associations between MBI (particularly impulse dyscontrol and decreased motivation) and atrophy in two medial temporal lobe regions: the entorhinal cortex and hippocampus [10]. These findings collectively suggest that pre-dementia neuropsychiatric symptoms (NPS) may stem from incipient brain atrophy and early-stage pathological changes associated with dementia. These neuropathological alterations, while still subtle and insufficient to manifest cognitive impairment, appear to manifest initially as NPS preceding measurable cognitive decline. This evidence strongly supports NPS as early biomarkers of dementia, potentially emerging prior to detectable cognitive deterioration. In detail, the prevalence of NPS differs in different types of dementia. Our study revealed that FTD exhibited the highest proportion of neuropsychiatric symptoms (NPS), with 100% (13 out of 13) of cases affected. In contrast, Alzheimer’s Disease (AD) demonstrated the lowest proportion of NPS at 56.5%, a statistically significant difference from FTD (p = 0.004). These findings are consistent with previous research, which indicates that FTD frequently presents with neuropsychiatric symptoms as an initial manifestation [8]. However, our results did not reflect this distinction, potentially due to the limited sample size, which may have rendered the findings susceptible to the influence of individual participants. The findings in Alzheimer’s Disease (AD) were consistent with previous studies [27], indicating the generalizability of emotional problems within the preclinical dementia population. The prevalence of emotional problems was highest among all-cause dementia. However, among other NDDs (Neurodegenerative diseases), the prevalence of psychiatric disorders was the most significant. The relationship between NPS and dementia is also reflected in its association with the types of dementia outcomes. According to the multivariate analysis comparing Alzheimer’s disease (AD) and non-Alzheimer’s disease (non-AD) conditions, sex, hallucinations, and disinhibition were found to have a significant influence. Participants with disinhibition or hallucination were more frequently diagnosed with non-AD dementia. This trend was verified in past research. This was identified in past studies, MCI patients mostly converted to Alzheimer’s dementia, while MBI converted to FTD and Lewy body dementia [28]. Prior studies have consistently associated hallucinations with dementia with Lewy bodies [8]. Additionally, according to a cohort study from Ontario Neurodegenerative Disease Research Initiative (ONDRI) (n = 481), NPS frequency and severity are markedly greater in FTD as compared to other neurodegenerative diseases [29], same with another large study [30]. In details, another observational study found that participants with FTD had more frequent appetite disturbances, disinhibition, and abnormal motor behaviors as compared to AD and VD [31]. This higher rate of decline to non-AD dementia with inhibition or hallucination may be explained by a common pathology.

The sequence of occurrence varies according to the type of NPS. In cases of all-cause dementia, sleep disturbances may manifest first, appearing approximately 47.5 months prior to memory decline and 66.6 months before the formal diagnosis of dementia. Conversely, psychiatric disorders tend to emerge the latest, indicating a heightened risk of dementia and demonstrating a strong association with more severe cognitive decline. Neurobehavioral symptoms have a certain relationship with cognitive decline [32], but deeper associations are still unidentified. According to previous research, NPS is related to the progression of cognitive decline to varying degrees. A prospective cohort study conducted on elderly community members found that non-psychotic symptoms can effectively predict cognitive decline (MCI) five years later. The strongest predictor is agitation (hazard ratio [HR] = 3.06, 95% CI = 1.89–4.93), followed by apathy (HR = 2.26, 95% CI = 1.49–3.41), while delusions and hallucinations do not predict the occurrence of MCI [25]. In another research, the highest proportion of elderly individuals with psychosis converted to dementia (25.8%) [28]. Another research compared inhibition between elderly people and young people, individuals indicates that elderly participants had greater difficulty in inhibited tasks [9]. These research before have all showed the close association between psychiatric disorders and dementia. Another research found that MBI could predict the progression to both the clinical and the neuropathologically confirmed diagnosis of AD, especially psychosis from psychiatric disorders domain, showed the strongest contribution to this association [4]. This association may result from neurodegenerative pathology. The aging of inhibitory control has been posited to be associated with the functionality of the right inferior frontal gyrus (rIFG), a region implicated in inhibitory processes [33]. Additionally, the decline in behavioral inhibition (BI) with age may be related to the dopaminergic system, which undergoes deterioration during the aging process [34]. Some studies show amyloid accumulation has been associated with an increase in depressive, anxiety and apathy symptoms over time in non-demented individuals [13, 35]. MBI domain of impulse dyscontrol accounts for the correlation between more severe NPS symptoms and higher LC signal in tau-positive participants [32]. However, in another cross sectional study, psychosis was the MBI domain with the largest effect on the association between MBI and the progression of AD [4].

Our research yielded several gender-related findings. While gender/sex differences in noncognitive manifestations are well described in dementia population [36–38], little is known about these differences in individuals without dementia. In our study, delusions and hallucinations were observed more frequently in women (p = 0.238, p < 0.001), which contrasts with previous research indicating that affective dysregulation is more common in females [5]. Additionally, no significant gender differences were found in the occurrence of affective problems, which diverges from earlier research indicating that women are more prone to emotional disturbances, particularly during the menopausal transition, irrespective of their prior mood history [19]. Previous studies suggest that the prevalence of MBI – specifically decreased motivation, impulse dyscontrol, apathy, agitation, and irritability – in a cognitively healthy population might be higher in males [21, 25, 39]. Sex may also be associated with dementia outcome, which is similar with past research [40]. In our cohort, males are much more likely to become AD, while females are more likely to become non-AD. Past studies found that females bear a higher burden of dementias than men in terms of prevalence, incidence, and also severity [41]. Neuropathological changes in entorhinal cortex may describe these differences between gender [4, 42, 43]. Moreover, 5-HT1A receptors could also play a role in those differences. Previous evidence have shown that female sex hormones modulate 5-HT1A receptors, which have been suggested as one of the explanations for higher prevalence of major depressive disorder in females [44]. In addition, previous studies have proposed that genetic variation in the serotonin receptor gene 5-HT2A is associated with the occurrence of hallucinations and delusions in patients with AD [45, 46]. These evidence shows that sex participates in the association of mild behavioral impairment with cognitive aging [47].

These symptoms are indicative of neurodegeneration and may be associated with G-Protein-Coupled Receptors (GPCRs), which represent a potential therapeutic target for interventions in neurodegenerative disorders [48].

Crucially, our findings underscore that late-onset neuropsychiatric symptoms in the elderly should be viewed as early signs of underlying neurodegeneration. However, recognizing these symptoms should serve as the starting point, not the endpoint, of the diagnostic process. To strengthen diagnostic validity, we strongly advocate that the identification of persistent late-onset NPS should trigger a comprehensive diagnostic workup that extends beyond clinical phenomenology. Specifically, integrating objective biological markers—including structural neuroimaging and, where available, fluid biomarkers (CSF or plasma Aβ/Tau) or amyloid-PET—is essential. This multi-modal approach will help clinicians more accurately distinguish prodromal dementia from primary psychiatric mimics, thereby facilitating earlier and more targeted interventions.

Limitation

Our study has several limitations inherent to its retrospective design and real-world clinical setting. First, regarding diagnostic precision, while diagnoses were established based on expert consensus and structural neuroimaging, we lacked routine confirmation via fluid biomarkers (CSF Aβ42/Tau) or amyloid-PET imaging. Consequently, our diagnoses represent clinical rather than pathological classifications. This reflects the current reality of outpatient practice, where invasive biomarker testing is not yet universally accessible.

Second, regarding cognitive assessment, the Mini-Mental State Examination (MMSE) was the primary screening tool available for all patients. More sensitive instruments for mild cognitive impairment, such as the MoCA or comprehensive neuropsychological batteries, were not consistently administered. This may limit the sensitivity in detecting subtle cognitive deficits.

Third, differentiating between NPS as prodromal features of dementia and those stemming from primary psychiatric disorders remains a diagnostic challenge without longitudinal biomarker tracking. Although we rigorously reviewed psychiatric history and treatment response to minimize misclassification, symptom overlap introduces a potential for diagnostic ambiguity.

Fourth, regarding the sample size, although the number of patients in the non-AD subgroups (e.g., FTD and VD) was relatively limited, a post-hoc power analysis (G*Power 3.1) revealed high statistical power (0.94–1.00) for our key comparisons. This indicates that the study possessed sufficient sensitivity to detect significant differences in NPS prevalence despite the unequal group sizes. Nevertheless, to maintain rigor, we interpret these subgroup-specific findings as exploratory and hypothesis-generating, warranting further validation in larger multi-center cohorts.

Finally, limitations inherent to retrospective data collection from routine outpatient visits must be acknowledged. Unlike prospective studies using standardized instruments at fixed intervals, our symptom identification relied on longitudinal medical records. Consequently, subtle or fluctuating symptoms (e.g., mild apathy, sleep disturbances, or irritability) might have been underreported or less consistently documented compared to overt behavioral disturbances. Furthermore, the determination of the onset time for subjective cognitive decline (SCD) relied on patient or caregiver reports, introducing potential recall bias. Additionally, due to the irregularity of follow-up intervals, we were unable to definitively determine the longitudinal trajectory of symptoms—specifically, whether they had fully resolved or persisted immediately prior to the formal diagnosis. However, despite these constraints, our study provides valuable ecological validity, reflecting how these symptoms manifest and are recorded in real-world psychiatric practice, serving as pragmatic early red flags for underlying neurodegeneration.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Yuhang Li: Conceptualization, Investigation, Writing – original draft; Dong Xu: Investigation; Chao Gu: Writing – review & editing; Liying Yang: Visualization; Yehao Jiang: Visualization, Ning Su: Writing – review & editing; Bo Hong: Writing – review & editing; Yuanyuan Liu: Writing – review & editing; Xia Li: Writing – review & editing; Shifu Xiao: Writing – review & editing; Ling Yue: Conceptualization, Writing – review & editing.

Funding

This work was supported by the China Ministry of Science and Technology (STI2030-Major Projects-2022ZD0213100;2023YFC3603200), Shanghai public health projects (GWVI-11.2-XD24).

Data availability

The data supporting the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Declarations

Ethical approval

The study protocol was approved by the Ethical Committee of the Shanghai Mental Health Center(2023-59),in accordance with the Declaration of Helsinki. All data were coded and registered anonymously. Clinical trial number: not applicable.

Standard declaration template for retrospective studies

This retrospective study was approved by the Ethical Committee of the Shanghai Mental Health Center (Approval No. 2023-59), with a waiver of individual informed consent. All data were anonymized to strictly protect patient privacy.

Competing interests

The authors declare no competing interests.