Impact of family function on psychological distress and its threshold effect in informal caregivers of elderly people with dementia Authors: Yi-xin Wang, Fan Zhang, Chong-qing Shi, Jing Shao Published: 2025 ABSTRACT ObjectsThe Circumplex model has been a subject of controversy. Our study aims to validate its curvilinear hypothesis by examining the relationship between family function and psychological distress in informal caregivers of people with dementia (PwD). If a curvilinear relationship is identified, it will provide empirical support for the Circumplex model. Conversely, if a linear relationship is found, the threshold effect can be further clarified, thereby offering a foundation for personalized psychological intervention strategies for caregivers at the family level.MethodsWe recruited a total of 267 elderly PwD and their informal caregivers as research subjects. Using a self-designed questionnaire, the Kessler Psychological Distress Scale, and the Family Adaptation and Cohesion Evaluation Scales to detect curvilinear or linear relationships, perform smooth curve fitting, and conduct threshold effect analysis.ResultsBetween-group differences were significant across the three family types (P = 0.001), as were within-group differences among the four family function types (P = 0.000, P = 0.000). A negative correlation was observed between family function and psychological distress. When the family score was ≤ 140, the psychological distress score decreased as the family function score increased (P < 0.001). However, when the family function score was > 140, the level of psychological distress remained unchanged (P = 0.552).ConclusionThe better the family function of caregivers, the less likely adverse psychological conditions are to occur. A family function score of 140 can serve as a node for assigning tiered, individualized interventions. Introduction Dementia is a syndrome caused by multiple diseases that damage neural cells and impair brain function over time, typically leading to a decline in cognitive abilities beyond the expected biological effects of aging [1]. It is currently the seventh leading cause of death worldwide and one of the primary reasons for the loss of self-care abilities and increased dependence among older people [1]. Informal caregivers are estimated to provide approximately 133 billion hours of care annually to the people with dementia (PwD) [2], which includes medical care, assistance with basic daily activities, and management of neurological and psychiatric symptoms associated with the condition [3]. These caregivers face a significant burden, with anxiety and depression being the most prevalent issues [4]. Psychological distress is a painful emotional experience that arises from various factors and encompasses both anxiety and depression [5]. The psychological distress experienced by informal caregivers not only affects the quality of care, but also has a considerable impact on the quality of life and disease prognosis for PwD [6]. In China, due to the underdevelopment of geriatric care facilities and the prevalence of traditional family values, “family-centered” is the preferred mode of care for most PwD. As the most fundamental social unit, a family member’s functional impairment or illness inevitably impacts other family members [7]. Olson’s Family Circumplex Model Theory is grounded in the Family System Theory, which views family function in terms of the effectiveness of emotional ties between family members, family rules, communication patterns, and responses to external events [8]. This model is composed of adaptability and cohesion. Family adaptability is categorized into four types based on scores: chaotic, flexible, structured, and rigid. Similarly, family cohesion is classified into four types: enmeshed, connected, separated, and disengaged. Matching the adaptability and cohesion levels pairwise would result in 16 sets of family function types: Families that score moderate on all four types across both subscales are considered balanced families. These families demonstrate moderate emotional distance among members, maintaining both intimacy and mutual independence, characteristics of well-adapted, healthy families; families that are extreme in all 4 types of the 2 subscales belong to the extreme families. Such families and their members often exhibit diametrically opposed characteristics in terms of cohesion and adaptability; the other 8 types are mid-range families (Fig. 1) [9]. Fig. 1Classification of Family Functions [9] Classification of Family Functions [9] The Circumplex Model Theory posits that there is a curvilinear relationship between family function and cohesion and adaptability. Both high and low levels of cohesion and adaptability can lead to various problems within a family. Optimal family function occurs when these dimensions are in a moderate state; in other words, a balanced family represents the ideal state of family dynamics. Scholars have tested Olson’s Circumplex Model Theory and have consistently supported its theoretical assumptions [10, 11]; however, some studies have reached different conclusions, indicating that families with higher levels of cohesion and adaptability demonstrate stronger family function, suggesting a linear relationship between them [12, 13]. In 1991, Olson revised his original curvilinear model to a three-dimensional (3-D) linear model theory [14]. This linear model posits that families with high scores in cohesion and adaptability exhibit effective function, while those with low scores tend to function poorly. Olson also contends that, despite the differences between the curvilinear and linear models, there is no fundamental conflict between them. Currently, most research addressing the testing of curvilinear models of family function focuses on children with behavioral problems [12, 15], and it remains unclear whether the family function of informal caregivers for PwD conforms to the curvilinear model. Therefore, our study aims to examine whether there is a curvilinear or linear relationship between psychological distress and family function among informal caregivers of PwD. If a curvilinear relationship is identified, it would provide empirical support for Olson’s theory. Conversely, if such a relationship does not exist, it would indicate a linear correlation between family function and cohesion and adaptability. It is worth noting that previous studies have shown that many research variables do not exhibit a linear relationship. Within a specific range, they may produce zero or positive effects, but when exceeding a certain threshold, the magnitude and/or direction of the effect may change. This phenomenon is known as the threshold effect [16]. If the linear hypothesis is confirmed, we hope to clarify the threshold effect through further research to assist clinical workers in providing personalized interventions to prevent and address the adverse psychological conditions of informal caregivers for PwD. Methods Design and participants A convenience sampling method was employed to select PwD and their informal caregivers from June 2022 to June 2023 at memory clinics, neurology clinics, and inpatient departments of three grade-A hospitals in Wuhan, Hubei Province, China. PwD were eligible for inclusion in this study if they were aged ≥ 60 years old [17] and met the World Health Organization International Classification of Diseases, 11th Revision (ICD-11) diagnostic criteria for dementia [18]. Informal caregivers were eligible to participate if they (1) were ≥ 18 years old; (2) had a length of care ≥ 3 months, ≥ 8 h per week, and were responsible for the primary caregiving duties [6, 19]; and (3) were conscious and able to communicate. Employed caregivers were excluded from the study. The G*Power 3.1.9.7 software was utilized to calculate the required sample size based on the following parameters: an effect size of 0.15, an alpha level of 0.05, and a power of 0.90 [20], with an estimated sample size of 196. Accounting for a 15% potential attrition rate, 226 subjects were required. Additionally, to ensure the stability of the piecewise model for threshold effects testing, a sample size range of 240–300 is recommended to obtain robust statistical results. Our study ultimately enrolled 267 pairs of subjects, all of whom were included in the final analysis. Instruments Data was collected in this study by using a combination of a questionnaire designed by the research team, the Kessler psychological distress scale, and the Family adaptation and cohesion evaluation scales II-CV. Measurement of demographic variables Demographic data and clinical characteristics of PwD were collected, including age (years), gender, education level, marital status, numbers of children, type of medical expenses payment, Activity of daily living (ADL), Mini-Mental State Examination (MMSE), types of dementia, duration of illness (years); caregiver-related information included age, gender, education level, marital status, employment situation, monthly family income (yuan), domicile place, relationship with PwD, whether to live with the PwD, average daily care duration (h), patient care duration (years), physical status. Kessler psychological distress scale (K10) We utilized the Kessler Psychological Distress Scale to evaluate the informal caregivers of PwDs’ psychological distress. This scale was developed by Kessler et al. [21], and the Chinese version was translated by Zhou et al. [22]. It adopted the 5-point Likert scale and consist of two parts: anxiety and depression, with a total score ranging from 10 to 50 points. The scoring interpretation is as follows: 10–15 indicates that the individual is likely well; 16–21 suggests a mild mental disorder; 22–29 indicates a moderate mental disorder; and 30–50 signifies a severe mental disorder. A higher score reflects greater psychological distress. The Cronbach’ alpha for this study was 0.93. Family adaptation and cohesion evaluation scales II-CV (FACES II-CV) The Family Adaptation and Cohesion Evaluation Scales II-CV was employed to assess the family function of informal caregivers. The FACES-II scale was developed by Olson et al. [10], and the Chinese version was translated and revised by Fei et al. [23]. It consists of a total of 30 items, including 16 items that measure cohesion and 14 items that assess adaptability. Each item is rated on a 5-point scale, ranging from 1 (never) to 5 (always). The Cronbach’ alpha for the two subscales were 0.88 and 0.91, respectively. Barthel index (BI) This study used the Barthel Index (BI) to assess the level of daily living activities in PwD [24]. The BI consists of 10 items, with a total score ranging from 0 to 100 points. Scores of 0–40 represent severe dependence, 45–60 indicate moderate dependence, 65–95 denote mild dependence, and 100 signifies complete independence. The higher the total score, the better their ability to engage in daily activities. Its Cronbach’s alpha is 0.90. Mini-mentalstate examination(MMSE) This scale is simple and easy to administer, making it the preferred tool for cognitive screening [25]. It is divided into five sections: orientation, memory, attention, calculation, recall, and language ability. The total score is 30 points, and a score < 25 points is considered indicative of cognitive impairment. Its Cronbach’s alpha is 0.82. Procedure During the project preparation phase, a research team was established, comprising a neurology specialist (specializing in dementia diagnosis and management), a team leader (specializing in geriatric psychology and clinical research methodology), primary nurses, and other researchers. Concentrated training was provided to all investigators, enabling each team member to understand the content, purpose, significance, and methods of the survey questionnaire. This training also included communication skills with PwD and their families during the investigation; in the formal survey phase, face-to-face on-site questionnaires were administered only after obtaining informed consent from the participants (a combination of self-report and caregiver proxy approach was used to collect all scales and questionnaires). For participants with visual impairments or illiteracy, interviewers verbally presented the questions and recorded responses accurately. Participants without such barriers completed the questionnaires through self-administration under the guidance of the research team; upon completion of the survey, questionnaires were immediately collected and promptly reviewed for any omissions or errors, with respondents guided to supplement or correct their responses as necessary. Quality control Participants were selected in strict accordance with the inclusion and exclusion criteria. To minimize selection bias, recruitment was conducted concurrently across multiple departments (neurology departments, geriatrics departments, and memory clinics) in three hospitals. Throughout data collection, the team leader performed on-site supervision and quality control by performing daily spot checks to ensure questionnaire and scale completeness, regularly conducting inter-rater reliability assessments, and delivering immediate corrective feedback for any inconsistencies identified. To avoid information bias, this study employed coded anonymity for data collection, with all participants identified by unique identification codes. Additionally, double-entry data verification was implemented, and outliers were analyzed promptly to minimize data management bias. Data analysis Two independent researchers entered the questionnaire data. Descriptive statistics were performed using SPSS 26.0 to summarize the general information of the subjects, analyze the distribution of family functional classifications, and generate scatter plots. Additionally, risk factors for psychological distress were identified using univariate analysis and multiple linear regression analysis. Empower Stats (R 4.0) was employed for smooth curve fitting and threshold effect analysis to examine the relationship between psychological distress and family function and to identify any turning points. Results Demographic characteristics A total of 267 pairs of elderly PwD and their informal caregivers were included. 53.93% of PwD were male, 46.07% were female. More than half of informal caregivers are female (62.17%). Additional details are provided in Table 1. Table 1Demographic characteristics and univariate analysis of research subjects (n = 267)PwDn (%) P Caregiversn (%) P Gender 0.858 Gender 0.454Male144(53.93)Male101(37.83)Female123(46.07)Female166(62.17) Age(years) 0.334 Age(years) 0.19360 ~ 6983(31.09)18 ~ 297(2.62)70 ~ 7990(33.71)30 ~ 59144(53.93)80 ~ 8964(23.97)60 ~ 89113(42.32)≥ 9030(11.21)≥ 903(1.12) Education level 0.305 Education level 0.000 Primary school or below153(57.30)Primary school or below60(22.47)Junior school92(34.46)Junior school92(34.46)High school or above22(8.24)High school or above115(43.07) Marital status 0.158 Marital status 0.697Single3(1.12)Single12(4.49)Married180(67.42)Married249(93.26)Divorced or Widowed84(31.46)Divorced or Widowed6(2.25) Numbers of Children 0.754 Employment situation 0.38004(1.50)Full-time job89(33.33)153(19.85)Part-time job36(13.48)2114(42.70)Retire121(45.32)3 or more96(39.95)No21(7.87) Type of medical expenses payment 0.120 Monthly family income (yuan) 0.000 Medical insurance241(90.26)≤ 300067(25.09)Own expense13(4.87)3000–500099(37.08)Other13(4.87)≥ 5000101(37.83)Activity of daily living, ADL 0.000 Domicile place 0.003 Complete independent13(4.87)Rural81(30.34)Mild dependence94(35.20)Urban186(69.66)Moderate dependence88(32.96) Relationship with PwD 0.168Severe dependence72(26.97)Spouse81(30.34)Mini-Mental State Examination, MMSE 0.000 Child153(57.30)Mild45(16.85)Relatives26(9.74)Moderate102(38.20)Other7(2.62)Severe120(44.95) Whether to live with the PwD 0.063 Types of dementia 0.432Yes184(68.91)Alzheimer’s disease (AD)110(41.20)No83(31.09)Vascular dementia (VaD)110(41.20) Average daily care duration (h) 0.270Mixed dementia (MD)34(12.73)≤ 536(13.48)Other13(4.87)5–848(17.98) Duration of illness(years) 0.360≥ 8183(68.54)≤ 171(26.59) Patient care duration (years) 0.0741–271(26.59)≤ 167(25.09)2–565(24.34)1–267(25.09)5–1065(24.34)2–567(25.09)≥ 1030(11.24)5–1031(11.61)≥ 1035(13.12) Physical status 0.001 Worse39(14.61) General129(48.31) Good99(37.08) Table 1Demographic characteristics and univariate analysis of research subjects (n = 267)PwDn (%) P Caregiversn (%) P Gender 0.858 Gender 0.454Male144(53.93)Male101(37.83)Female123(46.07)Female166(62.17) Age(years) 0.334 Age(years) 0.19360 ~ 6983(31.09)18 ~ 297(2.62)70 ~ 7990(33.71)30 ~ 59144(53.93)80 ~ 8964(23.97)60 ~ 89113(42.32)≥ 9030(11.21)≥ 903(1.12) Education level 0.305 Education level 0.000 Primary school or below153(57.30)Primary school or below60(22.47)Junior school92(34.46)Junior school92(34.46)High school or above22(8.24)High school or above115(43.07) Marital status 0.158 Marital status 0.697Single3(1.12)Single12(4.49)Married180(67.42)Married249(93.26)Divorced or Widowed84(31.46)Divorced or Widowed6(2.25) Numbers of Children 0.754 Employment situation 0.38004(1.50)Full-time job89(33.33)153(19.85)Part-time job36(13.48)2114(42.70)Retire121(45.32)3 or more96(39.95)No21(7.87) Type of medical expenses payment 0.120 Monthly family income (yuan) 0.000 Medical insurance241(90.26)≤ 300067(25.09)Own expense13(4.87)3000–500099(37.08)Other13(4.87)≥ 5000101(37.83)Activity of daily living, ADL 0.000 Domicile place 0.003 Complete independent13(4.87)Rural81(30.34)Mild dependence94(35.20)Urban186(69.66)Moderate dependence88(32.96) Relationship with PwD 0.168Severe dependence72(26.97)Spouse81(30.34)Mini-Mental State Examination, MMSE 0.000 Child153(57.30)Mild45(16.85)Relatives26(9.74)Moderate102(38.20)Other7(2.62)Severe120(44.95) Whether to live with the PwD 0.063 Types of dementia 0.432Yes184(68.91)Alzheimer’s disease (AD)110(41.20)No83(31.09)Vascular dementia (VaD)110(41.20) Average daily care duration (h) 0.270Mixed dementia (MD)34(12.73)≤ 536(13.48)Other13(4.87)5–848(17.98) Duration of illness(years) 0.360≥ 8183(68.54)≤ 171(26.59) Patient care duration (years) 0.0741–271(26.59)≤ 167(25.09)2–565(24.34)1–267(25.09)5–1065(24.34)2–567(25.09)≥ 1030(11.24)5–1031(11.61)≥ 1035(13.12) Physical status 0.001 Worse39(14.61) General129(48.31) Good99(37.08) Demographic characteristics and univariate analysis of research subjects (n = 267) Curvilinear or linear relationship test The descriptive statistics for the four levels of cohesion and adaptability in family functions indicate that the four types of families exhibit similar numbers in terms of cohesion. Notably, in the adaptability domain, close to half of the families (45.31%) were classified as chaotic type, which was the most prevalent pattern. Mid-range families represent the largest group (39.33%), while extreme families constitute the smallest group (29.96%) (Table 2). Table 2Classification of family function and differential testing of research subjects (n = 267)VariablesRate (%)Median,25th ,75th percentile P Cohesion disengaged64(23.97)31.00,24.00,36.000.000separated65(24.34)27.00,24.00,34.50connected75(28.09)23.00,17.50,23.50enmeshed63(23.60)17.00,14.00,20.00 Adaptability rigid21(7.87)24.00,18.00,29,000.000structured50(18.73)21.00,16.00,27.00flexible75(28.09)19.00,15.50,22.00chaotic121(45.31)16.00,14.00,19.00 Family type balanced82(30.71)17.00,14.00,23.500.001mid-range105(39.33)21.00,17.00,26.00extreme80(29.96)23.50,18.75,29.00 Family function group Lower FACESII-CV score48(17.98)26.50,22.00,35.000.000Balanced FACESII-CV score172(64.42)21.00,16.00,27.00Higher FACESII-CV score47(17.60)15.00,14.00,17.50 Table 2Classification of family function and differential testing of research subjects (n = 267)VariablesRate (%)Median,25th ,75th percentile P Cohesion disengaged64(23.97)31.00,24.00,36.000.000separated65(24.34)27.00,24.00,34.50connected75(28.09)23.00,17.50,23.50enmeshed63(23.60)17.00,14.00,20.00 Adaptability rigid21(7.87)24.00,18.00,29,000.000structured50(18.73)21.00,16.00,27.00flexible75(28.09)19.00,15.50,22.00chaotic121(45.31)16.00,14.00,19.00 Family type balanced82(30.71)17.00,14.00,23.500.001mid-range105(39.33)21.00,17.00,26.00extreme80(29.96)23.50,18.75,29.00 Family function group Lower FACESII-CV score48(17.98)26.50,22.00,35.000.000Balanced FACESII-CV score172(64.42)21.00,16.00,27.00Higher FACESII-CV score47(17.60)15.00,14.00,17.50 Classification of family function and differential testing of research subjects (n = 267) According to Olson’s Circumplex Model Theory, we hypothesize that a significant difference between groups, coupled with no significant difference within groups, would indicate that the sample conforms to the Circumplex model [10]. The results of the between-group difference test for the three family types showed statistical significance (P = 0.001). Additionally, the within-group difference test for the four family function types among the three family types also demonstrated statistical significance (P = 0.000, P = 0.000) (Table 2). Therefore, we preliminarily conclude that the sample in this study does not conform to the Circumplex model distribution. We categorized cohesion and adaptability into four levels and family function scores into three levels based on their respectively scores (Fig. 2). Groups with higher family adaptability scores tended to exhibit lower mean scores for psychological distress. Similarly, groups with higher family cohesion scores also demonstrated lower mean scores for psychological distress (Table 2). The mean scores for psychological distress were the highest in the lower FACES II-CV group, followed by the balanced group, while the higher FACES II-CV group had the lowest scores (Table 2). So, we believe that there is a linear relationship between the family function and psychological distress among informal caregivers of elderly PwD, which means that better family function is associated with reduced anxiety and depression in caregivers. Fig. 2Scatter plot of family function scores for research subjects. *Four cohesion categorized groups: disengaged (< 55), separated (55 ~ 63.9), connected (64 ~ 72), enmeshed (>72); Four adaptability categorized groups: rigid (< 44), structured (44 ~ 50.9), flexible (51 ~ 57), chaotic (>57). Family function scores were divided according to the cut-off values recommended by Olson et al. [20]. Using the Mean (M) ± 1 standard deviation (SD) of the study population as the balanced family, the remaining were lower FACES scores and Higher FACES scores Scatter plot of family function scores for research subjects. *Four cohesion categorized groups: disengaged (< 55), separated (55 ~ 63.9), connected (64 ~ 72), enmeshed (>72); Four adaptability categorized groups: rigid (< 44), structured (44 ~ 50.9), flexible (51 ~ 57), chaotic (>57). Family function scores were divided according to the cut-off values recommended by Olson et al. [20]. Using the Mean (M) ± 1 standard deviation (SD) of the study population as the balanced family, the remaining were lower FACES scores and Higher FACES scores Smooth curve fitting and threshold effect analysis To determine additional factors significantly impacting psychological distress, we first analyzed general demographic variables using univariate analysis. Subsequently, variables with statistical significance from the univariate analysis were included in a multiple regression analysis. The results indicated that PwDs’ ADL (β = 0.183, P = 0.019), caregivers’ monthly family income (β=-0.241, P = 0.000), and caregivers’ physical status (β = 0.183, P = 0.019) were risk factors for psychological distress (Table 3). Table 3Multiple regression analysis of research subjectsBβt P ToleranceVIF(Constant)24.566-8.2000.000--PwDs’ ADL1.6470.1832.3610.0190.5211.918PwDs’ MMSE1.2480.1151.4970.1350.5311.882Caregivers’ monthly family income-2.459-0.241-3.5270.0000.6671.498Caregivers’ domicile place-0.120-0.007-0.1010.9200.6561.524Caregivers’ physical status-1.955-0.167-2.9660.0030.9821.018 Table 3Multiple regression analysis of research subjectsBβt P ToleranceVIF(Constant)24.566-8.2000.000--PwDs’ ADL1.6470.1832.3610.0190.5211.918PwDs’ MMSE1.2480.1151.4970.1350.5311.882Caregivers’ monthly family income-2.459-0.241-3.5270.0000.6671.498Caregivers’ domicile place-0.120-0.007-0.1010.9200.6561.524Caregivers’ physical status-1.955-0.167-2.9660.0030.9821.018 Multiple regression analysis of research subjects After controlling for PwDs’ ADL, caregivers’ monthly family income, and caregivers’ physical status, smooth curve fitting analysis revealed an inverse association between family function and psychological distress, with psychological distress exhibiting a continuous downward trend as family function scores increased (Fig. 3). Fig. 3Smooth curve fitting graph for family function and psychological distress scores Smooth curve fitting graph for family function and psychological distress scores The threshold effect analysis module of the Empower Stats software was used to examine the relationship between family function and psychological distress. The results indicated that the key turning point for the family function score was 140. When the family score ≤ 140, each 1-point decrease in family function scores was associated with an approximate 0.3-point increase in psychological distress levels, and this difference was statistically significant (P < 0.001); When family function scores > 140, there was no corresponding decrease in psychological distress levels, and the difference between the two was not statistically significant (P = 0.552), indicating a relative plateau phase. According to the results presented in Table 4, a family function score of 140 can serve as a monitoring indicator for psychological distress. Table 4Threshold effect analysis of family function and psychological distress scoresItemsβ(95%CI) P Model1One line effect-0.226(-0.269, -0.184)< 0.001Model2Turning point(K)140< K effect 1-0.257(-0.306, -0.209)< 0.001>K effect 20.073(-0.168, 0.315)0.552Effect 2 − 10.331(0.068, 0.593)0.014*One line effect: the overall linear regression coefficient of the association between family functioning and psychological distress without considering the threshold; Turning point (K): the critical value of the family function score at which of association with psychological distress changes significantly< K effect 1: The regression coefficient of family function on psychological distress when family function score is below the turning point (K = 140); >K effect 2: The regression coefficient of family function on psychological distress when family function score is higher than the turning point (K = 140); Effect 2 − 1: The difference in regression coefficients between effect 2 and effect 1(> K effect 2 minus < K effect 1), used to test the statistical significance of the threshold effect Table 4Threshold effect analysis of family function and psychological distress scoresItemsβ(95%CI) P Model1One line effect-0.226(-0.269, -0.184)< 0.001Model2Turning point(K)140< K effect 1-0.257(-0.306, -0.209)< 0.001>K effect 20.073(-0.168, 0.315)0.552Effect 2 − 10.331(0.068, 0.593)0.014*One line effect: the overall linear regression coefficient of the association between family functioning and psychological distress without considering the threshold; Turning point (K): the critical value of the family function score at which of association with psychological distress changes significantly< K effect 1: The regression coefficient of family function on psychological distress when family function score is below the turning point (K = 140); >K effect 2: The regression coefficient of family function on psychological distress when family function score is higher than the turning point (K = 140); Effect 2 − 1: The difference in regression coefficients between effect 2 and effect 1(> K effect 2 minus < K effect 1), used to test the statistical significance of the threshold effect Threshold effect analysis of family function and psychological distress scores *One line effect: the overall linear regression coefficient of the association between family functioning and psychological distress without considering the threshold; Turning point (K): the critical value of the family function score at which of association with psychological distress changes significantly < K effect 1: The regression coefficient of family function on psychological distress when family function score is below the turning point (K = 140); >K effect 2: The regression coefficient of family function on psychological distress when family function score is higher than the turning point (K = 140); Effect 2 − 1: The difference in regression coefficients between effect 2 and effect 1(> K effect 2 minus < K effect 1), used to test the statistical significance of the threshold effect Discussion This study supports the theory that the relationship between family function and psychological distress in informal caregivers of PwD may not always be curvilinear. This is consistent with some previous research results [12, 15, 26]. In other words, the more adaptive or cohesive the caregivers are, the fewer negative emotions they exhibit. Koutra et al. confirmed that the characteristics of family dysfunction can trigger emotional reactions toward patients, exacerbate the burden on caregivers caring for family members with mental illness, and further increase the level of psychological distress experienced by these caregivers [27]. Souza also believed that there is a correlation between impaired family function and anxiety and depression, suggesting that the emotional comfort provided by effective family function can enhance the physical and mental health of patients [28]. Good family function creates a supportive environment for family members, promoting their physical and mental well-being while reducing the incidence of depression, anxiety, and depressive symptoms; dysfunctional families often struggle to provide necessary care for patients, which can exacerbate anxiety and depression [29]. There has always been controversy over Olson’s curvilinear hypothesis, but this study confirmed its inapplicability through empirical research in a sample of informal caregivers of elderly PwD. This suggests that researchers should fully consider the appropriateness of this theory when using it. Although, Olson modified the original hypothesis of the curvilinear model to a 3-D linear model theory on the basis of summarizing the previous studies, most of the studies using the family function model in China are still based on Olson’s curvilinear model, which lacks the validation in the context of Chinese culture and the development of the theoretical model of family function suitable for the Chinese population. Our study examined the relationship between family function and psychological distress using the FACES II-CV, as Olson’s curvilinear model was constructed and designed based on the FACES II scale. Furthermore, the Chinese version is the only one that has been fully validated among caregivers in mainland China [30]. Although the FACES series has been updated to versions III and IV [30, 31], with improvements in dimensions and items, they have not yet undergone localization validation within the Chinese population. Consequently, their suitability cannot be ensured at present. Future research should prioritize the localization and cultural adaptation of the FACES scale, implementing targeted updates and validation tailored to the Chinese cultural context. This will provide a more accurate and reliable measurement tool for research and intervention practices. The results of this study confirm that the PwDs’ ADL, caregivers’ monthly family income, and caregivers’ physical status are risk factors for caregivers’ psychological distress, which is consistent with the findings of previous studies [32–34]. As PwDs’ condition progresses, their ability to perform activities of daily living gradually declines, and may even be completely lost. Self-care ability is a fundamental skill necessary for individuals to live independently; as a patient’s self-care ability diminishes, their dependence on caregivers increases, which adds to the caregiver’s burden and can lead to adverse mental health outcomes [32]. Similarly, lower family income levels and poorer physical status among caregivers undoubtedly exacerbate this burden, ultimately resulting in psychological distress for the caregivers [35]. This study identified through curve fitting and threshold effect analysis that psychological distress scores decreases with increasing family function scores. However, when the family function scores exceed 140 points, the downward trend becomes less pronounced, entering a relatively stable plateau period. This has never been discovered in previous studies and provides empirical guidance for implementing differentiated interventions tailored to caregivers with varying family function levels. When family function scores are ≤ 140 points, according to Maslow’s Hierarchy of Needs theory [36], caregivers in this group typically face multiple pressures. Their physiological needs (e.g., sleep deprivation) and safety needs (e.g., concerns about the patient wandering or experiencing accidents) remain unmet over the long term, leading to psychological distress driven by “survival-related stress”. Interventions for this subgroup should focus on “alleviating basic burdens and addressing core needs”. For instance, providing respite care for caregivers and delivering training on efficient caregiving skills. When family function scores exceed 140 points, the psychological saturation effect posits that psychological benefits tend to diminish once the intensity or frequency of stimulation surpasses an individual’s adaptive capacity [37]. At this level, caregivers’ brain sensitivity to supportive stimuli gradually decreases [38]; consequently, additional support provides reduced emotional benefits, slowing the alleviation of psychological distress. At this stage, caregivers’ basic needs are already met, so simply enhancing family function has a limited impact on reducing psychological distress. Therefore, intervention priorities should shift from “alleviating basic burdens” to “promoting well-being and self-actualization,” such as establishing peer support groups, offering mindfulness training, or providing stress management interventions. In short, future practices should integrate diverse resources to build a multi-level support system, thereby maximizing resource efficiency and optimizing caregivers’ mental health outcomes. Limitations Firstly, the study adopts a cross-sectional design, which can only identify correlations and threshold characteristics between family function and psychological distress, but cannot establish a causal or temporal relationship between the two. Secondly, the subjects were recruited exclusively from three tertiary hospitals in Wuhan, Hubei Province, resulting in a relatively concentrated sample sources. Future studies should aim to validate the generalizability of these findings by using a multi-center, large-sample design that includes a broader population of informal caregivers of PwD. Thirdly, the study relied solely on questionnaire surveys to collect data on family function and psychological distress. These self-reported measures may be subject to social desirability bias (such as caregivers under-reporting severe psychological distress or overestimating family function). Future research could incorporate objective indicators to enhance the validity of the findings. Conclusion This study analyzed the correlation between family function and psychological distress among informal caregivers of PwD. It not only verified the linear relationship between these variables but also further elucidated the threshold effect of family function on psychological distress. The findings provide an empirical basis for developing personalized psychological intervention strategies for caregivers at the family level in the future.