Care burden and outcomes in young carers during and after the COVID‑19 pandemic: psychological distress and cognitive–emotional aspects

Abstract

Global social and policy concerns regarding young carers’ care burden are increasing, but the outcomes of their caring responsibilities, especially in Asian countries, remain inadequately explored. We empirically examined the associations of care burden with psychological distress and cognitive–emotional aspects using cross‑sectional online surveys of young carers aged 15–19 living throughout Japan in 2021 (during the COVID‑19 pandemic) and 2024 (post‑pandemic). Each survey sampled a different group of young carers. The Kessler Psychological Distress Scale and Positive and Negative Outcomes of Caring were used to assess outcomes, which were analyzed in relation to care complexity and intensity. The mean psychological distress score was 7.33, with significantly higher scores observed during the pandemic compared with 2024. However, both positive and negative scores on the cognitive and emotional outcomes instrument were significantly elevated post-pandemic. We observed a heavy care burden, associated with worsened psychological distress. Regarding cognitive–emotional aspects, greater care task complexity was associated with an increase in both negative and positive responses. Policies should offer tailored support for young carers, especially during emergencies, ensure that they receive mental health support, tackle issues associated with young carers’ isolation, and deliver accessible and youth-friendly social resources.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-22652-5.

Introduction

The circumstances surrounding young carers have been increasingly capturing the attention of researchers and policymakers worldwide^1,2. “Young carers” can be broadly defined as children or adolescents who provide ongoing care and assume responsibility for family members (e.g., for grandparents, parents, or siblings)^3–5. Although caregiving has long been regarded as the purview of adults, a growing body of literature has revealed how young carers assume significant responsibilities that place them at risk of adverse outcomes in various life domains^6,7.

Indeed, the care burden borne by young carers encompasses various responsibilities (e.g., personal care, household chores, emotional support, and financial contributions)^5,8, which in turn lead to caregiving demands that have been associated with reduced academic achievement, increased stress levels, and significant educational disadvantages^9–11. Researchers have comprehensively delved into the impact of such care burden on the mental, cognitive, and emotional aspects of young carers. For instance, a systematic review reported the existence of various studies on young carers’ levels of depression, anxiety, and stress, as well as disruptions in their cognitive development and emotion regulation¹².

However, the experience of being a young carer does not have exclusively negative consequences. Instead, studies have identified the potential for various positive outcomes arising from these responsibilities, including the development of improved coping mechanisms and problem-solving skills, accelerated progression toward maturity, cultivation of empathy and compassion, strengthened resilience in the face of adversity, and a greater sense of autonomy and self-reliance^13–15. The mixed evidence on the consequences of being a young carer warrants a balanced and comprehensive understanding of the multifaceted impacts of caregiving on adolescents.

The extant support systems for young carers vary considerably across countries. In some developed countries, including the United Kingdom, Australia, and Canada, policy initiatives have led to the development of respite care, counseling services, and educational accommodations for young carers^16,17. However, implementation remains uneven even in developed countries, with some such as the United States and Japan still relying on inadequate informal networks that cannot comprehensively handle the complex needs of young carers^16,17. These disparities highlight both the progress made and significant gaps in support across different national contexts.

Despite the growing body of literature, significant gaps persist. First, there is a notable scarcity of empirical research on young carers in Asian countries^18,19, where awareness of young carers as a distinct group has emerged only recently^16,17. In Japan, empirical investigations into care burden outcomes also remain limited^20,21. Second, international global public health crises such as the COVID-19 pandemic often amplify the challenges faced by vulnerable groups, including young carers²². As the pandemic brought about unprecedented disruptions (e.g., school closures and social distancing), which impacted young carers’ normal routines, their access to external support from schools, friends, and formal services was also limited^23–25. The literature has focused on the well-being of young carers during COVID-19²⁶, but their situation during the post-pandemic period should also be examined to allow for future comparisons. Our study addresses these gaps by providing empirical evidence useful for future international comparisons and policy implications regarding social support measures for young carers during both emergency and peaceful times.

Specifically, we aimed to investigate the psychological distress and cognitive and emotional aspects of young carers along with their relationships with care burden. We hypothesized that the care burden associations are as follows. First, caring task complexity and intensity are associated with worsened psychological distress and cognitive and emotional aspects. Second, care skill development and care recipients’ provision of emotional feedback are associated with positive emotions (e.g., satisfaction).

We used a unique individual dataset of Japanese young carers collected in 2021 and 2024 to empirically examine the aforementioned hypotheses. We chose these two periods for the following reasons. First, in Japan, coinciding with the outbreak of the COVID-19 pandemic, the social issues surrounding young carers are just beginning to be recognized, with the first initiatives at the governmental and academic levels in 2021^17,20,21, when we conducted our first survey. Second, in May 2023, the Japanese government officially reclassified COVID-19 from a special category (equivalent to Class 2) to a Class 5 infectious disease under the Infectious Diseases Control Act²⁷. After this transition, restrictions on people’s movement were no longer imposed and support policies such as subsidies for COVID-19 treatment drugs were abolished; since then, COVID-19 has been handled under the normal medical system. In light of the above social circumstances, our 2021 survey captures the situation of both pandemic-related restrictions and limited societal recognition of young carers, whereas the 2024 survey captures the post-pandemic recovery and heightened societal awareness of young carers. We use the individual data collected in these two periods to assess how shifting societal awareness and public health crises jointly shape the psychological and cognitive–emotional outcomes of young carers in Japan.

Methods

Data

We conducted online surveys with young carers in 2021 and 2024 in collaboration with JMAR, a Japanese research institute. The survey targeted young carers aged 15–19 years. Regarding participant selection, an initial screening survey was administered to a sample of those aged 15–19 years registered as monitors with JMAR. This initial group was selected via systematic random sampling to mitigate potential biases based on gender or residence area. From this randomly screened pool, only the monitors who confirmed that they met the conditions for this study (i.e., being responsible for caring for cohabiting grandparents, parents, or siblings) were selected for the final analysis. In total, data of 1581 young carers (816 in 2021 and 765 in 2024) were collected. It is important to note that the survey in 2024 was not a follow-up of the 2021 survey but sampled a different group of young carers following the same sampling design.

Ethical considerations

This study was approved by the Research Ethics Committee of Osaka Prefecture University, Graduate School of Economics on January 14, 2021 and the Research Ethics Committee of Osaka Metropolitan University, Graduate School of Economics on February 22, 2024 (No. 2023-3). All procedures were performed in accordance with relevant guidelines and regulations, including the Declaration of Helsinki. All participants and/or their legal guardians provided written informed consent before enrollment. All personal identifiers have been removed from the manuscript and supplementary materials. No identifying images, videos, or clinical details of participants are included in this paper.

Dependent variables

Psychological distress

We used the Kessler Psychological Distress Scale (K6) to assess psychological distress in young carers^28–30. This six-item self-reported questionnaire was designed to measure non-specific psychological distress experienced during a period (e.g., typically in the past 30 days). Participants were asked to rate how frequently they experienced symptoms such as feeling “nervous,” “hopeless,” “restless or fidgety,” “so sad that nothing could cheer them up,” “that everything was an effort,” and “worthless.” Each item was rated on a 5-point Likert-type scale (0: None of the time; 4: All of the time), with total scores calculated by summing item scores (range: 0–24) and higher scores indicating higher levels of psychological distress. In this study, a score of 13 or higher represented severe psychological distress³¹, a score between 5 and 13 represented moderate psychological distress, and a score of 5 or lower represented mild psychological distress^32,33.

Cognitive and emotional aspects

We used the Positive and Negative Outcomes of Caring (PANOC-YC20) scale to capture young carers’ cognitive and emotional aspects¹⁴. This instrument was developed in a study conducted in the United Kingdom³⁴ and has been applied in many studies on care outcomes^14,35,36. Based on this instrument, we generated three dependent variables: positive response scores, negative response scores, and a dummy variable indicating whether a young carer faces a high risk regarding cognitive and emotional aspects because of care burden. Supplementary Table A1 shows details of the items of the PANOC-YC20.

The 20 items of the instrument were rated on a 3-point scale (0: Never; 1: Some of the time; 2: A lot of the time) and divided into the two separate 10-item subscales of positive response (items 1, 2, 3, 4, 7, 8, 15, 18, 19, and 20) and negative response (items 5, 6, 9, 10, 11, 12, 13, 14, 16, and 17). Scores for each subscale ranged from 0 to 20, with higher scores corresponding to more pronounced positive/negative effects of caring activities. There is potential for concern when the score is less than 12 on the positive response subscale and concomitantly greater than 8 on the negative response subscale³⁴. Participants with these scores for each subscale were defined as the high-risk group, and we used a dummy variable scored 1 when the participant was in the high-risk group and 0 when it was not applicable.

Independent variable

Care burden

Following the methodology in a prior study²⁰, three measures were used to capture young carers’ care burden. We used “the number of care types provided” to capture care burden complexity and “the overall percentage of care burden” and “primary carer status” to capture care burden intensity.

First, the number of care types provided was measured by asking the question, “Do you provide the following types of care for your family members?” Participants were presented with some care type options regarding direct personal care (e.g., assistance with hygiene, mobility, and medication management) and indirect support tasks (e.g., managing household chores and organizing medical appointments).

Second, we assessed care burden as a percentage, reflecting how much of the young carers’ daily time and effort was devoted to caring compared with other caregivers in the family. This was captured by asking the question, “Regarding the role as a caregiver in your family, what percentage of the care burden does each caregiver bear (the overall ratio includes time, money, and effort spent on caregiving)?”. Young carers stated the percentage of care provided by themselves and by all other caregivers in the same family (the ratio of each caregiver was added up to 100%). The overall percentage of care burden of young carers ranged from 0 to 100, representing the intensity of their care burden.

Third, we assessed primary carer status, reflecting whether the participant was the primary carer in the family. This was based on the question, “Who in your household is primarily responsible for caring for your family members?” Participants were presented with the choices of “you” (young carers themselves) and other family members including grandfather, grandmother, father, mother, brother, sister, younger brother, younger sister and others.

Other covariates

Based on a review of some previous studies on young carers’ care burden^20,36,37, we first controlled for demographic variables including age and gender. We used the young carers’ actual age as the age variable. Regarding the definition of the gender variable, we provided three options to the young carers—boy, girl or others—based on a European study³⁶. However, in our sample, the proportion in the category of “others” was extremely small, accounting for only 0.89% of the total respondents in 2021 and numbering zero in 2024. Therefore, in our estimation, we only set a female dummy variable for analyzing gender differences (“girl” = 1, “boy” and “others” = 0), while considering that women are more involved in family caregiving among Asian cultures^38,39.

Second, we also controlled covariates including friendships of care recipients and main income source in the family³⁷. Regarding the friendships of care recipients, we asked young carers the following question: “Does the primary care recipient in your family have friends they keep in touch with?” The four response options were “many,” “only one,” “no one,” and “I don’t know.” We generated a friendship dummy variable by setting the response of “many” to 1 and all other responses to 0, considering care recipients may receive support (mental, emotional, network, information, etc.) outside the family, which may also impact the mental well-being of young carers. Regarding family income sources, we asked young carers about their households’ main source of income with seven response options including their parent(s), themselves, sibling(s), grandparent(s), public assistance received, other sources, and “I don’t know.” We generated a main income dummy variable by setting parents as the income source to 1 and other responses to 0, in order to capture family economic status. All study variables are summarized in Table 1.

Table 1.

Settings of variables.

Variables	Settings
Mental health outcomes
K6 scale score	K6
	Score range: 0–24
Psychological distress level based on K6	0: mild psychological distress
	1: moderate psychological distress
	2: severe psychological distress
Cognitive and emotional aspects
PANOC-YC20 positive scores	Positive response scores of the PANOC-YC20
	Score range: 0–20
PANOC-YC20 negative scores	Negative response scores of the PANOC-YC20
	Score range: 0–20
High-risk group	1: If applicable (less than 12 on the positive response scale and greater than 8 on the negative response scale)
	0: If not
Care burden
Number of care types provided	Types: 0–11
Overall percentage of care burden	Range: 0–100
Primary carer status	Being the primary carer in family or not
	1: If applicable 0: If not
Covariates
Age	Age range: 15–19
Gender (girl)	1: If applicable (girl)
	0: If not (boy and others)
Friendship	Care recipients have many friends
	1: If applicable 0: If not
Main source of income	Main source was the income of parent(s)
	1: If applicable 0: If not
COVID-19	During COVID-19 in 2021
	1: If applicable 0: If not (in 2024)

K6: Kessler psychological distress scale, PANOC-YC20: Positive and negative outcomes of caring scale¹⁴.

Estimation framework

Ordinary least squares model

Based on a review of studies examining the relationship between care burden and the K6 or PANOC-YC20^{28–30,36,40}, we first employed an ordinary least squares regression model as the baseline specification^41,42. The model is specified as shown herein:

1

where is the dependent variable representing young carer ’s psychological distress (K6) and/or cognitive and emotional aspects (positive/negative response subscale scores); is the independent variable representing young carer ’s care burden based on three measures (i.e., number of care types provided, overall percentage of care burden, and primary carer status); represents the covariates (i.e., age, gender, friendships, and family income source); represents the survey period; represents intercept variables; and represents unobserved random variables (errors).

Ordered probit model

As aforementioned, we divided psychological distress into three levels based on K6 scores (mild, moderate, and severe). Accordingly, we applied an ordered probit model to analyze the different associations at different levels of psychological distress^41,42. The model is specified as follows:

2

where is an ordered dependent variable representing psychological distress level based on K6 scores. The observed ordinal outcome was extracted through the following:

3

Moreover, the probability of obtaining a particular outcome was given by:

4

To assess the validity of the proportional odds (parallel regression) assumption underlying the ordered probit models, we conducted an approximate likelihood-ratio test of the equality of coefficients across response categories. We evaluated whether the estimated coefficients could be constrained to be equal across different cut-points of the ordinal outcome distribution. A non-significant result would indicate that the proportional odds assumption was not violated, and the use of the ordered probit specification was appropriate.

Probit model

A probit model was used to analyze whether a young carer could be included in the high-risk group based on PANOC-YC20 scores, with the estimation model being as follows:

5

where is a dummy variable for whether a young carer is at high risk and represents the cumulative distribution function of the standard normal distribution. Based on the estimated coefficient results, we also calculated the marginal effect (ME) results of the above models. For statistical analysis, Stata/SE version 18 software was used to estimate all models.

Results

Descriptive statistics

A comparison of descriptive statistics for the main variables during and after COVID-19 and their t-test/chi-square test results are summarized in Table 2. Regarding psychological distress, the K6 mean score was 7.33, and the scores during COVID-19 were significantly higher than those post-COVID-19. Figure 1 shows the distribution of K6 scores among young carers in 2021 and 2024. In 2021, 34.93%, 41.67%, and 23.41% of the respondents had mild, moderate, and severe psychological distress, respectively. In 2024, 44.58%, 38.69%, and 16.73% of the respondents had mild, moderate, and severe psychological distress, respectively. The distribution of young carers’ psychological distress levels differed significantly between the two survey periods.

Table 2.

Comparison of descriptive statistics between 2021 and 2024 survey samples.

Variables	Overall (N = 1581)	During COVID-19 in 2021 (N = 816)	Post COVID-19 in 2024 (N = 765)	t-test p-value		Chi-square test p-value
Mental health outcome
K6 score, mean (SD)	7.33 (6.45)	7.93 (6.48)	6.69 (6.37)	0.000	***
Psychological distress level (K6)						0.000	***
0: mild psychological distress, frequency (%)	626 (39.60)	285 (34.93)	341 (44.58)
1: moderate psychological distress, frequency (%)	636 (40.23)	340 (41.67)	296 (38.69)
2: severe psychological distress, frequency (%)	319 (20.18)	191 (23.41)	128 (16.73)
Cognitive and emotional aspects
PANOC-YC20 positive scores, mean (SD)	9.52 (5.91)	8.81 (5.95)	10.29 (5.78)	0.000	***
PANOC-YC20 negative scores, mean (SD)	5.68 (5.64)	5.38 (5.61)	6.00 (5.66)	0.029	**
High-risk group						0.765
1: If applicable, frequency (%)	299 (19.91)	152 (18.63)	147 (19.22)
0: If not, frequency (%)	1282 (81.09)	664 (81.37)	618 (80.78)
Care burden
Number of care types provided, mean (SD)	2.18 (1.59)	2.16 (1.53)	2.19 (1.65)	0.659
Overall percentage of care burden, mean (SD)	48.11 (38.56)	45.94 (39.10)	50.42 (37.67)	0.021	**
Primary carer status						0.379
1: If applicable, frequency (%)	586 (37.07)	294 (36.03)	292 (38.17)
0: If not, frequency (%)	995 (62.93)	522 (63.97)	473 (61.83)
Other characteristics
Age, mean (SD)	16.56 (1.15)	16.79 (1.13)	16.31 (1.11)	0.000	***
Gender						0.000	***
Girl, frequency (%)	1045 (66.10)	557 (68.26)	488 (63.79)
Boy, frequency (%)	522 (33.02)	245 (30.02)	277 (36.21)
Others, frequency (%)	14 (0.89)	14 (1.72)	0 (0.00)
Friendship of care recipient						0.232
1: If applicable, frequency (%)	783 (49.53)	416 (50.98)	367 (47.97)
0: If not, frequency (%)	798 (50.47)	400 (49.02)	398 (52.03)
Main source of income						0.187
1: Parent(s), frequency (%)	1300 (82.23)	681 (83.46)	619 (80.92)
0: Other sources, frequency (%)	281 (17.77)	135 (16.54)	146 (19.08)

***p < 0.01; **p < 0.05; *p < 0.1. SD: standard deviation, K6: Kessler psychological distress scale, PANOC-YC20: Positive and negative outcomes of caring scale¹⁴. Estimates are presented as mean (SD) or frequency (%). p-values were obtained using the Student’s t-test for continuous variables and the chi-square test for categorical variables.

Fig. 1.

Distribution of K6 scores among young carers (N = 1581) Note. K6: Kessler psychological distress scale.

Regarding cognitive and emotional aspects, the mean score for the positive and negative response subscales of the PANOC-YC20 was 9.52 and 5.68, respectively. The mean scores for both subscales were significantly different among the samples between the two survey periods. The distributions of scores and correlations between the positive and negative response scores are shown in Fig. 2. The correlation coefficient between the positive and negative response subscale scores among the total sample (N = 1581) was 0.321 (p < 0.01), indicating a slightly weak positive correlation. Positive response scores were normally distributed and showed a prominent peak in the middle range (around 10), indicating that many young carers showed relatively poor positive outcomes based on the evaluation criterion of high-risk groups (positive response scores less than 12)³⁴. Negative response scores were positively skewed and showed a prominent peak in the lower range (between 0 and 4), indicating that many young carers showed relatively poor negative outcomes. The proportion of young carers in the high-risk group was 19.91% among the total sample (N = 1581).

Fig. 2.

Heatmap of the distributions of scores for the positive and negative response subscales of the PANOC-YC20 (N = 1581). Note. ***p < 0.01. PANOC-YC20: Positive and negative outcomes of caring scale.

Regarding care burden variables, young carers provided more than two care types on average, and 37.07% were the primary carers in the family. Regarding overall care burden, the proportion was 48.11% on average, with a statistically significant difference between during and post-COVID-19. The average age of young carers was 16.56 years, and 66.10% were girls. In our sample, 49.53% of the care recipients had two or more friends, and 82.23% of the young carers reported their parent(s) as being the main income source.

Associations of care burden with K6 scores and psychological distress

The estimated associations of care burden with K6 scores and psychological distress are summarized in Table 3.

Table 3.

Associations of care burden with K6 scores and psychological distress.

	K6		Psychological distress
	Scores Coef (SE)		Mild level ME (SE)		Moderate level ME (SE)		Severe level ME (SE)
	OLS		Ordered probit		Ordered Probit		Ordered probit
Panel A: Associations with the number of care types provided
Number of care types provided	0.276	**	− 0.015	**	0.004	**	0.011	**
	(0.115)		(0.007)		(0.002)		(0.005)
Age	0.491	***	− 0.028	***	0.008	***	0.021	***
	(0.141)		(0.009)		(0.003)		(0.007)
Gender (girl)	0.695	**	− 0.329		0.009		0.024
	(0.334)		(0.023)		(0.006)		(0.017)
Friendships of care recipients	− 0.797	**	0.055	**	− 0.015	**	− 0.041	**
	(0.323)		(0.022)		(0.006)		(0.016)
Main income source	− 0.571		0.031		− 0.008		− 0.021
	(0.447)		(0.028)		(0.008)		(0.021)
COVID-19	1.022	***	− 0.082	***	0.022	***	0.060	***
	(0.328)		(0.022)		(0.006)		(0.016)
Constant	− 1.531
	(2.385)
N	1581		1581		1581		1581
Parallel regression assumption test			chi2(6) = 4.53 Prob > chi2 = 0.6052
Panel B: Associations with the overall percentage of care burden
Overall percentage of care burden	0.014	***	− 0.0008	***	0.0002	***	0.0006	***
	(0.004)		(0.0003)		(0.0001)		(0.0002)
Age	0.427	***	− 0.025	**	0.007	**	0.018	***
	(0.142)		(0.009)		(0.003)		(0.007)
Gender (Girl)	0.757	**	− 0.037		0.010		0.027
	(0.332)		(0.023)		(0.006)		(0.017)
Friendships of care recipients	− 0.777	**	0.055	**	− 0.015	**	− 0.040	**
	(0.323)		(0.022)		(0.006)		(0.016)
Main income source	− 0.424		0.020		− 0.005		− 0.015
	(0.450)		(0.028)		(0.008)		(0.021)
COVID-19	1.097	***	− 0.087	***	0.023	***	0.063	***
	(0.329)		(0.022)		(0.006)		(0.016)
Constant	− 0.742
	(2.357)
N	1581		1581		1581		1581
Parallel regression assumption test			chi2(6) = 3.85 Prob > chi2 = 0.6964
Panel C: Associations with the primary carer status
Primary carer status	1.194	***	− 0.079	***	0.021	***	0.058	***
	(0.347)		(0.022)		(0.006)		(0.016)
Age	0.422	***	− 0.024	**	0.006	**	0.017	**
	(0.142)		(0.009)		(0.003)		(0.007)
Gender (girl)	0.782	**	− 0.039	*	0.011	*	0.028	*
	(0.331)		(0.023)		(0.006)		(0.017)
Friendships of care recipients	− 0.789	**	0.056	**	− 0.015	**	− 0.041	**
	(0.323)		(0.021)		(0.006)		(0.016)
Main income source	− 0.390		0.017		− 0.004		− 0.012
	(0.451)		(0.028)		(0.008)		(0.021)
COVID-19	1.062	***	− 0.085	***	0.023	***	0.062	***
	(0.328)		(0.022)		(0.006)		(0.016)
Constant	− 0.453
	(2.361)
N	1581		1581		1581		1581
Parallel regression assumption test			chi2(6) = 4.21 Prob > chi2 = 0.6476

***p < 0.01; **p < 0.05; *p < 0.1. SE: standard error, OLS: ordinary least squares, ME: marginal effect, K6: Kessler psychological distress scale. Friendships of care recipients: whether they had more than two friends they kept in touch with; main income source: the main income source of the household was the young carers’ parent(s).

Estimation results from both ordinary least squares and ordered probit models consistently indicated that more complex and intensive care burden was associated with greater psychological distress among young carers. The number of care types was positively associated with K6 scores (coef. = 0.276, p < 0.05). As types of care increased, the probabilities of having moderate (ME = 0.004, p < 0.05) and severe psychological distress (ME = 0.011, p < 0.05) also increased. Similarly, a high level of overall percentage of care burden was linked to elevated K6 scores (coef. = 0.014, p < 0.01) and higher risks of having moderate (ME = 0.0002, p < 0.01) and severe distress (ME = 0.0006, p < 0.01). Being the primary carer in the family was significantly associated with higher K6 scores (coef. = 1.194, p < 0.01) and substantially higher probabilities of having moderate (ME = 0.021, p < 0.01) and severe distress (ME = 0.058, p < 0.01).

Among covariates, older respondents (age: coef. = 0.422 to 0.491, p < 0.01) and girls (coef. = 0.695 to 0.782, p < 0.05) consistently reported greater psychological distress. Care recipients with more friends were more likely to relieve young carers’ distress (coef. =  − 0.797 to − 0.777, p < 0.05; ME for mild distress = 0.055 to 0.056, p < 0.05). The pandemic was strongly associated with psychological distress among young carers; it was associated with increased K6 scores (coef. = 1.022 to 1.097, p < 0.01) and higher levels of moderate and severe distress.

In order to examine the robustness of our results, we re-estimated the models by introducing the interaction terms between care burden indicators and the COVID-19 period dummy variable as shown in Table 4. Under both ordinary least squares and ordered probit specifications, the interaction terms were generally not statistically significant, indicating that the associations between care burden and psychological distress were broadly stable across the pandemic and post-pandemic periods. At the same time, the coefficients for the COVID-19 dummy itself remained positive and significant in most specifications (as shown in Table 3), indicating an overall increase in K6 score and psychological distress levels during the pandemic, regardless of care burden status. Moreover, from the joint significance tests of each care burden indicator and its interaction with the COVID-19 period, we observed significant results in the case of the number of care types provided, and even more significance for the overall percentage of care burden and the primary carer status. Based the above robustness check, while the moderating role of the COVID-19 dummy variable was limited, the associations between care burden and psychological distress (both score and level) remained consistent and robust across alternative specifications.

Table 4.

Robustness check of the associations of care burden with K6 scores and psychological distress.

	K6 Scores Coef (SE)		Psychological distress level Coef (SE)		K6 Scores Coef (SE)		Psychological distress level Coef (SE)		K6 Scores Coef (SE)		Psychological distress level Coef (SE)
	OLS		Ordered probit		OLS		Ordered probit		OLS		Ordered probit
Number of care types provided	0.265		0.039
	(0.170)		(0.027)
Number of care types provided × COVID-19	0.023		0.004
	(0.230)		(0.037)
Overall percentage of care burden					0.013	**	0.002	**
					(0.006)		(0.001)
Overall percentage of care burden × COVID-19					0.001		− 0.00004
					(0.009)		(0.001)
Primary carer status									1.240	**	0.206	**
									(0.488)		(0.085)
Primary carer status × COVID-19									− 0.090		0.005
									(0.679)		(0.117)
COVID-19	0.971	*	0.209	**	1.050	**	0.232	**	1.095	***	0.223	***
	(0.574)		(0.098)		(0.505)		(0.093)		(0.398)		(0.073)
Joint significance test	Prob > F = 0.052		Prob > chi2 = 0.084		Prob > F = 0.007		Prob > chi2 = 0.014		Prob > F = 0.003		Prob > chi2 = 0.002

***p < 0.01; **p < 0.05; *p < 0.1. K6: Kessler psychological distress scale, SE: standard error, OLS: ordinary least squares. All models include the same set of covariates as in Table 3. For brevity, coefficients of control variables are not reported. Joint significance test: p-values were obtained using the F-test for OLS models and the chi-square test for ordered probit models.

Associations of care burden with PANOC-YC20 scores and high-risk group classification

The estimated associations of care burden with PANOC-YC20 scores and the high-risk group are summarized in Table 5.

Table 5.

Associations of care burden with PANOC-YC20 scores and the high-risk group classification.

	Panel A: Association with the number of care types provided						Panel B: Association with the overall percentage of care burden						Panel C: Association with the primary carer status
	Positive response scores Coef (SE)		Negative response scores Coef (SE)		High-risk group ME (SE)		Positive response scores Coef (SE)		Negative response scores Coef (SE)		High-risk group ME (SE)		Positive response scores Coef (SE)		Negative response scores Coef (SE)		High-risk group ME (SE)
	OLS		OLS		Probit		OLS		OLS		Probit		OLS		OLS		Probit
Number of care types provided	0.520	***	0.544	***	0.011	*
	(0.091)		(0.094)		(0.006)
Overall percentage of care burden							0.002		0.020	***	0.0003
							(0.004)		(0.004)		(0.0002)
Primary carer status													0.171		1.481	***	0.024
													(0.311)		(0.306)		(0.020)
Age	− 0.130		0.171		0.010		− 0.160		0.074		0.008		− 0.159		0.077		0.008
	(0.130)		(0.123)		(0.009)		(0.133)		(0.125)		(0.009)		(0.133)		(0.126)		(0.009)
Gender (girl)	− 0.376		− 0.332		0.013		− 0.328		− 0.231		0.015		− 0.325		− 0.208		0.016
	(0.307)		(0.299)		(0.021)		(0.314)		(0.302)		(0.021)		(0.314)		(0.302)		(0.021)
Friendships of care recipients	1.608	***	− 0.179		− 0.034	*	1.649	***	− 0.139		− 0.033	*	1.648	***	− 0.154		− 0.034	*
	(0.292)		(0.282)		(0.020)		(0.295)		(0.283)		(0.020)		(0.294)		(0.283)		(0.020)
Main income source	− 0.768	**	− 1.378	***	− 0.066	***	− 0.787	**	− 1.180	***	− 0.063	**	− 0.786	**	− 1.171	***	− 0.063	**
	(0.378)		(0.378)		(0.025)		(0.384)		(0.380)		(0.025)		(0.383)		(0.379)		(0.025)
COVID-19	− 1.412	***	− 0.630	**	− 0.008		− 1.409	***	− 0.525	*	− 0.006		− 1.416	***	− 0.583	**	− 0.007
	(0.294)		(0.284)		(0.020)		(0.298)		(0.285)		(0.020)		(0.297)		(0.285)		(0.020)
Constant	11.365	***	3.426				12.833	***	4.976	**			12.874	***	5.333	**
	(2.199)		(2.089)				(2.217)		(2.090)				(2.221)		(2.092)
N	1581		1581		1581		1581		1581		1581		1581		1581		1581

***p < 0.01; **p < 0.05; *p < 0.1. PANOC-YC20: Positive and negative outcomes of caring scale; friendships of care recipients: whether they had more than two friends they kept in touch with; main income source: the main income source of the household was young carers’ parent(s). SE: standard error, ME: marginal effect, OLS: ordinary least squares.

As evident from Table 5, as the number of care types provided increased, both positive response scores (coef. = 0.520, p < 0.01) and negative response scores (coef. = 0.544, p < 0.01) increased, and the probability of falling into the high-risk group (ME = 0.011, p < 0.1) also increased modestly. However, in the case of the overall percentage of care burden (coef. = 0.020, p < 0.01) and being the primary carer (coef. = 1.481, p < 0.01), we only observed positive associations between care burden and the negative response scores.

Among covariates, care recipients having more than two friends was consistently associated with higher positive response scores (coef. = 1.608 to 1.649, p < 0.01) and a lower likelihood of high-risk classification (ME =  − 0.034 to − 0.033, p < 0.1). The main income source (parent(s)’s income) was negatively related to both positive (coef. =  − 0.787 to − 0.768, p < 0.05) and negative response scores (coef. =  − 1.378 to  − 1.171, p < 0.01), and it was also associated with a reduced probability of falling into the high-risk group (ME =  − 0.066 to  − 0.063, p < 0.05). During the pandemic in 2021, both positive response scores (coef. =  − 1.416 to  − 1.409, p < 0.001) and negative response scores (coef. =  − 0.630 to  − 0.525, p < 0.1) decreased significantly.

For the robustness check, we re-estimated the models by introducing interaction terms between care burden indicators and the COVID-19 period dummy as shown in Table 6. The interaction between the number of care types provided and COVID-19 was statistically significant for both positive and negative response scores, indicating that the associations between providing more types of care and caregiving experiences were amplified during the pandemic. By contrast, interactions for the overall percentage of care burden × COVID-19 and primary carer status × COVID-19 were not significant, suggesting that their effects did not vary systematically between during the pandemic and post-pandemic periods. Joint significance tests confirmed that the combined effect of caregiving burden measures and their interactions with pandemic was consistently significant for positive and negative responses, but not for high-risk group membership.

Table 6.

Robustness check of the associations of care burden with PANOC-YC20 scores and high-risk group classification.

	Panel A: Association with the number of care types provided						Panel B: Association with the overall percentage of care burden						Panel C: Association with the primary carer status
	Positive response score Coef (SE)		Negative response score Coef (SE)		High-risk group ME (SE)		Positive response score Coef (SE)		Negative response score Coef (SE)		High-risk group ME (SE)		Positive response score Coef (SE)		Negative response score Coef (SE)		High-risk group ME (SE)
	OLS		OLS		Probit		OLS		OLS		Probit		OLS		OLS		Probit
Number of care types provided	0.372	***	0.288	**	0.044
	(0.126)		(0.129)		(0.030)
Number of care types provided × COVID-19	0.308	*	0.534	***	− 0.004
	(0.180)		(0.188)		(0.044)
Overall percentage of care burden							0.007		0.022	***	0.000
							(0.006)		(0.006)		(0.001)
Overall percentage of care burden × COVID-19							− 0.008		− 0.005		0.002
							(0.008)		(0.008)		(0.002)
Primary carer status													0.468		1.592	***	0.016
													(0.432)		(0.436)		(0.109)
Primary carer status × COVID-19													− 0.578		− 0.216		0.143
													(0.611)		(0.604)		(0.150)
COVID-19	− 2.080	***	− 1.789	***	− 0.021		− 1.005	**	− 0.306		− 0.124		− 1.202	***	− 0.504		− 0.083
	(0.492)		(0.478)		(0.123)		(0.477)		(0.431)		(0.120)		(0.369)		(0.337)		(0.095)
Joint significance test	Prob > F = 0.000		Prob > F = 0.000		Prob > chi2 = 0.151		Prob > F = 0.459		Prob > F = 0.000		Prob > chi2 = 0.236		Prob > F = 0.537		Prob > F = 0.000		Prob > chi2 = 0.315

***p < 0.01; **p < 0.05; *p < 0.1. PANOC-YC20: Positive and negative outcomes of caring scale, SE: standard error, ME: marginal effect; OLS: ordinary least squares. All models include the same set of covariates as in Table 3. For brevity, coefficients of control variables are not reported. Joint significance test: p-values were obtained using the F-test for OLS models and the chi-square test for probit models.

Discussion

This study revealed that care burden was associated with psychological distress among young carers. Furthermore, while caregiving presented challenges in relation to carers’ cognitive and emotional aspects by exposing them to negative experiences, the related tasks also allowed them to have positive experiences associated with caregiving.

First, care task variety (complexity), overall percentage of care, and being the primary carer (intensity) were associated with amplified psychological distress. These findings mirror those of a recent systematic review, wherein nine out of 10 quantitative studies reported significant associations between informal caring and elevated depression and anxiety in youth¹². The caregiving roles of young carers have been shown to be negatively correlated with life satisfaction^43,44, while care time negatively impacts mental health^18,45. The need to engage in many care task types complicates caregiving processes, potentially causing young carers to feel overwhelmed and consequently affecting their mental health. Moreover, an excessive care burden is highly likely to infringe upon the basic social roles expected of young carers (e.g., study time and duration and socialization).

Our empirical findings can be situated within the stress process theory⁴⁶, which conceptualizes caregiving as involving both primary stressors (e.g., care demands and time intensity) and secondary stressors (e.g., role conflict, loss of social opportunities)⁴⁷. Our results among Japanese young carers align with this theoretical framework: as stressors accumulate (care burden increases), psychological distress increases, particularly among female caregivers and older adolescents, who may face heightened vulnerability pathways^36,37,48.

Second, while care task complexity was associated with increased negative responses regarding cognitive and emotional aspects, it was in parallel positively associated with positive responses. Compared with a previous study in European countries that reported a negative association between care burden and care outcome (only negative responses increased as care time increased)³⁶, we observed a unique duality of care outcomes within the Japanese context (both negative and positive responses increased as care burden increased). This divergence can be explained by cultural specificities in East Asian countries. For instance, the longstanding influence of filial piety places a moral emphasis on intergenerational caregiving as a virtue and social expectation^49,50. This dual pattern of care outcomes can also be understood through role enhancement theory^51,52—assuming caregiving roles may involve performing complex care tasks that allows young carers to learn skills that their peers do not have, making them feel proud and improving their positive cognition and feedback despite added strain¹⁵.

Third, comparisons across two survey waves revealed that psychological distress declined after the pandemic, yet both positive and negative responses to care responsibilities increased. During the lockdown early in the pandemic, isolation limited both social networks and external support, which magnified distress⁵³. In the post-pandemic period, the return to school and the reinstatement of peer networks and community services reduced overall distress⁵⁴. However, the rise in negative responses to care responsibilities even in the later stages of the pandemic calls for continued social support for young carers.

At the same time, with the gradual deepening of Japanese society’s awareness of young carers since 2021¹⁷, the positive feedback from society and the increase in social support have also increased young carers’ affirmation of the caregiving roles they undertake and the sense of growth that comes with it⁵⁵. Our findings thus illustrate how public health crises and evolving social recognition jointly shape young carers’ caregiving experiences.

Fourth, our findings have several implications for practice and policy. First, support measures must recognize the dual nature of caregiving among young carers: while reducing excessive role demands that undermine well-being, interventions should also harness the opportunities for skill development and resilience that caregiving can foster. Tailored and tiered approaches are therefore needed. Low-intensity carers may benefit from light-touch guidance and peer support, whereas those facing high-intensity or complex responsibilities require comprehensive interventions, including mental health services, respite opportunities, and academic accommodation. Second, our empirical evidence highlights the necessity of crisis preparedness. During emergencies such as pandemics, policies should prioritize continuity of access to mental health care, peer contact, and formal support services to mitigate isolation and role overload. Finally, the growing recognition of young carers since 2021 shows that national awareness can shift quickly¹⁷, with tangible effects on their daily lives. Positioning Japan within the broader international community, our study suggests that strengthening recognition and tailoring interventions to local cultural contexts are essential to reducing risks while enhancing the potential positive experience of caregiving.

Limitations

This study has several limitations. First, we calculated care burden complexity simply by summarizing the number of care types provided, despite each type of care carrying distinct challenges and potentially contributing differently to the overall burden. Future studies could develop indicators that can capture the complexity of each care type from multiple angles.

Second, although we collected data at two time points (during and post-COVID-19), the design remained essentially cross-sectional for each wave. Consequently, we could not establish definitive causal relationships between care burden and the outcomes of interest. Future research employing true longitudinal or panel designs could trace individual trajectories and test causal pathways.

Third, our nationally representative samples provided a unique window into the caregiving experiences of young carers in Japan, within an East Asian cultural context. Cross-cultural studies are needed to assess which patterns observed in Japan are universal and which are culturally specific, thereby informing the transferability of our findings to other social and cultural contexts.

Conclusion

We aimed to investigate how informal caregiving and the associated burden are related to young carers’ psychological distress and cognitive and emotional aspects, and whether these relationships shifted between during and after the COVID-19 pandemic. Care intensity (i.e., overall percentage of care burden and primary carer status) was positively associated with the K6 score, psychological distress level, and scores for the negative response subscale of the PANOC-YC20. Care complexity (i.e., number of care types) showed the same trend as care intensity but had a dual result, having a positive relationship with the positive response subscale of the PANOC-YC20. Covariates such as age, gender, friendships of care recipient, and main income source of family also mattered in relation to young carers’ psychological distress and cognitive and emotional aspects.

By situating our analysis in Japan, where societal recognition of young carers has only recently gained momentum¹⁷, we contribute culturally specific evidence to the global literature. These findings not only underscore the importance of targeted interventions but also open avenues for future research. As one direction, future studies should focus on the actual needs for social support among young carers and examine the factors that affect their needs. Further comparative research across countries will also be critical for identifying which aspects of caregiving experiences are universal and which are context-dependent, thereby informing policies that both alleviate care burden and effectively improve understanding and social recognition.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: Z.W. and B.N.; Methodology: Z.W. and B.N.; Software: Z.W.; Validation: B.N.; Formal Analysis: Z.W.; Data Curation: B.N.; Writing—Original Draft: Z.W. and B.N.; Writing—Review and Editing: Z.W. and B.N.; Visualization: Z.W.; Supervision: B.N.; Project Administration: B.N.; Funding Acquisition: B.N. All authors have read and agreed to the published version of the manuscript.

Data availability

Data supporting the findings of this study are available from Bing Niu upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.