Community-based traditional Chinese medicine health service use among older adults and provider-reported delivery barriers in Wuhan, China: an explanatory sequential mixed-methods study

Abstract

Background

Population ageing and multimorbidity are increasing the need for accessible community-based care, while service use may vary according to service availability, affordability, and delivery capacity. In China, community-based traditional Chinese medicine (TCM) health services represent one model of primary care support for prevention and chronic disease management, but evidence rarely integrates older adults’ service use with frontline providers’ delivery experiences within the same setting.

Methods

An explanatory sequential mixed-methods design was used in Wuhan, China, from December 2024 to July 2025. The quantitative phase involved a structured survey of 1,195 adults aged ≥ 60 years who were recruited on site at selected community health service centres. The qualitative phase included in-depth interviews with 11 primary care providers. Quantitative data were analysed using descriptive statistics, Pearson’s χ² tests or Fisher’s exact test, and multivariable binary logistic regression. Two multivariable binary logistic regression models were fitted: a primary model excluding awareness of services and an exploratory model including it. Firth’s penalised likelihood was used as a sensitivity analysis for the small uninsured subgroup.

Results

Among the 1,195 older adults recruited during designated survey sessions at seven selected community health service centres, 729 (61.0%) reported receiving at least one community-based TCM health service during the past 12 months. In bivariate analyses, the proportion reporting use during the past 12 months differed significantly by sex, education level, medical insurance status, monthly household income, smoking status, drinking status, number of chronic conditions, awareness of community-based TCM health services, perceived convenience of access, and perceived complexity of the service process (all P < 0.05). In the primary multivariable model excluding awareness, higher educational attainment, higher monthly household income, and a greater number of chronic conditions were associated with higher odds of reported use. Participants without medical insurance had lower odds of reported use than those covered by urban employee basic medical insurance (aOR = 0.30, 95% CI: 0.11–0.77). The Firth penalised logistic regression sensitivity analysis yielded a directionally consistent estimate for the small uninsured subgroup. Thematic analysis of interviews with primary care providers identified three themes of provider-reported or provider-perceived contextual barriers to service delivery: perceived limitations in policy and support systems, workforce and material resources, and barriers related to community awareness and acceptance.

Conclusions

Among older adults recruited on site at selected community health service centres, reported use of community-based TCM health services during the past 12 months was associated with educational attainment, monthly household income, number of chronic conditions, and medical insurance status. Interviewed primary care providers reported service-delivery challenges related to policy and support systems, workforce and material resources, and their perceptions of community awareness and acceptance. These findings describe individual-level associations within the study sample and provider-reported service-delivery challenges, and may inform further research and local service planning.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12913-026-15625-5.

Introduction

Population ageing is a major global public health challenge with important implications for sustainable development. By 2050, people aged 60 years or older are projected to account for approximately 22.00% of the global population [1]. China is one of the countries with the most pronounced population ageing worldwide; in China, people aged 60 years or above are expected to reach 395 million [2]. As life expectancy increases, chronic diseases and multimorbidity have become increasingly prevalent in older adults [3, 4], substantially compromising physical and mental health and imposing heavy socioeconomic burdens [5]. According to a study published in The Lancet Public Health [6], the prevalence of diabetes among Chinese adults is estimated at 23.90% for those aged 60–69 years and 27.30% for those aged ≥ 70 years. In China, the societal cost of diabetes is projected to increase from USD 250.20 billion in 2020 to USD 460.40 billion by 2030, with per-capita costs rising substantially [7]. Ensuring healthy ageing and meeting the health service needs of older people in the community is therefore an urgent public health priority.

Community-based traditional Chinese medicine (TCM) health services have become an integral component of public health systems, particularly for elderly populations, by enhancing access to care, facilitating chronic disease management through holistic approaches, and strengthening health promotion rooted in traditional preventive philosophies [8]. As frontline providers, primary care providers (PCPs) undertake a broad spectrum of responsibilities, including health education, chronic disease management, basic diagnostic guidance, medication counseling, rehabilitation support, and referrals [9–12]. In practice, PCPs are expected to carry out these duties under heavy workloads, while maintaining continuous communication with patients, caregivers, and multidisciplinary teams. However, the delivery of these services is often constrained by systemic and institutional barriers [13–15]. At the policy level, discrepancies between national directives and local implementation, fragmented funding mechanisms, inadequate reimbursement, and weak cross-sector coordination can undermine service delivery [16, 17]. At the operational level, shortages of qualified personnel, limited standardized training, outdated equipment, and insufficient infrastructure restrict service capacity and efficiency. On the community side, low health literacy, skepticism toward TCM, and ineffective communication strategies may further reduce engagement and adherence. These interrelated challenges place substantial strain on PCPs and compromise both the accessibility and quality of TCM health services for the aging population. Examining these challenges from the perspective of PCPs can provide important context for understanding service-delivery conditions in community settings.

Although some studies have examined community-based TCM health service use and its associated factors [18], fewer studies have considered older adults’ reported use alongside primary care providers’ experiences of delivering these services within the same setting [19–21]. Because the two perspectives arise from different participant groups, provider accounts can provide service-delivery context but cannot directly represent older adults’ personal motivations, preferences, affordability concerns, or experiences.

This study therefore aimed to: (1) describe reported use of community-based TCM health services during the past 12 months among older adults recruited at selected community health service centres; (2) identify individual-level factors associated with reported use; (3) explore delivery barriers reported by primary care providers; and (4) compare the quantitative and qualitative findings to identify complementary, discordant, or distinct insights.

Methods

Study design

This study employed an explanatory sequential mixed-methods design, conducted in two phases [22]. In Phase 1, a quantitative survey examined reported use of community-based TCM health services during the past 12 months and associated individual-level factors among older adults recruited at selected community health service centres. In Phase 2, semi-structured interviews explored primary care providers’ experiences of service delivery and the barriers they reported in routine practice. Preliminary quantitative findings informed the focus of the interview guide.

Because the two phases involved different participant groups, the qualitative phase was not intended to identify older adults’ personal reasons for using or not using services. Instead, it provided provider-side contextual information relevant to selected quantitative findings. The two datasets were integrated at the interpretation stage through a joint display and interpretive comparison [23]. Integration focused on identifying complementarity, discordance, and findings without a direct counterpart in the other dataset. The integrated interpretations were contextual and hypothesis-generating rather than causal. The mixed-methods study was reported with reference to the Good Reporting of A Mixed Methods Study (GRAMMS) [24] recommendations, and the qualitative component was reported in accordance with the Consolidated Criteria for Reporting Qualitative Research (COREQ) [25].

Quantitative phase

Participants

A stratified random sampling procedure was used to select the survey sites, followed by on-site participant recruitment at the selected sites. First, the 13 administrative districts of Wuhan were stratified into central urban districts and outer urban districts according to their administrative and geographical characteristics. The central urban stratum comprised Jiang’an, Hanyang, Qiaokou, Wuchang, Hongshan, Qingshan, and Jianghan districts, whereas the outer urban stratum comprised Dongxihu, Caidian, Jiangxia, Huangpi, Xinzhou, and Hannan districts. Lists of community health service centres within the two strata were used as the sampling frames for survey-site selection. A random-number table was then used to select centres independently within each stratum. In total, seven community health service centres were included, comprising four centres from the central urban stratum and three from the outer urban stratum.

Participants were recruited on site during designated survey sessions held in conjunction with community activities for older residents and routine health examinations at the selected centres. The inclusion criteria were: (1) community-dwelling adults aged ≥ 60 years; (2) residence in the same community for ≥ 1 year; (3) sufficient comprehension, literacy, and communication ability to complete the questionnaire; and (4) voluntary participation with written informed consent. Individuals with severe psychiatric or organic mental disorders, severe hearing or visual impairment that prevented questionnaire completion, or stroke or a major cardiovascular event during the preceding six months were excluded. Questionnaires were either self-administered or administered face to face by trained investigators using standardised wording and a fixed question sequence. Previous use of community-based TCM health services was not an eligibility criterion; both users and non-users were eligible to participate.

Individual participants were not randomly sampled with known selection probabilities from a complete register of older residents. Random selection therefore applied only at the survey-site level, whereas final participant recruitment was site based and non-probability in nature.

For fieldwork planning, the planned target sample size was estimated using the formula for a single population proportion, [26]. Because no reliable prior estimate of community-based TCM health service use was available, an expected proportion of 50% was used to obtain a conservative estimate. Assuming a two-sided α of 0.05 and an absolute margin of error of 3%, the minimum required sample size was 1,068. This number was increased by approximately 20% to provide a buffer for incomplete or unusable questionnaires, yielding 1,282, and the target number of questionnaires distributed was rounded up to 1,300. This calculation was used to guide the planned fieldwork scale and should not be interpreted as providing design-based precision for a population estimate of all older adults in Wuhan.

Data collection

Data were collected from December 2024 to May 2025 in Wuhan, Hubei Province. Before participation, participants were informed of the study’s purpose and procedures and provided written informed consent. Participants were informed that participation was voluntary and that they could withdraw at any time, for any reason, without penalty.

Questionnaire development and measurements

A structured questionnaire was developed to assess sociodemographic and socioeconomic characteristics, health status and health-related behaviours, and awareness, use, experiences, and perceptions of community-based TCM health services. The questionnaire was developed with reference to Andersen’s Behavioural Model of Health Services Use and relevant literature. The draft questionnaire underwent two rounds of Delphi expert consultation involving 12 experts in community-based TCM health services, TCM nursing, and health management for older adults. The experts rated the relevance of each item and provided comments on item clarity and applicability. The research team revised the questionnaire according to the first-round findings, and the experts reassessed the revised items in the second round. Based on the second-round relevance ratings, the questionnaire-level content validity index, calculated as the average of the item-level CVIs (S-CVI/Ave), was 0.95 for the 23 items retained in the final questionnaire.

During the Delphi process, the service-recommendation item was removed because the experts considered it to overlap substantially with the existing satisfaction item. The revised questionnaire was subsequently pilot-tested among 30 community-dwelling older adults. Following pilot testing, the additional willingness-to-pay item was removed because it was interpreted inconsistently and frequently left unanswered. The wording, response options, and skip instructions of several other items were also revised to improve clarity and fieldwork feasibility.

The questionnaire comprised four sections: (1) sociodemographic and socioeconomic characteristics; (2) health status and health-related behaviors; (3) awareness and use of community-based TCM health services; and (4) service experiences and perceptions. Supplementary file A provides the complete questionnaire administered in this study. The number of chronic conditions was derived from a multiple-response checklist of physician-diagnosed conditions and categorised as none, one, two, or three or more conditions. The checklist included hypertension, diabetes mellitus, coronary heart disease, dyslipidemia, osteoporosis, chronic obstructive pulmonary disease, chronic gastritis, liver cirrhosis, asthma, stroke or cerebrovascular disease, Parkinson’s disease, peptic ulcer, chronic kidney disease, chronic cor pulmonale, Alzheimer’s disease, and other physician-diagnosed chronic conditions.

Internal consistency was not assessed for the questionnaire as a whole because it comprised heterogeneous item types and was not designed as a unidimensional scale.

Data analysis

All statistical analyses were conducted using Stata 18.0. Categorical variables were reported as frequencies and percentages. Differences in reported use across participant subgroups were examined using Pearson’s χ² tests; Fisher’s exact test was used for living arrangement because of sparse cells. All tests were two-sided, and a P-value < 0.05 was considered statistically significant.

The binary outcome was self-reported receipt of at least one listed community-based TCM health service during the past 12 months, derived from Item 14. Participants who selected at least one specific service in Item 14 were classified as users and coded as 1, whereas those who selected “None of the above” were classified as non-users and coded as 0. Andersen’s Behavioral Model of Health Services Use [27] was used as a conceptual framework to organise the available covariates into predisposing, enabling, and need-related factors. Sex, age group, marital status, and education were treated as predisposing factors; medical insurance, living arrangement, household income, perceived convenience, perceived high cost, and perceived policy support as enabling factors; and number of chronic conditions as a need-related factor. For the purposes of this analysis, smoking and drinking status were treated as personal health-practice variables within the predisposing domain of Andersen’s model and were retained in both multivariable models to maintain consistency with the conceptual organisation of the covariates. The framework was used to guide covariate organisation and interpretation rather than to test the complete theoretical model. Multivariable binary logistic regression was used to examine associated factors. Categorical predictors were entered as indicator variables using the reference categories specified in Supplementary file C Table S2. Model 1, designated as the primary model, included sex, age group, marital status, education, medical insurance, living arrangement, household income, smoking, drinking, number of chronic conditions, perceived convenience, perceived high cost, and perceived policy support. Perceived complexity of the service process was examined only in the bivariate analyses.

Model 2 additionally included awareness as a five-category variable, with “not at all aware” as the reference category, and was treated as exploratory because awareness and service use were measured concurrently. The models were compared using the likelihood-ratio test, Akaike information criterion, and Bayesian information criterion, and model fit was assessed using the Hosmer–Lemeshow test. Adjusted odds ratios and 95% CIs were reported. All 1,195 participants had complete data for the model variables. As a sensitivity analysis for sparse data in the uninsured category, the same two models were refitted using Firth’s penalised likelihood logistic regression, retaining the same covariates and reference categories.

Qualitative methods

Participants

Participants were recruited using purposive and snowball sampling. Potential participants were approached face to face and were provided with information about the study before deciding whether to participate. The target group comprised PCPs directly involved in delivering community-based TCM health services to older adults in the participating communities. The inclusion criteria were: (1) direct involvement in the delivery of community-based TCM health services; (2) voluntary participation and provision of informed consent. The exclusion criterion was participation in other relevant studies during the research period. Three potential participants declined due to work-related reasons, and 11 PCPs participated in the study. Interviews continued until the research team judged that additional interviews were no longer generating information relevant to the study aim. This judgement was made provisionally after the ninth interview; two further PCPs were then interviewed, and no new subthemes relevant to the study aim were identified. This was treated as a pragmatic stopping point rather than as a claim of theoretical saturation.

Data collection

This qualitative study was conducted through in-depth interviews from June to July 2025. All interviews were conducted by YKY, a female researcher. Before each interview, YKY introduced herself to the participant as a researcher and explained that the study aimed to understand the delivery of community-based TCM health services from the perspective of primary care providers. Interviews were conducted in a private room at the participants’ community health service centres, with no other individuals present.

A preliminary semi-structured interview guide was developed based on a review of the literature on TCM health services and consultations with clinicians (Supplementary file B). Following preliminary analysis of the quantitative survey findings, the guide was further refined with particular attention to insurance-related differences, number of chronic conditions, awareness of community-based TCM health services, and perceived convenience. The qualitative interviews aimed to explore PCPs’ service-delivery experiences and their perspectives on provider-side conditions relevant to these quantitative findings. The interviews were not intended to elicit older adults’ own motivations, preferences, affordability concerns, or service experiences. The revised guide was pilot-tested with three participants to improve the clarity and relevance of the questions.

With the consent of the interviewees, all interviews were audio-recorded for approximately 30 ~ 60 min. Brief field notes were recorded during and immediately after the interviews. No repeat interviews were conducted. During the interviews, YKY used follow-up questions, requested clarification, and summarised or repeated key points when necessary to confirm her understanding of the participants’ accounts and obtain more complete information.

Data analysis

Within 24 h after the interviews, the researchers transcribed, organized, and repeatedly reviewed the interview recordings. NVivo 14 software was used to assist in the analysis. The transcripts were analysed using thematic analysis [28]. This approach was selected because the qualitative phase focused on recurring patterns in PCPs’ accounts of service-delivery barriers. Two researchers independently read the transcripts, generated initial codes, and grouped related codes into subthemes and themes. The research team then discussed and refined the developing themes, and differences in coding or interpretation were resolved through discussion and consensus.

Rigour and reflexivity

YKY conducted all interviews and had a professional background in nursing and training in qualitative research. YKY and YFQ both held master’s degrees, and YFQ was an academic researcher with experience in primary healthcare research in China. YKY had no prior professional relationship with the participants, and the research team had no prior professional or institutional relationship with the participating community health service centres.

The wider research team had backgrounds in nursing, primary healthcare research, and TCM-related academic institutions. The team recognised that these professional and institutional backgrounds could influence the formulation of interview questions, coding priorities, and interpretation of the data. To reduce the influence of preconceptions, the interviewer used open-ended, non-leading questions and did not express personal views about TCM services during the interviews. Peer debriefing and research-team meetings were also used to compare alternative interpretations and to check the developing themes against the original transcripts and supporting quotations.

To further enhance the credibility of the qualitative findings, a summary of the preliminary themes and interpretations was provided to three participants for member checking. All three considered the interpretations consistent with their interview accounts, and no substantive changes to the thematic structure or interpretations were required. The research team also examined accounts that did not fit the developing patterns and retained variations in participants’ emphasis on reimbursement, staffing, equipment, and health communication.

Integration

After completing separate analyses of the quantitative and qualitative data, the findings were integrated at the interpretation stage using a joint display and interpretive comparison. Quantitative findings concerning insurance status, number of chronic conditions, education, household income, awareness of services, and perceived convenience were compared with PCP-reported themes concerning reimbursement and coordination, workforce and material resources, and community communication and acceptance. Each comparison was classified as complementary, discordant, or without a direct counterpart. Because the two phases involved different respondent groups, provider accounts were not treated as direct explanations of individual older adults’ behaviour or as evidence of causal mechanisms. The integrated interpretations were used to identify questions and contextual hypotheses for further study.

Ethical statement

This study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. Ethical approval for both the quantitative survey of older residents and the qualitative interviews with primary healthcare providers was obtained from the Ethics Committee of the Nursing College, Hubei University of Chinese Medicine (approval number 2025024). Before participation, all older residents and primary healthcare providers were informed of the purpose, procedures, and potential risks of the study and provided written informed consent.

Results

Quantitative study

Participant characteristics and reported use of community-based TCM health services during the past 12 months

Of the 1,300 questionnaires distributed, 1,195 were retained for the final analysis, corresponding to 91.92% of the questionnaires distributed. As shown in Tables 1, 729 of the 1,195 participants (61.00%) reported receiving at least one community-based TCM health service during the past 12 months, whereas 466 (39.00%) reported receiving none of the listed services during this period. The sample included 641 men (53.64%) and 554 women (46.36%). Most participants were aged 60–80 years (1,107, 92.64%). One chronic condition was the most common category (470, 39.33%), followed by two conditions (326, 27.28%), no chronic condition (279, 23.35%), and three or more conditions (120, 10.04%).

Table 1.

Socio-demographic characteristics of the participants (n = 1,195)

Variable	Category	Sample size, n (%)	Participants reporting use, n	Reported use, %	χ²	P value
Sex	Male	641 (53.64)	408	63.65	4.07	0.044
	Female	554 (46.36)	321	57.94
Age group	60–70 years	794 (66.44)	488	61.46	0.21	0.901
	71–80 years	313 (26.19)	188	60.06
	≥ 81 years	88 (7.36)	53	60.23
Marital status	Married	889 (74.39)	539	60.63	2.12	0.346
	Divorced	99 (8.28)	67	67.68
	Widowed	207 (17.32)	123	59.42
Education level	No schooling	170 (14.23)	74	43.53	58.69	< 0.001
	Primary school	260 (21.76)	129	49.62
	Junior middle school	309 (25.86)	202	65.37
	Senior high/technical secondary school	239 (20.00)	165	69.04
	College or above	217 (18.16)	159	73.27
Medical insurance	Urban employee basic medical insurance	332 (27.78)	231	69.58	62.19	< 0.001
	Urban resident medical insurance	370 (30.96)	247	66.76
	New rural cooperative medical insurance	408 (34.14)	198	48.53
	Commercial medical insurance	57 (4.77)	45	78.95
	No medical insurance	28 (2.34)	8	28.57
Living arrangement	Living alone	177 (14.81)	96	54.24	—	0.117a
	With spouse	532 (44.52)	321	60.34
	With children/grandchildren	271 (22.68)	177	65.31
	With spouse and children/grandchildren	212 (17.74)	134	63.21
	With other relatives	3 (0.25)	1	33.33
Monthly household income (RMB)	≤ 1,999	133 (11.13)	46	34.59	52.99	< 0.001
	2,000–2,999	230 (19.25)	133	57.83
	3,000–4,999	322 (26.95)	213	66.15
	5,000–9,999	329 (27.53)	207	62.92
	≥ 10,000	181 (15.15)	130	71.82
Smoking status	Current smoker	328 (27.45)	196	59.76	10.95	0.004
	Former smoker	202 (16.90)	144	71.29
	Never smoker	665 (55.65)	389	58.50
Drinking status	Current drinker	436 (36.49)	268	61.47	16.29	< 0.001
	Former drinker	181 (15.15)	133	73.48
	Never drinker	578 (48.37)	328	56.75
Number of chronic conditions	No chronic condition	279 (23.35)	91	32.62	131.50	< 0.001
	One chronic condition	470 (39.33)	309	65.74
	Two chronic conditions	326 (27.28)	234	71.78
	Three or more chronic conditions	120 (10.04)	95	79.17
Awareness of community-based TCM health services	Not at all aware	334 (27.95)	38	11.38	509.32	< 0.001
	Slightly aware	132 (11.05)	113	85.61
	Moderately aware	351 (29.37)	244	69.52
	Quite aware	337 (28.20)	296	87.83
	Fully aware	41 (3.43)	38	92.68
Perceived convenience of accessing community-based TCM health services	No	314 (26.28)	163	51.91	14.80	< 0.001
	Yes	881 (73.72)	566	64.25
Perceived complexity of the service process	No	817 (68.37)	516	63.16	5.04	0.025
	Yes	378 (31.63)	213	56.35
Perceived price of community-based TCM health services as expensive	No	643 (53.81)	402	62.52	1.34	0.246
	Yes	552 (46.19)	327	59.24
Perceived policy support	No	430 (35.98)	248	57.67	3.13	0.077
	Yes	765 (64.02)	481	62.88

Note: Reported use was defined as self-reported receipt of at least one community-based TCM health service listed in Item 14 during the past 12 months; participants who selected “None of the above” were classified as non-users. Reported use (%) was calculated as the number of users divided by the total number of participants in each category × 100. Pearson’s χ² test was used unless otherwise specified. ᵃ Fisher’s exact test was used for living arrangement because of sparse expected cell counts

Among the 729 participants who reported having received at least one community-based TCM health service, oral Chinese herbal medicine was the most frequently reported service type (449, 61.59%), followed by Tuina/massage (316, 43.35%), external TCM therapies (312, 42.80%), acupuncture (304, 41.70%), and cupping (302, 41.43%). The complete distribution of service types is presented in Supplementary file C Table S1.

Bivariate analyses of reported use of community-based TCM health services during the past 12 months

Bivariate analyses showed that, within the study sample, the proportion reporting use of community-based TCM health services during the past 12 months differed significantly by sex (χ² = 4.07, P = 0.044), education level (χ² = 58.69, P < 0.001), medical insurance status (χ² = 62.19, P < 0.001), monthly household income (χ² = 52.99, P < 0.001), smoking status (χ² = 10.95, P = 0.004), drinking status (χ² = 16.29, P < 0.001), number of chronic conditions (χ² = 131.50, P < 0.001), awareness of community-based TCM health services (χ² = 509.32, P < 0.001), perceived convenience of accessing community-based TCM health services (χ² = 14.80, P < 0.001), and perceived complexity of the service process (χ² = 5.04, P = 0.025). Reported use was higher among men than women. The proportion reporting use was generally higher among participants with higher educational attainment and household income, although the pattern across income categories was not strictly monotonic. Across insurance groups, the observed proportion reporting use of community-based TCM health services was highest among participants with commercial medical insurance and lowest among those without medical insurance; however, the uninsured subgroup was small (n = 28), and the observed proportion for this subgroup should therefore be interpreted cautiously. Reported use was progressively higher across chronic-condition categories, from 32.62% among participants with no chronic condition to 79.17% among those with three or more conditions. It was higher among participants who perceived access to community-based TCM health services as convenient than among those who did not, and lower among those who perceived the service process as complex than among those who did not. The proportion reporting use was lowest among participants who were not at all aware of community-based TCM health services and highest among those who were fully aware; however, the intermediate awareness categories did not follow a strictly monotonic gradient. No statistically significant differences were observed by age group, marital status, living arrangement, perceived price of community-based TCM health services, or perceived policy support (all P > 0.05). For living arrangement, Fisher’s exact test was used because of sparse cell counts in the “With other relatives” category. Detailed results are presented in Table 1.

Multivariable logistic regression analyses

Both multivariable logistic regression models included all 1,195 participants. Model 1, which excluded awareness and was designated as the primary model, was statistically significant overall (likelihood-ratio χ² = 286.61, df = 31, P < 0.001). The Hosmer–Lemeshow test indicated no evidence of poor model fit (χ² = 5.21, df = 8, P = 0.735). Model 2, which additionally included awareness as a five-category variable, was also statistically significant overall (likelihood-ratio χ² = 858.09, df = 35, P < 0.001) and showed no evidence of poor fit according to the Hosmer–Lemeshow test (χ² = 8.47, df = 8, P = 0.389). Adding awareness significantly improved statistical fit compared with Model 1 (likelihood-ratio χ² = 571.48, df = 4, P < 0.001).

Within the study sample, the primary model excluding awareness showed that education level, monthly household income, number of chronic conditions, and medical insurance status were associated with reported use of community-based TCM health services during the past 12 months. Compared with participants with no schooling, those with junior middle school education (aOR = 1.96, 95% CI: 1.25–3.06, P = 0.003), senior high or technical secondary education (aOR = 2.11, 95% CI: 1.29–3.47, P = 0.003), and college education or above (aOR = 3.48, 95% CI: 2.00–6.04, P < 0.001) had higher odds of reporting service use. Compared with participants whose monthly household income was ≤ 1,999 yuan, all higher-income groups had higher odds of reported use, with aORs ranging from 2.40 to 3.62. The odds of reported use also increased across chronic-condition categories: the aORs were 5.38 (95% CI: 3.75–7.71) for one condition, 7.63 (95% CI: 5.11–11.39) for two conditions, and 9.72 (95% CI: 5.58–16.94) for three or more conditions. Participants without medical insurance had lower odds of reported use than those covered by urban employee basic medical insurance (aOR = 0.30, 95% CI: 0.11–0.77, P = 0.012).

In the exploratory extended model, awareness of community-based TCM health services was jointly associated with reported service use during the past 12 months (Wald χ² = 298.10, df = 4, P < 0.001). Compared with participants who were not at all aware, all four awareness categories had substantially higher odds of reported use, although the magnitude of the estimates did not follow a monotonic gradient. After awareness was included, college education or above, higher household income, number of chronic conditions, and lack of medical insurance remained associated with reported use. The estimated associations for chronic-condition categories increased substantially, from aORs of 5.38–9.72 in Model 1 to 18.12–67.41 in Model 2. Because awareness and service use were assessed concurrently, the associations observed in Model 2 should not be interpreted as evidence of temporal ordering or a causal effect of awareness on service use. Complete results for both models are presented in Table 2. Perceived convenience of access was not statistically associated with reported use in the primary model (aOR = 1.26, 95% CI: 0.92–1.74, P = 0.154). In the exploratory model including awareness, the corresponding estimate was 0.99 (95% CI: 0.64–1.55, P = 0.974).

Table 2.

Multivariable logistic regression models of factors associated with reported use of community-based TCM health services

Variable	Category	Model 1 aOR (95% CI)	P value	Model 2 aOR (95% CI)	P value
Sex	Male	Ref.		Ref.
	Female	0.94 (0.66–1.32)	0.708	0.89 (0.54–1.48)	0.654
Age group	60–70 years	Ref.		Ref.
	71–80 years	0.82 (0.59–1.13)	0.216	0.85 (0.54–1.35)	0.502
	≥ 81 years	0.93 (0.53–1.63)	0.797	1.04 (0.46–2.34)	0.930
Marital status	Married	Ref.		Ref.
	Divorced	1.37 (0.73–2.56)	0.325	1.09 (0.45–2.63)	0.853
	Widowed	1.22 (0.73–2.04)	0.44	1.25 (0.61–2.56)	0.544
Education level	No schooling	Ref.		Ref.
	Primary school	1.10 (0.71–1.71)	0.671	1.38 (0.70–2.73)	0.353
	Junior middle school	1.96 (1.25–3.06)	0.003	1.87 (0.95–3.65)	0.068
	Senior high/technical secondary	2.11 (1.29–3.47)	0.003	1.92 (0.92–4.00)	0.083
	College or above	3.48 (2.00–6.04)	< 0.001	4.44 (2.02–9.75)	< 0.001
Medical insurance	Urban employee basic MI	Ref.		Ref.
	Urban resident MI	1.01 (0.71–1.45)	0.942	0.93 (0.56–1.53)	0.771
	New rural cooperative MI	0.69 (0.47–1.00)	0.052	0.86 (0.50–1.47)	0.574
	Commercial MI	1.80 (0.85–3.82)	0.124	0.69 (0.25–1.87)	0.463
	No medical insurance	0.30 (0.11–0.77)	0.012	0.12 (0.03–0.40)	0.001
Living arrangement	Living alone	Ref.		Ref.
	With spouse	1.22 (0.73–2.04)	0.454	1.29 (0.62–2.72)	0.497
	With children/grandchildren	1.32 (0.84–2.08)	0.229	1.27 (0.67–2.42)	0.460
	With spouse and children/grandchildren	1.19 (0.67–2.11)	0.559	1.56 (0.67–3.62)	0.299
	With other relatives	2.16 (0.12–38.87)	0.602	0.90 (0.04–18.61)	0.947
Variable	Category	Model 1 aOR (95% CI)	P value	Model 2 aOR (95% CI)	P value
Monthly household income (RMB)	≤ 1,999	Ref.		Ref.
	2,000–2,999	2.40 (1.46–3.95)	0.001	2.40 (1.14–5.07)	0.021
	3,000–4,999	2.92 (1.79–4.76)	< 0.001	3.11 (1.50–6.44)	0.002
	5,000–9,999	2.76 (1.67–4.54)	< 0.001	3.06 (1.46–6.40)	0.003
	≥ 10,000	3.62 (2.02–6.49)	< 0.001	5.62 (2.36–13.40)	< 0.001
Smoking status	Current smoker	Ref.		Ref.
	Former smoker	1.43 (0.90–2.26)	0.129	1.39 (0.72–2.66)	0.325
	Never smoker	1.12 (0.76–1.64)	0.567	0.89 (0.51–1.56)	0.681
Drinking status	Current drinker	Ref.		Ref.
	Former drinker	1.27 (0.80–2.00)	0.308	1.10 (0.57–2.13)	0.771
	Never drinker	0.87 (0.61–1.25)	0.448	0.84 (0.50–1.43)	0.527
Number of chronic conditions	No chronic condition	Ref.		Ref.
	One chronic condition	5.38 (3.75–7.71)	< 0.001	18.12 (10.69–30.71)	< 0.001
	Two chronic conditions	7.63 (5.11–11.39)	< 0.001	34.06 (18.41–62.98)	< 0.001
	Three or more chronic conditions	9.72 (5.58–16.94)	< 0.001	67.41 (29.49–154.12)	< 0.001
Perceived convenience of accessing community-based TCM health services	No	Ref.		Ref.
	Yes	1.26 (0.92–1.74)	0.154	0.99 (0.64–1.55)	0.974
Perceived price of community-based TCM health services as expensive	No	Ref.		Ref.
	Yes	1.03 (0.78–1.36)	0.832	1.03 (0.69–1.53)	0.886
Perceived policy support	No	Ref.		Ref.
	Yes	1.03 (0.77–1.39)	0.825	0.68 (0.45–1.04)	0.075
Awareness of community-based TCM health services	Not at all aware	Not included		Ref.
	Slightly aware	Not included		138.85 (63.49–303.65)	< 0.001
	Moderately aware	Not included		40.90 (24.02–69.66)	< 0.001
	Quite aware	Not included		209.71 (108.39–405.74)	< 0.001
	Fully aware	Not included		717.43 (155.17–3316.98)	< 0.001

Note: The outcome was self-reported use of at least one community-based TCM health service during the past 12 months (1 = user; 0 = non-user). Model 1 was the primary model; Model 2 additionally included awareness as a five-category variable and was exploratory. aOR, adjusted odds ratio; CI, confidence interval; Ref., reference category. Detailed variable coding is provided in Supplementary file C. Estimates for participants living with other relatives and those without medical insurance should be interpreted cautiously because of small subgroup sizes. The chronic-condition variable was collapsed into four categories. The number of chronic conditions was categorised as none, one, two, or three or more conditions

Firth penalised logistic regression sensitivity analysis

In the Firth penalised logistic regression sensitivity analysis, participants without medical insurance continued to have lower odds of reporting use of community-based TCM health services during the past 12 months than those covered by urban employee basic medical insurance. The association remained statistically significant in both the model excluding awareness (aOR = 0.32, 95% CI: 0.13–0.81, P = 0.016) and the model including awareness (aOR = 0.14, 95% CI: 0.04–0.44, P = 0.001). These estimates were broadly consistent with the corresponding conventional logistic regression estimates of 0.30 (95% CI: 0.11–0.77, P = 0.012) and 0.12 (95% CI: 0.03–0.40, P = 0.001), respectively. Nevertheless, the uninsured subgroup was small (n = 28, including eight participants reporting service use), and the estimates should be interpreted cautiously. Detailed results are presented in Table 3.

Table 3.

Sensitivity analysis using Firth’s penalised likelihood logistic regression: adjusted association between having no medical insurance and reported use of community-based TCM health services

Model / estimation method	No medical insurance aOR (95% CI)	P value
Conventional logistic Model 1	0.30 (0.11–0.77)	0.012
Firth Model 1	0.32 (0.13–0.81)	0.016
Conventional logistic Model 2	0.12 (0.03–0.40)	0.001
Firth Model 2	0.14 (0.04–0.44)	0.001

Qualitative study

Eleven PCPs participated in the study, with details presented in Table 4. Their ages ranged from 28 to 50 years old, and they had diverse professional backgrounds, including physicians, rehabilitation therapists, pharmacists, and nurses. The work experience of the participants spanned from 3 to 25 years. After the ninth interview, the research team provisionally judged that additional interviews were no longer generating information relevant to the study aim. Two further PCPs were interviewed, and no new subthemes relevant to the study aim were identified. This was treated as a pragmatic stopping point rather than as evidence of theoretical saturation. Table 4 lists the individual characteristics of the participants.

Table 4.

Characteristics of the participants (n = 11)

Coding	Age (years)	Gender	Original profession	Degree	Position	Work experience (years)
P1	38	Female	TCM Physician	Master’s	PCP	10
P2	45	Male	Rehabilitation Therapist	Undergraduate	PCP	15
P3	32	Female	Pharmacist	Undergraduate	PCP	6
P4	50	Male	TCM Physician	Junior college	PCP	25
P5	42	Female	TCM Physician	Undergraduate	PCP	12
P6	35	Male	TCM Physician	Master’s	PCP	7
P7	32	Female	Nurse	Undergraduate	PCP	5
P8	33	Female	Nurse	Junior college	PCP	8
P9	37	Female	Rehabilitation Therapist	Undergraduate	PCP	12
P10	28	Female	Nurse	Undergraduate	PCP	3
P11	40	Female	Nurse	Junior college	PCP	15

Three major themes were identified: (1) Lack of a Sound Policy and Support System; (2) Failure to Secure Adequate Human and Material Resources; (3) Barriers to Community Awareness and Acceptance of TCM Healthcare Services (Table 5).

Table 5.

Themes and subthemes

Themes	Subthemes
Lack of a sound policy and support system	Limited insurance hinders TCM services
	Fragmentation stems from poor coordination
Failure to secure adequate human and material resources	Workforce shortages
	Insufficient training systems
	Outdated or inadequate equipment
	Limitations of community venues
Barriers to community awareness and acceptance of TCM health services	Ineffective health communication strategies
	Persistent cultural and cognitive biases

Theme 1: Lack of a sound policy and support system

Although national policies have increasingly emphasized the development of TCM and the strengthening of primary health care, the PCPs interviewed in this study consistently reported that, in their own practice, the delivery of community-based TCM health services was constrained by what they perceived as incomplete policy frameworks and limited institutional support. Participants described TCM service delivery across health promotion, chronic disease management, and preventive care as depending not only on a skilled workforce but also on systematic policy guidance, sustained funding, and interdepartmental collaboration. Two issues recurred across their accounts. Participants reported that health insurance coverage for TCM items was narrow, which they perceived as limiting the accessibility and scalability of these services. They also reported that coordination among the relevant departments was weak, which they perceived as impeding resource integration. These findings reflect the perceptions of 11 providers working in the participating centres and are presented as provider accounts rather than as an assessment of the health system itself.

Sub-Theme 1: Limited insurance hinders TCM services

Participants reported that the limited coverage of TCM health services under the current health insurance scheme was a barrier to the provision and accessibility of these services for older adults. In their accounts, health insurance was described not only as protecting basic medical needs but also as shaping which services older adults chose to use. Several participants suggested that broader inclusion of TCM items in insurance coverage would increase use and reduce out-of-pocket payments. Participants reported that, in their institutions, only a small portion of TCM treatments-such as acupuncture and therapeutic massage-are reimbursed, and that reimbursement rates for these items were low. They further reported that commonly used services such as constitution identification, health consultation, and moxibustion were generally not reimbursed, and perceived this as making service delivery difficult to sustain. These statements describe providers’ perceptions of local reimbursement arrangements; they were not verified against insurance schedules and should not be read as a description of insurance policy.

In terms of medical insurance compensation, some items can be reimbursed, but the proportion and scope of reimbursement need to be improved. (P1)

The coverage of medical insurance is narrow, and some rehabilitation programs are not covered. (P2)

There are few items of medical insurance compensation involving TCM nursing, only part of moxibustion and massage nursing items can be reimbursed, and the reimbursement amount is not high, which limits the promotion of TCM nursing services. (P8)

The medical insurance compensation covers relatively few items related to TCM healthcare. Only a few items such as moxibustion and massage can be reimbursed, and the reimbursement ratio is not high either. (P10)

Sub-Theme 2: Fragmentation stems from poor coordination

Participants reported that coordinated collaboration across departments and administrative levels was necessary for delivering TCM health services in community settings, but that, in their experience, structured interdepartmental communication mechanisms were underdeveloped. They described TCM initiatives as involving internal departments such as TCM, rehabilitation, and nursing, as well as external administrative bodies including medical insurance, public health, and elderly affairs offices. Participants perceived that the absence of such coordination affected continuity of care, information sharing, and resource allocation, and they reported fragmented management, inefficient referral pathways, separate information systems, and unclear allocation of responsibilities between departments.

One of the challenges we face is insufficient coordination with other departments. A possible solution would be to strengthen interdepartmental communication and establish a more structured referral mechanism. (P1)

Another issue is the lack of collaboration with higher-level medical institutions. It is difficult to carry out referrals or consultations for complicated cases. (P2)

In addition to the previously mentioned issues, we also face poor collaboration with other departments. For example, when providing rehabilitation services to the elderly, we need close cooperation with the TCM and nursing departments. However, communication and coordination mechanisms are underdeveloped, which affects service continuity and quality. (P4)

One challenge is the insufficient integration between TCM and Western nursing practices. We suggest offering more TCM-specific training for nurses and organizing regular knowledge exchange sessions to better integrate both approaches. (P5)

Theme 2: Failure to secure adequate human and material resources

Participants reported shortages of both human and material resources in the institutions in which they worked. In relation to human resources, they described a shortage of trained professionals, particularly in TCM nursing and rehabilitation. In relation to material resources, they reported outdated or insufficient equipment, limited diagnostic tools, and constrained service venues. Participants perceived these shortages as limiting the range, efficiency, and continuity of the TCM health services they were able to offer to older adults.

Sub-Theme 1: Workforce shortages

Participants reported that shortages of professional staff were a recurrent problem in their institutions, particularly in TCM rehabilitation, preventive care, and nursing.

Human resources are insufficient, and there are few professional TCM prevention and health care personnel. (P5)

Human resources are insufficient, and there is a shortage of TCM professionals. (P6)

There were few qualified TCM nurses, and most staff had western medicine nursing backgrounds and lacked systematic TCM nursing training. (P8)

There is a shortage of human resources, difficulty recruiting qualified TCM nurses, and a heavy workload for existing staff. (P11)

Sub-Theme 2: Insufficient training systems

Several PCPs reported that, although they are responsible for delivering TCM healthcare services to older residents, they have not received systematic training from qualified TCM professionals. Participants also reported that many nurses in their institutions had been trained in Western-medicine settings and, in their view, lacked sufficient TCM-specific knowledge.

Compound nursing talents who are proficient in both nursing professional knowledge and familiar with TCM knowledge are relatively scarce. (P7)

There are relatively few professional TCM nurses. Most nurses have a background in Western medicine nursing and lack systematic training in TCM nursing knowledge. (P8)

There are relatively few professional TCM nursing staff. Most of the nursing staff still mainly rely on Western medicine nursing knowledge, and there are also few opportunities for TCM nursing skills training. (P10)

Sub-Theme 3: Outdated or inadequate equipment

Participants reported that limited financial support for TCM development in their institutions was accompanied by constrained service space and a limited number of TCM diagnostic, treatment, and rehabilitation devices. They perceived that these constraints meant that residents’ requests could not always be met and that the range of TCM health services they could provide was restricted.

There are relatively few Chinese medicine pharmacists, and errors are more likely during busy periods. The decoction equipment in the pharmacy is also ageing and needs to be replaced, but funding is limited. (P3)

Some advanced rehabilitation equipment is too costly for community health service centers, and existing equipment is insufficient in both quantity and function, affecting treatment effectiveness and efficiency. (P9)

The number of devices such as moxibustion machines and massage beds is limited, sometimes requiring patients to wait. (P10)

The shortage of funds for purchasing equipment has restricted the renewal and replenishment of the equipment. (P11)

Sub-Theme 4: Limitations of community venues

Participants reported constraints on physical space and variable hygiene conditions in some community settings. They perceived that inconsistent environmental standards affected the implementation of standardised nursing procedures and the preparation of treatment areas.

The hygienic conditions of community sites are uneven, which is not conducive to doing nursing work strictly according to the standards. (P7)

When carrying out activities in community activity centers, the preparation of venues and equipment is cumbersome. (P10)

Theme 3: Barriers to community awareness and acceptance of TCM health services

According to the descriptions of PCPs, there is an obvious sense of distance between PCPs and patients. In their accounts, many of the older residents they saw were unfamiliar with the community-based TCM health services available to them, and participants attributed this partly to what they regarded as insufficient publicity by their own institutions. Because these observations were reported by providers rather than by older adults, they represent providers’ perceptions of residents’ awareness, trust, and acceptance, and are not evidence of older adults’ own views. No data on awareness, trust, or reasons for non-use were collected directly from older adults in the qualitative phase.

Sub-Theme 1: Ineffective health communication strategies

Participants reported that TCM publicity in their communities consisted mainly of health lectures and free-clinic sessions. They described these formats as limited in variety and, in their view, constrained by time and workload, so that residents’ questions could not always be answered promptly; participants perceived the resulting effect to be small. Participants also reported that the older residents who attended their centres obtained information about TCM mainly from television, radio, relatives, and friends, and perceived that this produced a limited and unsystematic understanding of community-based TCM health services.

The publicity efforts were general, mainly through community bulletin boards and WeChat groups. (P1)

Propaganda efforts are not enough, propaganda methods are relatively single. There were health lectures and free clinic activities, but the publicity was not enough, and many residents did not know the information about the activities. (P2)

Although health lectures and free clinic activities are also carried out, the frequency is not large and the publicity is not in place, so many residents do not know about it. And the way of publicity is more old-fashioned, basically pull banners, leaflets, the effect is not very good. (P4)

At present, the publicity is mainly carried out through community bulletin boards, WeChat groups and other ways, but the publicity method is relatively single and the coverage is limited. (P9)

Sub-Theme 2: Persistent cultural and cognitive biases

Several PCPs reported that, in their perception, many residents had limited awareness of TCM and regarded TCM treatment as slow-acting and of uncertain efficacy, and were therefore not confident in it. Participants also reported that publicity for TCM nursing and preventive care was, in their view, weak, that they lacked effective ways of communicating the TCM concept of preventive treatment of disease and the advantages of these services to older adults. As above, these are provider perceptions rather than reports from older adults themselves.

Some people think that TCM is slow and unwilling to try it. (P1)

The problem is that the residents do not pay enough attention to the prevention and health care of TCM. (P5)

The frequency of free clinic activities is relatively low, and the form of health lectures is relatively traditional. Most of them are unilateral explanations by medical staff, with insufficient interaction, and the enthusiasm of residents to participate is not particularly high, resulting in unsatisfactory publicity effect. (P7)

Integrated findings

Integration was undertaken at the interpretation stage. Selected quantitative findings concerning educational attainment, monthly household income, medical insurance status, number of chronic conditions, and awareness of services were considered alongside the service-delivery barriers reported by providers. Table 6 presents areas of complementarity, discordance, and findings without a qualitative counterpart. Because the two phases involved different participant groups and the survey was cross-sectional, the provider accounts are used only to provide contextual information and do not explain individual older adults’ service use or establish causal pathways.

Table 6.

Joint display of quantitative findings, qualitative themes, and integrated interpretations

Quantitative	Qualitative	Relationship	Integrated interpretation
Participants without medical insurance had lower adjusted odds of reported use.	Providers reported that reimbursement coverage for TCM items was limited and that reimbursement rates for some covered items were low.	Discordance	The provider accounts do not directly explain the association between insurance status and reported use. The quantitative estimate was based on a small uninsured subgroup and should be interpreted cautiously.
Reported use increased across chronic-condition categories.	Providers reported shortages in staffing, training, equipment, and service venues.	Complementarity	The qualitative findings describe the service-delivery context but do not explain the association between chronic-condition burden and reported use.
Higher educational attainment and household income were associated with higher odds of reported use.	The provider interviews did not directly address individual users’ education or income.	No qualitative counterpart	The qualitative data collected from providers could not explain these socioeconomic associations.
Awareness was associated with reported use in the exploratory model.	Providers reported limited variety, coverage, and interactivity in community health communication.	Complementarity	The two findings concern service awareness and communication, but their temporal relationship cannot be determined because awareness and use were measured concurrently.

The insurance-related findings were classified as discordant. In the survey, participants without medical insurance had lower adjusted odds of reported use, whereas providers reported limited reimbursement coverage and low reimbursement rates for some TCM items. These findings were obtained from different respondent groups and addressed different aspects of insurance and reimbursement; therefore, the provider accounts cannot explain the quantitative association between insurance status and reported use. The uninsured subgroup was small, and the estimate should be interpreted cautiously. Educational attainment and monthly household income were associated with reported use but had no direct qualitative counterpart because the provider interviews focused on service-delivery conditions rather than individual users’ socioeconomic characteristics.

Discussion

This explanatory sequential mixed-methods study described reported use of community-based TCM health services during the past 12 months among 1,195 older adults recruited during designated survey sessions at seven community health service centres and explored service-delivery barriers reported by 11 primary care providers. Within this site-recruited sample, 61.0% reported receiving at least one listed service. Among those reporting service use, oral Chinese herbal medicine was the most frequently reported service type, followed by Tuina/massage, external TCM therapies, acupuncture, and cupping. In the primary model excluding awareness, higher educational attainment, higher monthly household income, and a greater number of chronic conditions were associated with higher odds of reported use, whereas the small subgroup without medical insurance had lower odds. The interviewed providers reported perceived barriers related to policy and organisational support, workforce and material resources, and community health communication. Because the two phases involved different participant groups and the survey was cross-sectional and based on on-site recruitment, the findings should be interpreted as distinct but potentially complementary evidence. The provider accounts do not demonstrate the causes or mechanisms underlying individual older adults’ service use.

Reported use increased across chronic-condition categories. This pattern is consistent with an explanatory mixed-methods study of Chinese older adults with multimorbidity, which reported greater health-service use among those with more coexisting conditions [18], and with a cross-sectional survey of Chinese patients with chronic diseases, which found greater use of TCM therapies among those with a higher disease burden [29]. Greater healthcare needs and more frequent contact with health services may partly explain this pattern, although these mechanisms were not directly measured and the temporal direction of the association cannot be established. The chronic-condition estimates increased substantially after awareness was added, from aORs of 5.38–9.72 in the primary model to 18.12–67.41 in the exploratory model, indicating sensitivity to model specification. Because awareness was measured concurrently with service use and might either precede use or increase following previous service contact, the model excluding awareness was treated as the primary analysis and the extended model as exploratory.

Higher educational attainment and monthly household income were also associated with higher odds of reported use. These variables may reflect broader socioeconomic differences in access to and engagement with health services, although the present study did not directly assess mechanisms such as health literacy, service navigation, or affordability. Participants without medical insurance had lower adjusted odds of reported use than those covered by urban employee basic medical insurance, and the direction of this estimate was retained in the Firth analysis. However, the uninsured subgroup included only 28 participants, of whom eight reported service use, and the estimate should therefore be interpreted cautiously. Participants with commercial insurance had the highest crude proportion of use, but their adjusted estimates were non-significant and changed direction across the two models; no substantive inference was therefore drawn for this subgroup. A province-level ecological study reported variation in TCM service utilisation according to regional socioeconomic, demographic, health-status, and healthcare-resource characteristics [30], but those contextual findings are not directly comparable with the individual-level associations observed here.

Perceived policy support was not statistically associated with reported use in either model. Although the point estimate was below 1 in the exploratory model, its 95% CI included 1. This pattern differs from the prominence of policy and reimbursement constraints in the provider interviews; however, the two findings were derived from different respondent groups and represented different constructs. They should therefore be regarded as non-equivalent or discordant findings rather than as direct confirmation or contradiction.

The interviewed providers reported that health insurance coverage for TCM items was narrow, that reimbursement rates were low, and that coordination among the relevant departments was weak, and they perceived these conditions as affecting the continuity and scalability of the services they delivered. Comparable provider-reported constraints have been described among village doctors in Shandong Province, where reimbursement arrangements and institutional support were among the factors associated with the promotion of appropriate TCM techniques [31]. Participants also reported shortages of TCM nursing and rehabilitation staff, limited systematic TCM training, and constraints in equipment and service venues. These accounts are consistent with a cross-sectional survey reporting low TCM health literacy among rural older adults in China [32], with the evaluation of a large-scale TCM nursing knowledge and skills training programme delivered to community nurses in Shanghai, which was introduced in response to precisely such training gaps [33], and with a social network analysis identifying personnel and resource support as central to the development of grassroots TCM services [34]. Finally, participants reported that community publicity was limited in variety, coverage, and interactivity, and perceived that some residents regarded TCM as slow-acting and of uncertain efficacy. It should be emphasised that all statements in this study about residents’ awareness, trust, and acceptance are providers’ perceptions. Older adults’ own reasons for using or not using these services were not collected: the survey contained no items on trust or on reasons for non-use, and the qualitative phase did not include older adults. The provider accounts therefore describe the conditions under which services were delivered; they cannot explain older adults’ behaviour.

Taken together, the two phases describe, on one side, individual-level characteristics associated with reported use within the study sample and, on the other, service-delivery conditions reported by providers working in comparable settings. The study does not establish that the conditions reported by providers produced the differences in reported use observed in the survey, and the findings should not be read as demonstrating a demand-supply gap or as identifying structural mechanisms. They may nevertheless be useful for generating hypotheses and for local service planning. Areas that could be examined in future research, and that local planners may wish to consider, include the scope and level of insurance reimbursement for appropriate TCM service items, coordination arrangements between primary care institutions and the relevant administrative departments, training and staffing for TCM nursing and rehabilitation, and the design of community health communication. Establishing whether changes in these areas are followed by changes in service use would require longitudinal or evaluative designs, population-based sampling frames, and data collected directly from older adults.

Limitations

This study has several limitations. First, random selection was applied only at the survey-site level, whereas individual participants were recruited on site during designated survey sessions at seven community health service centres. Older adults who did not attend these sessions were not represented, and information on non-attendees and the catchment populations of the participating centres was unavailable. The proportion of 61.0% should therefore be interpreted as a characteristic of the study sample rather than as a population-level estimate for older adults in Wuhan. In addition, centre identifiers were not retained in the de-identified dataset, so within-centre correlation could not be accounted for, which may have affected the precision of the standard errors and confidence intervals. Second, service use was self-reported for the preceding 12 months and defined as receipt of at least one listed service. This binary measure may be affected by recall bias and does not capture the frequency, continuity, or intensity of service use. The cross-sectional design prevents determination of temporal ordering or causal relationships. Awareness was measured concurrently with service use, and the substantial changes in several estimates after awareness was added indicate sensitivity to model specification; the model including awareness was therefore treated as exploratory. Third, some insurance categories contained few participants. In particular, the uninsured subgroup comprised 28 participants, of whom eight reported service use. Although Firth’s penalised likelihood was used as a sensitivity analysis, the corresponding estimates remained imprecise and should be interpreted cautiously. Finally, the quantitative and qualitative phases involved different participant groups. The provider interviews described service-delivery conditions from the providers’ perspectives but could not explain older adults’ own motivations, preferences, or service experiences. The qualitative sample was recruited from community health service centres in one city, which may limit the range and transferability of the findings. In addition, although the study questionnaire underwent expert consultation and pilot testing, test–retest reliability was not assessed.

Conclusions

This study found that more than half of the older adults recruited on site at seven community health service centres reported receiving community-based TCM health services during the preceding 12 months, with reported use varying across participants with different socioeconomic and health characteristics. Primary care providers also reported multiple challenges in delivering these services. Because the two phases involved different participant groups and the survey used a cross-sectional design with on-site recruitment, the findings do not explain individual older adults’ reasons for using or not using these services. Future studies should use more representative sampling approaches and directly examine older adults’ service-use experiences and perspectives.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AIC

Akaike information criterion

aOR

Adjusted odds ratio

BIC

Bayesian information criterion

CI

Confidence interval

COREQ

Consolidated Criteria for Reporting Qualitative Research

CVI

Content validity index

GRAMMS

Good Reporting of A Mixed Methods Study

MI

Medical insurance

PCP

Primary care provider

RMB

Renminbi

S-CVI/Ave

Scale-level content validity index based on the average method

TCM

Traditional Chinese medicine

Author contributions

Y.-K. Y.: Methodology, Formal analysis, Investigation, Data curation, Writing-original draft, Writing-review & editing. J.-Y. F.: Data curation, Writing-review & editing. J. G.: Methodology and Investigation. M.-J. T.: Data curation, Writing-review& editing. Y.-F. Q.: Conceptualization, Methodology. J.-L. L.: Investigation. J.-X. D.: Conceptualization, Supervision. F. Y.: Conceptualization, Supervision.

Funding

This study was supported by the National Center for Mental Health and Mental Illness Prevention Project (Grant: GXDXK2025B-12).

Data availability

The de-identified survey dataset and variable coding file are available from the corresponding author upon reasonable request. They have not been deposited in a public repository because the ethical approval for this study did not cover public data deposition. The interview transcripts are not available because they contain potentially identifiable information and are subject to confidentiality restrictions.

Declarations

Ethics approval and consent to participate

This study protocol was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. Ethical approval was obtained from the Ethics Committee of the Nursing College at Hubei University of Chinese Medicine (approval number 2025024). All participants were fully informed about the purpose, procedures, and potential risks of the study, and each provided written informed consent prior to participation.

Consent for publication

Not applicable.

Declaration of generative AI use

During the preparation of this work, the authors used ChatGPT solely for language editing and polishing to improve grammar, clarity, and readability. Generative AI was not used for the literature review, study design, randomization, data analysis, interpretation of results, or generation of scientific conclusions. After using this tool, the authors carefully reviewed and edited the manuscript as needed and take full responsibility for the content of the publication.

Competing interests

The authors declare no competing interests.