Online Health-Seeking Behavior Among Primary Caregivers of Children with Chronic Diseases: A Cross-Sectional Analysis of eHealth Literacy, Caregiver Characteristics, and Child Clinical Context

Abstract

Background

Primary caregivers of children with chronic diseases frequently use online resources to meet ongoing information needs, but the correlates and context of this behavior remain incompletely understood. This study assessed online health-seeking behavior (OHSB) and examined its associations with child clinical characteristics, caregiver sociodemographic factors, and caregiver psychosocial and cognitive characteristics.

Methods

A single-center cross-sectional study using convenience sampling was conducted at the First Affiliated Hospital of Xinjiang Medical University, China, from October 2025 to February 2026. A total of 365 primary caregivers completed an electronic questionnaire including the OHSB Scale, eHealth Literacy Scale (eHEALS), Zarit Burden Interview, Family Caregiver Task Inventory, and GAD-7. Confirmatory factor analysis, univariate analyses, Pearson correlations, and hierarchical multiple linear regression were performed. HC3 robust standard errors and an expanded covariate model were used in sensitivity analyses.

Results

The three-factor OHSB structure showed good fit (CFI = 0.960, RMSEA = 0.069). The mean OHSB score was 94.57 ± 44.91, and 71.8% of caregivers sought health information online at least weekly. OHSB was strongly correlated with eHealth literacy (r = 0.883, p < 0.001). The primary fully adjusted model explained 81.6% of OHSB variance; eHealth literacy was the strongest correlate (β = 0.735, p < 0.001). Higher education and urban residence also remained robustly associated with higher OHSB across sensitivity analyses, whereas associations for complications, caregiver burden, and anxiety were less stable. Expanded adjustment did not materially change the dominant association of eHealth literacy.

Conclusion

OHSB was common among caregivers of children with chronic diseases and was most strongly associated with eHealth literacy, alongside contextual and sociodemographic factors. Because OHSB and eHealth literacy are closely related self-reported constructs, the magnitude of their association should be interpreted cautiously. Targeted, literacy-sensitive digital health support may benefit caregivers with fewer educational or contextual resources.

Introduction

Children with chronic diseases often require long-term treatment, repeated follow-up, ongoing medication management, and sustained family involvement in daily care.¹ In pediatric chronic care, primary caregivers therefore play a central role in symptom monitoring, treatment implementation, health-related decision-making, and communication with health professionals.² These prolonged responsibilities may create substantial informational, emotional, and practical demands for families, and caregivers of children with chronic conditions frequently report poorer psychological well-being and greater caregiving burden than parents of healthy children.^2–4 Access to understandable and reliable health information is therefore an important component of ongoing family disease management.

Online health-seeking behavior (OHSB), referring to the active searching for, accessing, and using of health information through online channels, has become increasingly common among caregivers.⁵ Online resources may help caregivers acquire disease-related knowledge, prepare for clinical consultations, compare professional advice, and obtain experiential or emotional support from others facing similar situations.^5,6 Such needs may be particularly pronounced in families caring for children with chronic or medically complex conditions, for whom information demands often extend beyond individual clinical encounters.^2,7 Recent evidence also suggests that online information use can become closely integrated with ongoing pediatric care. In a Chinese pediatric cancer setting, Wang et al found that caregivers’ online health information-seeking behaviors were closely linked to the care pathway, illustrating how digital information seeking may accompany repeated decision-making and care transitions.⁸ However, online health information varies considerably in quality and credibility, and caregivers may encounter conflicting recommendations, information overload, or difficulty determining whether information is trustworthy and applicable to their child.^6,9

Caregivers’ engagement with online health information may be related to several aspects of their caregiving context. Previous studies have identified socioeconomic and demographic differences in digital health engagement and eHealth literacy among parents,^10,11 while characteristics of the child’s health status and care complexity may also influence information needs and health-seeking behavior.¹² Caregiver-related cognitive and psychological characteristics are similarly relevant. eHealth literacy reflects an individual’s perceived knowledge, confidence, and skills in locating, evaluating, and applying electronic health information, whereas OHSB reflects the behavioral use of online resources for information searching, supplementation of offline care, and support seeking.^10,13 Although closely related, these constructs represent different aspects of digital health engagement. In addition, caregiver burden, caregiving difficulty, stress, and anxiety may influence how caregivers respond to uncertainty and informational demands associated with chronic illness.^13,14 Taken together, existing evidence points to three broad domains potentially related to OHSB: the child’s clinical context, caregiver sociodemographic and contextual characteristics, and caregiver psychosocial and cognitive characteristics.

Despite increasing interest in parental use of digital health information, current evidence remains fragmented.¹⁵ Much of the literature has focused on general parent populations, individual disease groups, or qualitative descriptions of internet use.^15–17 Studies of parents and caregivers of children with long-term conditions show that online resources are used for both information and support,¹⁵ and disease-specific research has demonstrated that online information seeking may become closely integrated with ongoing care and decision-making.⁸ However, relatively few studies have examined OHSB among primary caregivers of children with a broad range of chronic diseases while simultaneously considering child clinical characteristics, caregiver sociodemographic context, and caregiver psychosocial and cognitive factors. In addition, because the Chinese version of the OHSB Scale has only recently been validated,¹⁸ further evaluation of its proposed three-factor structure in caregivers of children with chronic diseases is warranted.

Therefore, this study aimed to characterize OHSB among primary caregivers of children with chronic diseases and to examine its associations with child clinical characteristics, caregiver sociodemographic and contextual characteristics, and caregiver psychosocial and cognitive factors, including eHealth literacy, caregiver burden, caregiving difficulty, and anxiety. We also evaluated the three-factor structure of the Chinese OHSB Scale in this population. A clearer understanding of these relationships may help inform more targeted and context-appropriate digital health support for families managing pediatric chronic diseases.

Methods

Study Design and Setting

This was a single-center, cross-sectional study conducted at the First Affiliated Hospital of Xinjiang Medical University, China, between October 2025 and February 2026. Data were collected using an electronic questionnaire. The study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Xinjiang Medical University (Approval No. K202510-39). All procedures were conducted in accordance with the Declaration of Helsinki.¹⁹ Electronic informed consent was obtained from all participants before questionnaire completion.

Participants

The target population comprised primary caregivers of children with chronic diseases. Participants were recruited using convenience sampling in pediatric-related clinical settings of the study hospital during the study period.

A primary caregiver was defined as the family member who assumed the main responsibility for the child’s daily care and disease management.

The inclusion criteria were as follows: (1) being the primary caregiver of a child (aged 0–18 years) diagnosed with a chronic disease in accordance with the corresponding clinical diagnostic guidelines (chronic disease status was determined by diagnosis rather than illness duration);²⁰ (2) being aged 18 years or older; (3) being able to read and complete the electronic questionnaire independently or with minimal assistance; and (4) providing informed consent to participate.

The exclusion criteria were as follows: (1) the primary caregiver having a cognitive impairment or mental disorder that prevented completion of the survey; (2) duplicate questionnaire submission; and (3) questionnaires with substantial missing data or obviously inconsistent responses.

No a priori sample-size calculation was performed. The sample size was determined by the number of eligible primary caregivers who were available and agreed to participate during the predefined recruitment period. A total of 380 questionnaires were distributed. After excluding 15 invalid questionnaires, 365 valid questionnaires were included in the final analysis.

Data Collection and Variables

Data were collected through an electronic questionnaire platform. The questionnaire consisted of a general information form and standardized scales assessing online health-seeking behavior, eHealth literacy, caregiver burden, caregiving difficulty, and anxiety.

The general information form included child-related and caregiver-related variables. Child-related variables included sex, age, medical diagnosis, disease duration, current treatment modality, number of hospitalizations/emergency department visits in the past year, long-term medication status, comorbidities, complications, and care setting. Caregiver-related variables included relationship to the child, sex, age, educational level, occupation, monthly household income, number of children in the family, place of residence, preferred online channels for health information, and weekly frequency of online health-seeking behavior. In the raw dataset, medical diagnoses were recorded as free-text entries and were subsequently regrouped into broad clinical categories for statistical analysis.

Measures

Online Health-Seeking Behavior Scale (OHSB)

Online health-seeking behavior was assessed using the Online Health-Seeking Behaviour Scale, originally developed by Popovac and colleagues (2022)²¹ and translated into Chinese by Peng (2025).¹⁸ The scale consists of 37 items across three dimensions: information search (10 items), supplement or substitution for offline medical care (10 items), and support seeking (17 items). Each item is rated on a 5-point Likert scale ranging from 1 (“never”) to 5 (“very frequently”). The total score ranges from 37 to 185, with higher scores indicating a higher level of online health-seeking behavior. The Cronbach’s α of the Chinese version was 0.981, and in the present study it was 0.992. In this study, the total OHSB score was used as the primary outcome variable.

eHealth Literacy Scale (eHEALS)

Electronic health literacy was assessed using the eHEALS, originally developed by Norman and Skinner (2006)²² and translated into Chinese and preliminarily validated by Guo et al (2013).²³ The scale consists of 8 items measuring respondents’ perceived knowledge, comfort, and skills in finding, evaluating, and applying electronic health information. Each item is rated on a 5-point Likert scale ranging from 1 (“strongly disagree”) to 5 (“strongly agree”). The total score ranges from 8 to 40, with higher scores indicating a higher level of eHealth literacy. The Cronbach’s α of the Chinese version was 0.913, and in the present study it was 0.984.

Caregiver Burden

Caregiver burden was assessed using the Zarit Burden Interview (ZBI), originally developed by Zarit et al,²⁴ and subsequently developed and validated in a Chinese version by Lu et al,²⁵ The scale consists of 22 items. Each item is rated on a 5-point Likert scale ranging from 0 (“never”) to 4 (“nearly always”). The total score ranges from 0 to 88, with higher scores indicating a heavier caregiver burden. The Cronbach’s α of the Chinese version was 0.87, and in the present study it was 0.971.

Caregiving Difficulty

Caregiving difficulty was assessed using the Family Caregiver Task Inventory (FCTI), originally developed by Clark and Rakowski (1983)²⁶ and translated into Chinese and psychometrically evaluated by Lee and Mok (2011).²⁷ The scale consists of 25 items. Each item is rated on a 3-point scale ranging from 0 (“not difficult”) to 2 (“very difficult”). The total score ranges from 0 to 50, with higher scores indicating poorer caregiving competence (ie, greater difficulty in performing caregiving tasks). The Cronbach’s α of the Chinese version was 0.930, and in the present study it was 0.960. To avoid ambiguity in interpretation, this variable was analyzed as caregiving difficulty rather than caregiving ability.

Anxiety

Anxiety was assessed using the Chinese version of the Generalized Anxiety Disorder 7-item scale (GAD-7), originally developed by Spitzer et al (2006)²⁸ and translated into Chinese and psychometrically validated by He et al (2010).²⁹ The scale consists of 7 items assessing the frequency of anxiety symptoms over the past two weeks. Each item is rated on a 4-point scale ranging from 0 (“not at all”) to 3 (“nearly every day”). The total score ranges from 0 to 21, with higher scores indicating more severe anxiety symptoms. The Cronbach’s α of the Chinese version was 0.898, and in the present study it was 0.968.

Statistical Analysis

Data were analyzed using IBM SPSS Statistics (version 31.0), and confirmatory factor analysis (CFA) was conducted using IBM SPSS Amos (version 31.0). Continuous variables were summarized as means and standard deviations (SDs), and categorical variables as frequencies and percentages. Internal consistency of the study scales was assessed using Cronbach’s alpha.

The measurement structure of the OHSB Scale was examined using CFA. The hypothesized three-factor model was compared with a single-factor model, and model fit was evaluated using the comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). The factorial structure of the Family Caregiver Task Inventory was also examined according to its proposed five-factor structure.

Given the conceptual proximity between OHSB and eHealth literacy, their construct distinctiveness was further examined. A four-factor measurement model specifying the three OHSB dimensions and eHEALS as separate latent factors was compared with an alternative model in which eHEALS was combined with the OHSB information-search factor. Discriminant validity was additionally assessed using the heterotrait-monotrait ratio (HTMT). Bootstrap 95% confidence intervals for HTMT estimates were obtained from 5,000 resamples, with values around 0.90 interpreted cautiously in conjunction with the corresponding confidence intervals and CFA results.

Univariate analyses were conducted to compare OHSB total scores across participant characteristics. Independent-samples t tests were used for two-group comparisons, and one-way analysis of variance (ANOVA) was used for comparisons involving three or more groups. Pearson correlation analysis was used to examine relationships among the major continuous study variables, including OHSB, eHealth literacy, caregiver burden, caregiving difficulty, and anxiety.

Hierarchical multiple linear regression was performed with the OHSB total score as the dependent variable. Candidate variables for the primary model were selected according to conceptual relevance and univariate findings, while limiting redundancy among closely related indicators within the same domain. The primary model represented child clinical context, caregiver sociodemographic and contextual characteristics, and caregiver psychosocial and cognitive characteristics. In Model 1, educational level, residence, and complications were entered as background characteristics. In Model 2, eHealth literacy, caregiver burden, caregiving difficulty, and anxiety were additionally entered. Regression coefficients (B), standardized coefficients (β), 95% confidence intervals (CIs), and model fit statistics were reported.

An expanded multivariable model was also fitted as a sensitivity analysis using the broader set of eligible child clinical and caregiver sociodemographic variables collected in the study, together with eHEALS, ZBI, FCTI, and GAD-7 scores. Preferred online information channels and weekly online health-seeking frequency were not included as covariates because they directly characterize online health-seeking activity and therefore overlap behaviorally with the study outcome. Results of the expanded analysis are presented in Supplementary Table S1.

Regression assumptions were assessed by examining multicollinearity, residual normality, heteroscedasticity, and influential observations. Multicollinearity was evaluated using variance inflation factors (VIFs), residual normality using the Shapiro–Wilk and Jarque–Bera tests, and heteroscedasticity using the Breusch–Pagan test. Influential observations were identified using Cook’s distance. Because the primary model showed evidence of non-normal residuals and heteroscedasticity, HC3 robust standard errors were calculated. Sensitivity analyses excluding influential observations were also performed. A two-sided p value < 0.05 was considered statistically significant.

Results

Participant Characteristics

A total of 365 primary caregivers of children with chronic diseases were included in the final analysis (Figure 1). Among the children, 51.2% were boys, and the largest age group was 7–12 years (41.1%). Renal disease was the most common diagnostic category (26.0%), followed by rheumatologic/vasculitic (17.0%) and neurologic/developmental conditions (16.7%). Most children were receiving drug therapy (89.0%), and 54.2% had experienced 1–3 hospitalizations or emergency department visits in the preceding year. Most had no comorbidities (79.5%) or complications (86.6%). Among caregivers, 90.7% were parents, 63.0% were women, and 63.0% were aged 31–40 years. Urban and rural residence were nearly evenly distributed (51.8% vs 48.2%). Short-video platforms (65.5%) and search engines (60.8%) were the most commonly reported online information channels, and 71.8% reported seeking health information online at least once per week (Table 1).

Figure 1.

Participant flowchart. Electronic questionnaires were distributed to primary caregivers of children with chronic diseases (n = 380). All questionnaires were screened for validity and eligibility. Fifteen questionnaires were excluded because of duplicate submission, major missing data, or inconsistent responses. Finally, 365 valid questionnaires were included in the analysis.

Table 1.

Participant Characteristics (N = 365)

Domain/Variable	Category	n	%
Child-related characteristics
Child sex
	Male	187	51.2%
	Female	178	48.8%
Child age
	0–3 years	35	9.6%
	4–6 years	66	18.1%
	7–12 years	150	41.1%
	13–18 years	114	31.2%
Diagnosis group
	Renal	95	26.0%
	Rheumatologic/vasculitic	62	17.0%
	Neurologic/developmental	61	16.7%
	Respiratory/allergic airway	49	13.4%
	Hematologic/oncologic	34	9.3%
	Endocrine/metabolic	30	8.2%
	Cardiovascular	30	8.2%
	Other/unclear	4	1.1%
Disease duration
	<3 months	48	13.2%
	3 months-<1 year	93	25.5%
	1–3 years	93	25.5%
	3–5 years	131	35.9%
Treatment: drug therapy
	No	40	11.0%
	Yes	325	89.0%
Treatment: physical therapy
	No	342	93.7%
	Yes	23	6.3%
Treatment: rehabilitation training
	No	335	91.8%
	Yes	30	8.2%
Treatment: other
	No	339	92.9%
	Yes	26	7.1%
Hospitalizations/ED visits in past year
	0	120	32.9%
	1–3	198	54.2%
	4–7	25	6.8%
	>7	22	6.0%
Long-term medication
	Yes	172	47.1%
	No	177	48.5%
	Unclear	16	4.4%
Comorbidities
	0	290	79.5%
	1–3	62	17.0%
	>3	13	3.6%
Complications
	0	316	86.6%
	1–3	45	12.3%
	>3	4	1.1%
Care setting
	Inpatient	186	51.0%
	Outpatient	179	49.0%
Caregiver-related characteristics
Relationship to child
	Parent	331	90.7%
	Grandparent	28	7.7%
	Other relative	6	1.6%
Caregiver sex
	Male	135	37.0%
	Female	230	63.0%
Caregiver age
	20–30 years	78	21.4%
	31–40 years	230	63.0%
	41–50 years	28	7.7%
	≥51 years	29	7.9%
Education
	Primary or below	28	7.7%
	Junior middle school	91	24.9%
	High school/secondary	93	25.5%
	College	91	24.9%
	Bachelor’s	45	12.3%
	Master’s or above	17	4.7%
Occupation
	Enterprise employee	49	13.4%
	Government/public institution	99	27.1%
	Self-employed	88	24.1%
	Full-time caregiver	58	15.9%
	Retired	25	6.8%
	Other	46	12.6%
Monthly household income
	≤5000 CNY	131	35.9%
	5001–10000 CNY	199	54.5%
	10,001–20000 CNY	29	7.9%
	>20000 CNY	6	1.6%
Number of children in family
	1	108	29.6%
	2	152	41.6%
	3	76	20.8%
	≥4	29	7.9%
Residence
	Urban	189	51.8%
	Rural	176	48.2%
Internet-use characteristics
Preferred channel: search engines
	No	143	39.2%
	Yes	222	60.8%
Preferred channel: medical apps
	No	216	59.2%
	Yes	149	40.8%
Preferred channel: forums
	No	228	62.5%
	Yes	137	37.5%
Preferred channel: short-video platforms
	No	126	34.5%
	Yes	239	65.5%
Preferred channel: WeChat official accounts
	No	313	85.8%
	Yes	52	14.2%
Preferred channel: Bilibili
	No	354	97.0%
	Yes	11	3.0%
Preferred channel: AI tools
	No	297	81.4%
	Yes	68	18.6%
Online health-seeking frequency
	0 times/week	103	28.2%
	1–3 times/week	160	43.8%
	4–7 times/week	69	18.9%
	>7 times/week	33	9.0%

Confirmatory factor analysis and Construct Distinctiveness

Confirmatory factor analysis supported the hypothesized three-factor structure of the OHSB Scale. The three-factor model showed good fit (CFI = 0.960, TLI = 0.957, RMSEA = 0.069, SRMR = 0.032), whereas the one-factor model showed substantially poorer fit (CFI = 0.733, TLI = 0.716, RMSEA = 0.176, SRMR = 0.122). The proposed five-factor FCTI model showed modest fit (CFI = 0.861, TLI = 0.842, RMSEA = 0.095, SRMR = 0.062).

In the construct-distinctiveness analysis, the four-factor model specifying the three OHSB dimensions and eHEALS as separate latent factors showed good fit (CFI = 0.962, TLI = 0.960, RMSEA = 0.060, SRMR = 0.031), whereas the model merging eHEALS with the information-search factor showed poorer fit (CFI = 0.899, TLI = 0.893, RMSEA = 0.098, SRMR = 0.050). The HTMT value between the OHSB total item pool and eHEALS was 0.891 (95% CI, 0.860–0.918). HTMT values between the three OHSB dimensions and eHEALS ranged from 0.810 to 0.851 (Supplementary Table S2).

Scale Performance

Descriptive statistics and internal consistency estimates for the study scales are presented in Table 2. The mean OHSB total score was 94.57 ± 44.91, corresponding to an item mean of 2.56. Among the three dimensions, information search had the highest item mean (2.83 ± 1.24), followed by supplement/substitution for offline medical care (2.66 ± 1.13) and support seeking (2.33 ± 1.40). Cronbach’s α coefficients for the main study measures ranged from 0.960 to 0.992 (Table 2).

Table 2.

Descriptive Statistics and Internal Consistency of Study Scales

Scale/Subscale	Items	Theoretical Range	Mean ± SD	Observed Range	Cronbach α
OHSB total	37	37-185	94.57 ± 44.91	37–183	0.992
eHEALS total	8	8-40	22.11 ± 10.39	8–40	0.984
ZBI total	22	0-88	47.32 ± 18.02	10–87	0.971
FCTI total	25	0-50	16.15 ± 11.87	0–50	0.960
GAD-7 total	7	0-21	8.74 ± 6.10	0–21	0.968

Univariate Associations with OHSB

In univariate analyses, OHSB total score differed significantly according to child age, disease duration, hospitalizations/emergency department visits in the past year, long-term medication status, comorbidities, and complications, whereas no significant differences were observed by child sex or diagnosis category. Caregivers of children aged 0–3 years had the highest mean OHSB score (129.06 ± 41.24).

Among caregiver characteristics, significant differences in OHSB were observed according to relationship to the child, caregiver sex, age, educational level, occupation, household income, number of children in the family, and residence. OHSB scores were higher among urban than rural caregivers (123.85 ± 38.43 vs 63.12 ± 26.21, p < 0.001) and increased across categories of weekly online health-seeking frequency (Table 3).

Table 3.

Univariate Analysis of OHSB Total Score Across Participant Characteristics

Variable	Category	n	OHSB Total, Mean ± SD	Test Statistic	P value
Child sex	Male	187	92.94 ± 44.69	−0.708	0.479
	Female	178	96.28 ± 45.20
Child age	0–3 years	35	129.06 ± 41.24	8.466	<0.001
	4–6 years	66	90.50 ± 45.42
	7–12 years	150	88.51 ± 43.63
	13–18 years	114	94.31 ± 43.13
Diagnosis group	Renal	95	96.99 ± 47.17	1.422	0.195
	Rheumatologic/vasculitic	62	89.21 ± 43.27
	Neurologic/developmental	61	84.98 ± 44.37
	Respiratory/allergic airway	49	90.78 ± 41.32
	Hematologic/oncologic	34	105.03 ± 45.20
	Endocrine/metabolic	30	102.50 ± 38.05
	Cardiovascular	30	106.40 ± 50.45
	Other/unclear	4	75.50 ± 49.84
Disease duration	<3 months	48	108.96 ± 38.25	2.664	0.048
	3 months-<1 year	93	96.77 ± 46.15
	1-3 years	93	94.04 ± 49.00
	3-5 years	131	88.10 ± 42.29
Hospitalizations/ED visits in past year	0	120	91.10 ± 42.13	2.679	0.047
	1–3	198	92.79 ± 44.74
	4–7	25	105.88 ± 45.74
	>7	22	116.64 ± 54.47
Long-term medication	Yes	172	105.68 ± 46.61	11.055	<0.001
	No	177	85.65 ± 40.79
	Unclear	16	73.75 ± 42.74
Comorbidities	0	290	90.58 ± 44.51	5.711	0.004
	1–3	62	109.84 ± 42.98
	>3	13	110.69 ± 46.86
Complications	0	316	91.08 ± 43.89	7.354	<0.001
	1–3	45	116.87 ± 44.65
	>3	4	119.25 ± 60.01
Relationship to child	Parent	331	98.68 ± 44.35	20.384	<0.001
	Grandparent	28	45.07 ± 7.31
	Other relative	6	98.83 ± 42.78
Caregiver sex	Male	135	86.83 ± 42.73	−2.585	0.010
	Female	230	99.11 ± 45.62
Caregiver age	20-30 years	78	117.69 ± 44.65	19.754	<0.001
	31–40 years	230	93.19 ± 42.16
	41–50 years	28	88.71 ± 45.59
	≥51 years	29	48.93 ± 22.32
Education	Primary or below	28	47.07 ± 12.68	81.938	<0.001
	Junior middle school	91	60.15 ± 24.01
	High school/secondary	93	84.96 ± 36.72
	College	91	119.96 ± 36.40
	Bachelor’s	45	139.07 ± 30.04
	Master’s or above	17	155.88 ± 14.97
Occupation	Enterprise employee	49	105.51 ± 42.57	23.205	<0.001
	Government/public institution	99	118.30 ± 37.73
	Self-employed	88	80.53 ± 38.34
	Full-time caregiver	58	107.12 ± 50.52
	Retired	25	43.84 ± 6.55
	Other	46	70.41 ± 34.59
Monthly household income	≤5000 CNY	131	65.20 ± 31.30	39.138	<0.001
	5001–10000 CNY	199	109.31 ± 43.66
	10,001–20000 CNY	29	120.03 ± 39.28
	>20000 CNY	6	123.67 ± 34.15
Number of children in family	1	108	118.46 ± 43.40	33.618	<0.001
	2	152	97.74 ± 43.69
	3	76	70.71 ± 32.00
	≥4	29	51.48 ± 15.92
Residence	Urban	189	123.85 ± 38.43	17.742	<0.001
	Rural	176	63.12 ± 26.21
OHSB frequency (single-item)	0 times/week	103	46.83 ± 9.55	243.734	<0.001
	1–3 times/week	160	93.46 ± 33.83
	4–7 times/week	69	141.43 ± 22.48
	>7 times/week	33	150.94 ± 23.01

Correlations Among Key Variables

OHSB total score was strongly correlated with information search (r = 0.949), supplement/substitution for offline medical care (r = 0.920), and support seeking (r = 0.954), all p < 0.001. OHSB was also strongly positively correlated with eHEALS (r = 0.883, p < 0.001). In contrast, correlations of OHSB with ZBI, FCTI, and GAD-7 were weak. ZBI, FCTI, and GAD-7 were strongly intercorrelated, with coefficients ranging from 0.689 to 0.700 (all p < 0.001) (Table 4).

Table 4.

Pearson Correlations Among Key Study Variables

Variable	1. OHSB Total	2. Information Search	3. Supplement or Substitution for Offline Medical Care	4. Support Seeking	5. eHEALS	6. ZBI	7. FCTI	8. GAD-7
1. OHSB total	1.000
2. Information search	0.949***	1.000
3. Supplement or substitution for offline medical care	0.920***	0.929***	1.000
4. Support seeking	0.954***	0.827***	0.775***	1.000
5. eHEALS	0.883***	0.859***	0.781***	0.846***	1.000
6. ZBI	0.086	0.086	0.113*	0.064	−0.032	1.000
7. FCTI	−0.044	−0.016	0.009	−0.078	−0.120*	0.700***	1.000
8. GAD-7	0.041	0.093	0.122*	−0.030	−0.109*	0.690***	0.689***	1.000

Note: *p < 0.05; ***p < 0.001.

Abbreviations: OHSB, Online Health-Seeking Behavior Scale; eHEALS, eHealth Literacy Scale; ZBI, Zarit Burden Interview; FCTI, Family Caregiver Task Inventory; GAD-7, Generalized Anxiety Disorder-7.

Primary and Expanded Multivariable Analyses

Model 1, including educational level, residence, and complications, explained 62.8% of the variance in OHSB (R² = 0.628, F(3, 361) = 203.16, p < 0.001). After adding eHEALS, ZBI, FCTI, and GAD-7, Model 2 explained 81.6% of the variance (R² = 0.816, adjusted R² = 0.813, F(7, 357) = 226.61, p < 0.001), representing an additional 18.8% of explained variance (ΔR² = 0.188, ΔF(4, 357) = 91.46, p < 0.001).

In Model 2, higher educational level (B = 3.350, β = 0.096, p = 0.009), rural residence (B = −9.478, β = −0.106, p = 0.001), complications (B = 7.462, β = 0.064, p = 0.007), eHEALS (B = 3.179, β = 0.735, p < 0.001), ZBI (B = 0.175, β = 0.070, p = 0.045), and GAD-7 (B = 0.948, β = 0.129, p < 0.001) were statistically associated with OHSB, whereas FCTI was not significant (p = 0.090) (Table 5).

Table 5.

Hierarchical Multiple Linear Regression for Factors Associated with OHSB Total Score

Predictor	Model 1: B	Model 1: SE	Model 1: β	Model 1: p	Model 1: 95% CI	Model 1	Model 2: B	Model 2: SE	Model 2: β	Model 2: p	Model 2: 95% CI	Model 2
Education level (ordinal)	17.019	1.395	0.489	<0.001	14.276, 19.761		3.350	1.270	0.096	0.009	0.853, 5.847
Rural residence (vs urban)	−33.393	3.605	−0.372	<0.001	−40.483, −26.302		−9.478	2.911	−0.106	0.001	−15.203, −3.754
Complications (ordinal)	14.062	3.785	0.120	<0.001	6.620, 21.505		7.462	2.743	0.064	0.007	2.067, 12.857
eHEALS total score							3.179	0.178	0.735	<0.001	2.830, 3.529
ZBI total score							0.175	0.087	0.070	0.045	0.004, 0.346
FCTI total score							−0.227	0.134	−0.060	0.090	−0.490, 0.036
GAD-7 total score							0.948	0.255	0.129	<0.001	0.447, 1.450
R2						0.628						0.816
Adjusted R2						0.625						0.813
F						203.16 (df = 3, 361)***						226.61 (df = 7, 357)***

Note: Model 1 vs Model 2: ΔR² = 0.188, ΔF (4, 357) = 91.46, ***p < 0.001.

Abbreviations: B, unstandardized coefficient; SE, standard error; β, standardized coefficient; CI, confidence interval.

In the expanded multivariable model, R² was 0.846 and adjusted R² was 0.829. Using HC3 robust inference, disease duration, educational level, rural residence, eHEALS, ZBI, and GAD-7 remained statistically associated with OHSB. eHEALS showed the largest standardized coefficient (B = 3.035, β = 0.702, p < 0.001), whereas complications and FCTI were not statistically significant (Supplementary Table S1).

Regression Diagnostics and Sensitivity Analyses

VIF values in the primary fully adjusted model ranged from 1.06 to 3.29. Residuals deviated from normality according to the Shapiro–Wilk test (W = 0.984, p < 0.001) and Jarque–Bera test (p < 0.001), and the Breusch–Pagan test indicated heteroscedasticity (p = 0.006). Nineteen observations had Cook’s distance values greater than 4/n. With HC3 robust standard errors, educational level, rural residence, eHEALS, and GAD-7 remained statistically significant, whereas complications (p = 0.063) and ZBI (p = 0.051) no longer met the conventional p < 0.05 threshold. In analyses excluding influential observations, educational level, residence, and eHEALS remained statistically significant, while the statistical significance of several secondary associations varied across sensitivity analyses.

Discussion

As digital health resources become increasingly integrated into pediatric care, caregivers’ engagement with online health information has emerged as an important consideration in the long-term management of childhood chronic disease.³⁰ However, the existing evidence remains fragmented, and few studies have examined this behavior specifically among primary caregivers of children with chronic diseases within an integrated analytical framework that simultaneously accounts for child-related, sociodemographic, psychosocial, and cognitive factors. Addressing this gap is particularly important in pediatric chronic care, where information needs are ongoing and often complex.^31,32 Accordingly, the present study characterized online health-seeking behavior in this population and examined its associations across these domains. eHealth literacy showed the strongest association with OHSB, while educational level and residence were also consistently associated across the primary and sensitivity analyses; associations involving child clinical characteristics, caregiver burden, and anxiety were comparatively less stable. These findings are discussed below in relation to previous evidence.

Online health-seeking behavior has become increasingly integrated into the everyday management of pediatric chronic disease.⁹ In the present study, caregivers showed frequent engagement with online health information, with a mean OHSB total score of 94.57 and 71.8% reporting online health-seeking behavior at least once per week. Short-video platforms and search engines were the most frequently used channels. Among the three dimensions, information search had the highest item mean (2.83), followed by supplement or substitution for offline medical care (2.66) and support seeking (2.33). These findings indicate that caregivers used online resources for multiple purposes, including acquiring disease-related information, supplementing offline care, and seeking experiential or emotional support.^9,16 This pattern is broadly consistent with Treadgold et al’s qualitative synthesis of parents and carers of children with long-term physical conditions and with Wang et al’s study among Chinese caregivers of children with cancer, both of which showed that online information use in pediatric caregiving serves informational and supportive functions.^8,16 Whereas these previous studies relied on qualitative accounts or focused on a specific disease group, the present study quantified multiple dimensions of OHSB across caregivers of children with a broader range of chronic conditions.

eHealth literacy is increasingly recognized as an important component of engagement with digital health information.^11,13 In the present study, eHealth literacy showed the strongest association with OHSB, with a very strong positive correlation (r = 0.883) and the largest standardized coefficient in the primary fully adjusted model (β = 0.735). A similarly strong association was observed after broader covariate adjustment. This finding is broadly consistent with Kristjánsdóttir et al, who reported socioeconomic and demographic differences in eHealth literacy among parents of children requiring pediatric surgery,¹⁰ and with Hölgyesi et al, who found that parental electronic health literacy was related to the disease-management context in families of children with type 1 diabetes.¹³ Together, these studies support the relevance of caregivers’ perceived ability to locate, understand, evaluate, and apply electronic health information in pediatric care. The strong association observed in the present study should nevertheless be interpreted in the context of the close relationship between the two constructs. The four-factor model specifying eHEALS and the three OHSB dimensions as separate latent factors showed better fit than the model merging eHEALS with information search. However, the HTMT value for the OHSB total item pool and eHEALS was 0.891 (95% CI, 0.860–0.918), indicating substantial overlap at the total-scale level. Thus, although OHSB and eHealth literacy were empirically distinguishable, their conceptual and measurement proximity may have contributed to the magnitude of the observed correlation and to the high proportion of variance explained by the fully adjusted model (R² = 0.816). This possibility is also relevant given that both constructs were assessed using self-reported questionnaires administered at the same time.³³

Sociodemographic and contextual resources were also associated with caregivers’ engagement with online health information. Higher educational level was associated with higher OHSB scores, whereas rural residence was associated with lower OHSB, and both associations remained statistically significant across the primary and sensitivity analyses. These findings are consistent with previous evidence showing sociodemographic differences in digital health engagement and eHealth literacy among parents and caregivers.^15,34 Higher educational attainment may facilitate the comprehension and appraisal of online health information, while place of residence may reflect broader differences in digital infrastructure, healthcare accessibility, and opportunities to obtain and verify health information.^34,35 The observed urban–rural difference should therefore not be interpreted as an effect of residence alone, because it may also reflect correlated differences in socioeconomic resources and access to healthcare and digital services. More broadly, these findings suggest that digital health support for caregivers should account for differences in educational and contextual resources rather than assuming that access to online information is equivalent across populations.

Child clinical characteristics were also related to OHSB, although individual associations were less consistent after multivariable adjustment. In univariate analyses, OHSB differed according to child age, disease duration, hospitalizations or emergency department visits, long-term medication status, comorbidities, and complications, whereas diagnosis category was not significantly associated with OHSB. These findings are broadly consistent with previous evidence suggesting that caregivers’ health-information needs vary according to the child’s health status and the demands of ongoing care.^7,12 Caregivers early in the disease course may face unfamiliar treatment decisions and greater informational uncertainty, while repeated healthcare encounters, long-term medication use, and multiple health problems may generate continuing information needs.³⁶ The multivariable findings were more heterogeneous. Complications were associated with OHSB in the conventional primary model, but this association was attenuated when HC3 robust standard errors were applied and was not statistically significant in the expanded model. Disease duration remained associated with OHSB after broader adjustment. These results suggest that the child’s clinical context is relevant to online health seeking, but no single marker of disease complexity showed a uniformly consistent association across analyses. The absence of a significant association with diagnosis category further suggests that information needs may relate more closely to the course and management demands of illness than to diagnostic labels alone.¹²

Psychological and caregiving-related factors showed weaker and less consistent associations with OHSB than eHealth literacy and sociodemographic characteristics.^14,37 Caregiver burden and anxiety showed little or no bivariate correlation with OHSB but reached statistical significance in the conventional fully adjusted model, whereas caregiving difficulty was not statistically significant. The emergence of the burden and anxiety associations after multivariable adjustment is compatible with a conditional or suppression-like pattern rather than a straightforward direct relationship. These associations also varied across sensitivity analyses. Anxiety remained statistically significant with HC3 robust standard errors in the primary and expanded models but became borderline after influential observations were excluded from the expanded model. The association with caregiver burden was similarly sensitive to model specification, while caregiving difficulty remained non-significant. These findings should therefore be interpreted cautiously. The strong intercorrelations among caregiver burden, caregiving difficulty, and anxiety suggest that they may capture overlapping aspects of a broader caregiving stress context. One possibility is that greater anxiety is accompanied by increased uncertainty and vigilance regarding a child’s health and therefore greater engagement with online information. Conversely, repeated exposure to uncertain, conflicting, or alarming online health information may also be associated with greater anxiety. This bidirectional interpretation is consistent with previous evidence linking stress, anxiety, and health literacy among parents of children with chronic illness.¹⁴ Longitudinal studies are needed to clarify the temporal relationships among these factors.

Pediatric chronic care increasingly extends beyond the clinic, and the present findings highlight the importance of providing caregivers with clinician-endorsed, literacy-sensitive digital resources that account for differences in educational background, contextual resources, and changing informational needs across the disease course.^30,38 By integrating child clinical, sociodemographic, psychosocial, and cognitive characteristics within the same analytical framework and applying validated instruments together with regression diagnostics and sensitivity analyses, this study provides a multidimensional view of OHSB among primary caregivers of children with chronic diseases. Several limitations should be acknowledged. First, the cross-sectional design precludes causal inference and does not establish the temporal direction of associations involving OHSB, eHealth literacy, or anxiety. Second, all major constructs were assessed using self-reported questionnaires, raising the possibility of common-method variance; the substantial overlap between OHSB and eHealth literacy may have further increased shared measurement variance. Third, the very high internal consistency coefficients of several scales may partly reflect item redundancy rather than measurement precision alone. Fourth, the single-center convenience-sampling design limits generalizability, and use of an electronic questionnaire may have preferentially included caregivers who were more comfortable with digital technologies. Fifth, the inclusion of children with heterogeneous chronic diseases may have obscured disease-specific patterns of information seeking. Future multicenter and longitudinal studies incorporating objective measures of digital information use and disease-specific analyses would help clarify these relationships and further evaluate the distinction between eHealth literacy and OHSB.

Conclusion

Online health-seeking behavior is a prominent feature of caregiving in pediatric chronic disease. OHSB was most strongly associated with caregivers’ eHealth literacy, while educational level and residence also showed comparatively consistent associations across the primary and sensitivity analyses. Associations involving child clinical characteristics, caregiver burden, and anxiety were less stable across model specifications. These findings highlight the need to view digital health information use not simply as an issue of access, but also as one of capability and context. Accordingly, helping caregivers locate, appraise, and apply trustworthy online health information—with particular attention to those who are less educated, less resourced, or more anxious—may represent an important and equity-relevant component of pediatric chronic care. Longitudinal and multicenter studies are warranted to clarify the directionality of these associations and to confirm their generalizability.

Funding Statement

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data Sharing Statement

The data used in this study are available from the corresponding author upon reasonable request.

Ethics Approval and Consent to Participate

This study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Xinjiang Medical University (Approval No. K202510-39). The study adhered to the ethical principles of the Declaration of Helsinki. Participation was voluntary and responses were anonymous. Informed consent to participate was obtained electronically from all participants prior to survey initiation: participants reviewed the study information and confirmed their agreement by selecting an “I agree” checkbox before proceeding. Consent was further confirmed by completion and submission of the questionnaire. Participants were free to discontinue the survey at any time without penalty, and all data were kept confidential.

Author Contributions

Rouxianguli Aila contributed to the investigation, data acquisition and curation, and drafting of the manuscript. Xiaoqian Zhou contributed to the investigation, formal analysis, interpretation of the results, and critical revision of the manuscript. Jing Cao contributed to the investigation, formal analysis, interpretation of the results, and critical revision of the manuscript. Hong Zhang contributed to formal analysis, interpretation of the results, supervision, and critical revision of the manuscript. Shuwan Dong contributed to data visualization and critical revision of the manuscript. Gulisumuhan Abulaiti contributed to the conception and design of the study, supervision, interpretation of the results, drafting of the manuscript, and critical revision of the manuscript. All authors made a significant contribution to the work reported, whether in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising, or critically reviewing the article; gave final approval of the version to be published; agreed on the journal to which the article was submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that they have no conflicts of interest in this work.