Abstract
Background
Stroke prehospital time is exceptionally long in China. The multifaceted Stroke 1-2-0 Amplifying Version Engagement (SAVE) intervention has substantially reduced onset-to-door time for patients with acute stroke in Shanghai, China. However, there is limited evidence regarding the economic impacts of community educational interventions on reducing hospital delays. Therefore, this study aimed to assess the implementation and downstream healthcare costs associated with the educational program.
Methods
Data were extracted anonymously from the hospital information systems for all ischemic stroke patients admitted to the hospital from 2016 to 2019. Ischemic stroke patients who presented to the hospital within 2 days of stroke were included in the analysis. Cost data from different years were converted into 2019 Chinese Yuan (CNY) using a 5% discount rate based on China Guidelines for Pharmacoeconomic Evaluations. We used a multivariate generalized linear model (GLM) with a log-link to examine the intervention’s impact on length of stay (LOS) and costs.
Results
The SAVE intervention was estimated to cost CNY 1,768,067 for Xinzhuang County from October 2016 to December 2019. Most costs (59.51%) were for the mass media broadcast. 2,830 stroke patients met the inclusion criteria, including 490 in the pre-SAVE and 2,340 in the post-SAVE period. Following the multifaceted SAVE intervention, the mean (± standardized deviation [SD]) LOS decreased from 9.48 (± 3.80) days to 8.80 (± 3.55) days, representing a reduction of 7.17%. The mean (± SD) hospitalization costs per patient dropped from CNY 21,951 (± 10,411) to CNY 19,263 (± 12,773) (P < 0.001), with a 12.25% reduction in hospitalization costs. In the GLM model, the intervention was associated with reduced LOS (β coefficient [95% CI]: -0.070 [-0.11 to -0.030]; P < 0.001) and hospitalization costs (β coefficient [95% CI]: -0.11 [-0.17 to -0.053]; P < 0.001).
Conclusions
Our findings suggest that the SAVE intervention was associated with potential cost-offsets through its effect on reducing LOS and hospitalization costs. To further enhance stroke awareness at the national level, innovative approaches to educational delivery, such as mobile-based platforms and AI-enhanced strategies, represent promising hypotheses that warrant future evaluation.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12913-026-14191-0.
Keywords: Stroke 1-2-0, Intervention, Economic evaluation, Prehospital delay, LMICs
Main text
Stroke remains a leading cause of death globally; it is also ranked as the number one cause of death and disability in some low- and middle-income countries (LMICs), particularly in China [1]. Stroke death and disability are attributable to two primary barriers: failure to identify and failure to rescue [2]. Reperfusion therapy by intravenous thrombolysis or endovascular mechanical thrombectomy improves the likelihood of disability-free recovery after acute ischemic stroke (AIS) but should be initiated within 6 h of stroke onset [3–5]. The use of reperfusion therapy remains very low (< 10%) in China, partly due to poor awareness and significant prehospital delay, many people are disabled or even die without receiving any time-sensitive reperfusion therapy [6–8]. Recent data indicated that the median stroke onset-to-door time was around 24 h in China [9]. Super long stroke onset-to-door delay should be considered a global crisis, especially in LMICs [8]. Hence, reducing prehospital delay by improving laypeople’s symptom recognition is urgently needed [10, 11].
The lack of knowledge awareness about stroke signs or the emergency of treatment was one of the most frequently-reported barriers to acute stroke care [12, 13], particularly in China, where less than 50% of residents could recognize the stroke signs correctly. To improve stroke knowledge, the Stroke 1-2-0 tool was developed to help laypersons identify stroke warning signs by adapting the Face, Arm, Speech Test (F.A.S.T.) to the local emergency phone number of 1-2-0 in China [14]. Although FAST has been introduced to China for many years [15], the English word “FAST” is relatively difficult to remember for Chinese people due to linguistic barriers [16], limiting its effects in improving stroke awareness. In a national survey, four in five residents considered the Stroke 1-2-0 tool simple, practical, and easy to remember [17].
The community-based educational campaigns are an important strategy to promote stroke recognition tools and shorten the prehospital time in some high-income countries. Still, such programs are scarcely available in LMICs [18]. To promote the Stroke 1-2-0 tool, we designed a multilevel educational program entitled “Stroke 1-2-0 Amplifying Version Engagement (SAVE)”, which aims to encourage immediate action when any stroke symptoms occur and improve stroke outcomes in China [17]. In the pilot study covering Xinzhuang County of Shanghai, the SAVE intervention demonstrated promising effectiveness in reducing prehospital time [19]. However, limited data is available on the implementation costs and their downstream healthcare costs, which further impedes the adoption of health educational programs [18, 20]. Hence, this study aimed to fill gaps in the literature by comprehensively assessing the costs and resources needed to develop and execute the stroke educational program in China. This study also evaluated the impacts of the SAVE intervention on healthcare utilization and costs.
Methods
Study design
This study employed a pre-post observational study design. The effects of the SAVE intervention, in terms of the 3-hour hospital arrival rate, have been published elsewhere [19]. The SAVE intervention was implemented on October 1, 2016; the pre-SAVE period was defined as January 2016 to September 2016, and the post-SAVE period was defined as November 2016 to December 2019. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cohort studies [21].
Data source
Following the previously published study [19], data were extracted anonymously from the hospital information systems for all ischemic stroke patients admitted to the hospital from January 1, 2016, to December 31, 2019. As a designated member of the China Stroke Center Alliance, Minhang Hospital serves as the sole stroke center in Xinzhuang County, which has a population of approximately 250,000 residents. Patient demographic and clinical data were routinely collected within 24 h of admission for quality assurance purposes. To ensure data quality, the quality assurance procedure consisted of logic checks in the electronic data entry as well as annual audits of medical records and personnel training. All identifiable personal information, such as names, addresses, and phone numbers, was deleted to protect their privacy and prevent any potential misuse of their data.
We retrieved the following data from the electronic medical records, including age, sex, date and time of hospital admission, date and time of symptom onset, mode of transportation to the hospital, and National Institutes of Health Stroke Scale (NIHSS). To collect clinical data, patients or their caregivers were interviewed for their medical history, hospital arrival, and administered the NIHSS. The NIHSS score of 1–4 was categorized as minor stroke; 5–14 for moderate stroke; and ≥ 15 for major stroke [22].
Ethics approval
This study was reviewed and approved by the ethics committees of the Institutional Review Boards of Minhang Hospital of Fudan University.
Overview of SAVE intervention
The SAVE intervention was introduced to Xinzhuang, Shanghai, using a multifaceted outreach campaign implementation strategy, including broadcasting videos, printed materials, local news, and face-to-face community education, which has been reported elsewhere [19]. In brief, the SAVE intervention included: (1) The television campaign ran through free media advertising on the local TV Channel. The one-minute animated video, which was endorsed by the Chinese Stroke Association (CSA) , was broadcast six times per day. (2) The radio campaign ran through free radio advertising on the local station. The radio advertisement was broadcast six times per day. (3) The newspaper advertised the Stroke 1-2-0 pictures in the local newspaper. (4) The public poster campaign distributed 30,000 brochures to residents each year. (5) Face-to-face education sessions were offered by primary physicians every week.
Sample selection
As described previously [19], AIS patients who met the following criteria were included in the analysis: (1) diagnosed with ischemic stroke, which was confirmed with computed tomography or magnetic resonance image; (2) hospital admission within two days of stroke onset; and (3) lived in Xinzhuang during the campaign. Patients were excluded if they (1) had stroke onset within hospitals; (2) had dementia or psychological disorders; (3) had missing data on demographics, NIHSS score, date and time of symptom onset, and length and cost of hospital stay.
Healthcare utilization and costs
The relative difference in length of stay (LOS) and costs for AIS patients admitted to the hospital in pre- and post-SAVE was used to determine the potential impacts of SAVE intervention on healthcare utilization and cost [23]. The LOS and costs were estimated as the mean LOS and total hospitalization costs per admission due to AIS. We retrieved the sum of the costs for each AIS patient admitted to the hospital, including nursing, medical services, laboratory tests, and prescription drugs. Because the use of intravenous thrombolytic therapy was significantly improved following the SAVE intervention, the costs for intravenous thrombolytic therapy were not included in the analysis. Because increased thrombolysis use is an intended objective of the SAVE intervention, excluding these costs in the base case could underestimate short-term hospital expenditures while potentially overstating net savings. In contrast, thrombolysis-related costs are likely to be front-loaded and may be offset by long-term disability-related expenses, which could not be assessed within this short-term economic analysis. To address this limitation, thrombolysis costs were incorporated into the sensitivity analyses.
Cost of SAVE intervention
To measure the implementation cost of SAVE, we employed a micro-costing approach to quantify the resources needed to develop and implement the SAVE programs, following the recommendation for economic evaluations for implementation studies [24–26]. From the healthcare system’s perspective, direct cost data were collected for October 2016-December 2019. The healthcare system’s perspective was chosen over other perspectives because it reflects the direct medical costs paid by patients or healthcare providers. We quantified the quantity and unit cost from logistical and expenditure reports, further clarified by discussions with project staff and physicians. Total intervention costs included personnel costs (e.g., salary, benefits), travel, materials (e.g., posters), facilities, and campaign costs (e.g., planning, designing and executing media). The advertising fees, whether donated or subsidized by local communities, were based on fair market value. More details were described in eTable 1 in Supplemental Materials 1.
We used the average annual income in Shanghai for personnel costs, with an additional 40% fringe benefit rate. The annual income was obtained from the National Bureau of Statistics [27]. For in-person educational sessions, 50 volunteer physicians were recruited and provided with in-person educational sessions. The travel and training time for providing in-person educational sessions was based on the average income for full-time equivalent (FTE) healthcare providers. We also recruited ten project staff to distribute posters, brochures, and management (e.g., scheduling educational sessions). We used the average salary for community workers for project staff. Travel costs were determined from staff time and transportation modes. Cost data from different years were converted into 2019 Chinese Yuan (CNY) using a 5% cost-standardization rate based on China Guidelines for Pharmacoeconomic Evaluations [28], A range of 3–8% was also tested in the sensitivity analysis [28]. In this short-term retrospective economic analysis, the guideline-recommended rate was applied to standardize costs to 2019 CNY, ensuring comparability across different years.
Statistical analysis
Continuous variables are presented with mean and standard deviation (SD) and compared with Student’s t-test. Categorical variables are reported with frequencies and percentages and compared with chi-square tests. We also compared the length and hospital stay costs before and after the SAVE intervention. We employed a generalized linear model (GLM) with a gamma distribution and log link to calculate the mean incremental effect of SAVE on LOS and total hospital costs after adjustment for potential confounders, including age, sex, stroke severity, cigarette smoking, alcohol drinking, prior history of ischemic stroke or TIA, and comorbidities. The GLM employed a log-link function, where the β coefficient represents the logarithmic effect of the intervention on LOS and hospitalization costs. Adjusted incremental values were obtained by applying the inverse transformation to the β coefficient and integrating it with the model-predicted estimates. The data distribution of LOS and costs was demonstrated in eFigure 1 in Supplementary Material 1. These confounding factors were selected based on clinical relevance and prior literature demonstrating their influence on hospitalization costs and LOS in stroke patients [29].
Given the magnitude of the imbalance and the long post-intervention window, we conducted sensitivity analyses using propensity score models and inverse probability of treatment weighting (IPTW). The unweighted GLM was retained as the primary analytical approach due to the imbalanced number of study sample sizes and lengths of time in the Pre- and Post-SAVE intervention, to better reflect patient characteristics and intervention delivery as observed in clinical practice. Furthermore, propensity score matching is well-justified with a large sample size but may degrade inferences for imbalanced data or small sample sizes [30]. In addition, we also calculated the potential cost-offsets associated with the SAVE intervention. All analyses were conducted using SAS 9.4. Two-sided P < 0.05 was considered statistically significant.
Sensitivity analysis
Sensitivity analyses were performed to identify changes to the results under a range of conditions. First, the alternative inflation rates (3–8%) were used according to the China Guidelines for Pharmacoeconomic Evaluations [28]. Second, the costs for intravenous thrombolysis were included in the sensitivity analysis. Third, to balance the two groups, we performed propensity score analysis for patients in pre- and post-SAVE intervention. Fourth, we also tested the changes to the results due to different model specifications.
Results
Patient characteristics
From January 1 2016 to December 31 2019, among the 4,432 AIS patients who were admitted to the Minhang hospital, 137 patients were excluded due to missing data on the stroke onset time or hospital arrival time, 43 patients were excluded because their stroke occurred during hospitalization, 27 patients were excluded due to a history of dementia or psychological disorders, and 1,395 patients were excluded due to missing data on costs or NIHSS score. Therefore, a total of 2,830 patients with stroke were included in the analysis. The flowchart of the sample selection process is shown in Fig. 1.
Fig. 1.
Flowchart of sample selection
A total of 2,830 AIS patients were included in the analysis. 490 stroke patients were included in the pre-SAVE period, and 2,340 stroke patients were included in the post-SAVE period. As shown in Table 1, there were no statistically significant differences in age (P = 0.577), sex (P = 0.093), alcohol drinking (P = 0.056), prior history of ischemic stroke or TIA (P = 0.076), the onset of stroke during the daytime (P = 0.064), and comorbidities between AIS patients in the pre-and post-SAVE periods. Compared to the pre-SAVE period, the post-SAVE period had a higher proportion of AIS patients who had a mild stroke (75.64% vs. 65.10%; P < 0.001), suggesting that the patients were more alert to stroke signs.
Table 1.
Characteristics of stroke patients in the pre- and post-SAVE period
| Characteristics | Pre-SAVE | Post-SAVE | P-value |
|---|---|---|---|
| (n = 490) | (n = 2,340) | ||
| Age, years | 0.577 | ||
| 18–44 | 13 (2.65%) | 81 (3.46%) | |
| 45–64 | 155 (31.63%) | 685 (29.27%) | |
| 65–75 | 137 (27.96%) | 695 (29.70%) | |
| 75+ | 185 (37.76%) | 879 (37.56%) | |
| Sex | 0.093 | ||
| Female | 202 (41.22%) | 870 (37.18%) | |
| Male | 288 (58.78%) | 1470 (62.82%) | |
| Stroke severity | < 0.001 | ||
| 1 ≤ NIHSS < 5 | 319 (65.10%) | 1770 (75.64%) | |
| 5 ≤ NIHSS < 15 | 149 (30.41%) | 525 (22.44%) | |
| NIHSS ≥ 15 | 22 (4.49%) | 45 (1.92%) | |
| Cigarette smoker | 122 (24.90%) | 620 (26.50%) | < 0.001 |
| Alcohol drinker | 59 (12.04%) | 216 (9.23%) | 0.056 |
| Prior stroke or TIA | 107 (21.84%) | 430 (18.38%) | 0.076 |
| Medical conditions | |||
| Hypertension | 328 (66.94%) | 1473 (62.95%) | 0.095 |
| Diabetes mellitus | 134 (27.35%) | 658 (28.12%) | 0.729 |
| CH&SH | 11 (2.24%) | 58 (2.48%) | 0.760 |
| Atrial Fibrillation | 44 (8.98%) | 156 (6.67%) | 0.069 |
| Daytime onset | 356 (72.65%) | 1602 (68.46%) | 0.067 |
Abbreviation: SD, Standard Deviation; TIA, Transient Ischemic Attack; CH, Cerebral Hemorrhage; SH, Subarachnoid Hemorrhage; NIHSS, National Institutes of Health Stroke Scale; PVD, Peripheral Vascular DiseaseDaytime was defined as 6 AM to 6 PM
Cost of SAVE intervention
The SAVE intervention was estimated to cost CNY1,768,067 for Xinzhuang County from October 2016 to December 2019, with CNY 516,453 in 2017, CNY 519,316 in 2018, and CNY 523,526 in 2019 (Table 2). As shown in Fig. 2, most of the costs (59.51%) were for the mass media broadcast, including TV broadcast, radio broadcast, video production, new advertisement, and audio production. Other major cost drivers included the costs for personnel and travel (25.96%) and materials (6.69%), program activities (4.88%), and facilities (2.96%).
Table 2.
Implementation costs (CNY) of the SAVE intervention*
| Cost category | Implementation activity | 2016 (Oct-Dec) |
2017 (Jan-Dec) |
2018 (Jan-Dec) |
2019 (Jan-Dec) |
|---|---|---|---|---|---|
| Mass media | Newspaper advertisement | 1,968 | 7,938 | 7,980 | 8,000 |
| Television broadcast | 49,199 | 198,450 | 199,500 | 200,000 | |
| Video production | 9,840 | 9,923 | 9,975 | 10,000 | |
| Radio broadcast | 24,600 | 99,225 | 99,750 | 100,000 | |
| Audio production | 3,936 | 3,969 | 3,990 | 4,000 | |
| Personnel and travel | In-person education by physicians | 3,790 | 15,626 | 16,098 | 15,985 |
| Poster/brochure distribution by project staff | 27,929 | 107,572 | 108,202 | 108,529 | |
| Transportation | 1,655 | 5,833 | 6,224 | 6,240 | |
| Management by project coordinators | 2,793 | 10,757 | 10,820 | 10,853 | |
| Materials | Brochures | 7,380 | 29,768 | 29,925 | 31,500 |
| Posters | 1,476 | 5,954 | 5,985 | 6,300 | |
| Program activities | Training by experts | 5,325 | 5,072 | 4,830 | 4,600 |
| Physician trainee expense | 1,458 | 1,503 | 1,548 | 1,568 | |
| Project staff/coordinator trainee expense | 886 | 853 | 858 | 861 | |
| Planning for implementation | 53,251 | — | — | — | |
| Facilities and equipment rental | 892 | 910 | 879 | 949 | |
| Facilities | Facilities and equipment rental | 12,395 | 13,101 | 12,751 | 14,140 |
| Total: | 208,772 | 516,453 | 519,316 | 523,526 | |
*Cost data were converted into 2019 Chinese Yuan (CNY) using a 5% inflation rate
Fig. 2.
Implementation costs of the SAVE intervention
Healthcare utilization and costs
The mean (± SD) LOS after the SAVE intervention was significantly shorter at 8.80 (± 3.55) days in the post-SAVE period compared to 9.48 (± 3.80) days in the pre-SAVE period (Fig. 3; P < 0.001). Stroke patients at older age or with more severe stroke were more likely to have a longer hospital stay. In the GLM model, the SAVE intervention was associated with reduced LOS (Table 3; β coefficient [95% CI]: -0.070 [-0.11 to -0.030]; P < 0.001). The adjusted incremental reduction in LOS was 0.69 days per hospital admission.
Fig. 3.
Length and costs of hospital in pre- and post-SAVE intervention by age, sex, and stroke severity
Table 3.
Adjusted length and cost of hospitalization for ischemic stroke
| Characteristics | Length of Stay | Hospitalization Costs | ||||
|---|---|---|---|---|---|---|
| LSM | Adjusted β (95% CI) | P-value | LSM | Adjusted β (95% CI) | P-value | |
| SAVE campaign | ||||||
| Yes | 2.25 | -0.070 (-0.11 to -0.030) | < 0.01 | 9.98 | -0.11 (-0.17 to -0.053) | < 0.001 |
| No | 2.32 | 0 [Reference] | — | 10.10 | 0 [Reference] | — |
| Age | ||||||
| 18–44 | 2.21 | 0 [Reference] | — | 9.98 | 0 [Reference] | — |
| 45–64 | 2.26 | 0.048 (-0.044 to 0.14) | 0.303 | 10.05 | 0.074 (-0.066 to 0.21) | 0.302 |
| 65–75 | 2.31 | 0.097 (0.005 to 0.19) | 0.040 | 10.12 | 0.14 (0.001 to 0.28) | 0.049 |
| 75+ | 2.35 | 0.13 (0.042 to 0.23) | < 0.001 | 10.01 | 0.036 (-0.11 to 0.18) | 0.624 |
| Sex | ||||||
| Male | 2.30 | 0 [Reference] | — | 10.11 | 0 [Reference] | — |
| Female | 2.27 | -0.038 (-0.070 to -0.006) | 0.019 | 9.97 | -0.14 (-0.19 to -0.089) | < 0.001 |
| Stroke severity | ||||||
| 1 ≤ NIHSS < 5 | 2.12 | 0 [Reference] | — | 9.80 | 0 [Reference] | — |
| 5 ≤ NIHSS < 15 | 2.27 | 0.28 (0.21 to 0.36) | < 0.001 | 9.98 | 0.18 (0.13 to 0.23) | < 0.001 |
| NIHSS ≥ 15 | 2.40 | 0.23 (0.16 to 0.30) | < 0.001 | 10.45 | 0.65 (0.55 to 0.75) | < 0.001 |
| Cigarette smoker | 2.28 | -0.002 (-0.036 to 0.033) | 0.919 | 10.02 | 0.047 (-0.052 to 0.10) | 0.078 |
| Alcohol drinker | 2.28 | 0.004 (-0.048 to 0.056) | 0.877 | 10.01 | 0.061 (-0.018 to 0.14) | 0.132 |
| Prior stroke or TIA | 2.26 | -0.043 (-0.081 to -0.005) | 0.028 | 10.09 | -0.091 (-0.15 to -0.030) | 0.003 |
| Medical conditions | ||||||
| Hypertension | 2.28 | -0.012 (-0.044 to 0.020) | 0.446 | 10.07 | -0.070 (-0.12 to -0.021) | < 0.001 |
| Diabetes mellitus | 2.30 | 0.032 (-0.001 to 0.065) | 0.055 | 10.01 | 0.052 (0.009 to 0.10) | 0.046 |
| Atrial Fibrillation | 2.28 | -0.001 (-0.056 to 0.055) | 0.984 | 10.03 | 0.024 (-0.062 to 0.11) | 0.578 |
| Daytime onset | 2.28 | -0.013 (-0.044 to 0.018) | 0.416 | 10.02 | 0.030 (-0.018 to 0.078) | 0.213 |
Abbreviation: TIA, Transient Ischemic Attack; NIHSS, National Institutes of Health Stroke Scale; LSM: Least Squares Mean. Daytime was defined as 6 AM to 6 PM
The mean (SD) hospitalization costs per admission dropped from CNY 21,951 (± 10,411) to CNY 19,263 (± 12,773) CNY (P < 0.001). Older age and greater severity of stroke were associated with increased hospital admission costs. In the GLM model, the SAVE intervention was associated with lower hospitalization costs (β coefficient [95% CI]: -0.11 [-0.17 to -0.053]; P < 0.001). The adjusted incremental reduction in hospitalization costs was CNY 2,604 per hospital admission.
Potential cost-offsets associated with SAVE
The mean difference in the hospitalization costs between the pre- and post-SAVE intervention was CNY 2,688 (95% CI: 1,481 to 3,896) per admission. Over the period from 2016 to 2019, the estimated total cost-offset attributable to the SAVE intervention was CNY 4,522,321 (95% CI: 1,696,537 to 7,348,105), approximately three times the cost of implementing the intervention.
Sensitivity analysis
In the sensitivity analysis, when applying rates ranging from 3% to 8% to standardize costs to 2019 CNY, the costs associated with the 3-year SAVE program were CNY1,726,984 and CNY1,831,249, respectively. These cost estimates remained lower than the reduction in hospitalization costs observed during the same period (eTable 2 in the Supplemental Materials 1), indicating that the SAVE program continued to yield cost-offsets even when using different rates. Second, after incorporating the costs related to thrombolysis, the mean hospitalization costs during the post-SAVE period were significantly lower than those in the pre-SAVE period (eFigure 2 in the Supplemental Materials 1), further supporting the robustness of our findings. Furthermore, the impacts of the SAVE program on LOS and hospitalization costs remained consistent when utilizing different model specifications and propensity score techniques (eTable 3 in the Supplemental Materials 1). Specifically, the IPTW model demonstrated a statistically significant reduction in LOS with a β coefficient (95% CI) of -0.055 (-0.090 to -0.020; P < 0.001), and a significant decrease in hospitalization costs with a β coefficient (95% CI) of -0.18 (-0.23 to -0.12; P < 0.001).
Discussion
Our findings suggest that the multifaceted SAVE intervention is relative costly, with an annual expenditure exceeding CNY 500,000 in a community in Shanghai, largely driven by advertising expenses related to mass media. Despite its high cost, the SAVE intervention was effective in reducing prehospital delay and hospitalization costs. By shortening the prehospital delay, stroke patients were less likely to experience neurological deficits, which in turn reduced their needs for healthcare resources during and after hospitalization. Consequently, our findings indicate the potential cost offsets associated with the SAVE intervention. It is important to note, however, that this analysis represents a partial economic evaluation, focusing specifically on the downstream hospital costs linked to the intervention. Therefore, the findings should be interpreted as cost-offsets rather than net-savings from a full economic analysis.
Stroke educational campaigns effectively improve stroke knowledge and promote behavioral changes, but they generally require large, sustained funding [11, 18]. Policymakers face challenges in balancing the rising healthcare burden with implementing educational campaigns. However, to our knowledge, few studies have examined the economic impacts of educational campaigns [20, 31]. Hence, our analysis provided crucial empirical evidence for future work that scales up the intervention in other regions of China. This study provides important evidence on the cost of implementing the community-based campaign in China. On average, the annual cost of implementing the SAVE intervention was around CNY 500,000 in Shanghai, China. The advertising fees for mass media (e.g., newspaper, radio, and TV) were the major cost drivers, but these costs were donated by the local community. In addition, more than 25% of the total intervention cost was for personnel and travel, including salary and benefits. Even though physicians and project staff worked as volunteers, their personnel costs were also included in the analysis to obtain a more conservative estimate. It is important to recognize that Shanghai, as a megacity, incurs costs related to personnel, mass media, and venue rentals that may differ significantly from those in less economically developed regions. Given the relatively high costs of mass media intervention, innovative, more cost-effective and scalable approaches, particularly mobile-based platforms and AI-enhanced strategies, should be explored to enhance stroke awareness. For instance, the costs for digital social media campaigns in Nepal can be as low as 0.24 EUR to reach 1,000 users [32].
Our analysis also demonstrated the economic impacts of implementing the SAVE intervention. On average, the SAVE intervention was associated substantial reduction in hospitalization costs. In the post-SAVE period, a total of 2,340 stroke patients were admitted to hospitals. With an estimated reduction of CNY 2,604 in hospitalization costs per admission, the overall decrease in hospitalization expenses is projected to reach CNY 6,093,360 in the post-SAVE period. At the meanwhile, the total implementation costs of the SAVE intervention are only CNY 1,768,067 during the same period. Hence, the estimated cost-offsets for Xinzhuang, Shanghai, are expected to exceed 4 million CNY. With over 2.4 million incident stroke cases annually across China, one of the savings in hospitalization would be more substantial if successfully implementing the educational campaign nationwide. However, it is important to note that this projection is illustrative rather than predictive, as extrapolating per-patient cost offsets from a single hospital in Shanghai to the national level of stroke incidence may overstate the broader policy implications. Furthermore, considering the improved functional independence of stroke survivors, the educational campaign would have even more profound economic implications by reducing caregiver and society burdens.
Although Stroke recognition tools (e.g., F.A.S.T.) have been shown to improve stroke awareness, early detection, and reduce prehospital delay for stroke [20, 33–36], the effects of the education campaign were not sustainable, particularly after the COVID-19 pandemic. In fact, the United Kingdom relaunched the FAST campaign to reduce stroke burden in 2022 because of the prolonged prehospital time [37]. A potential explanation is due to the lack of effective and sustainable intervention strategies that are customized for different populations and settings [38]. Mass media campaigns are expensive, and their effects on behavior change are still unknown [11]. Traditional educational campaigns also require enormous resources for implementation, which is challenging for LMICs like China. Even though our findings suggested that the healthcare costs may offset the implementation costs over three years, it is still unknown whether the economic benefits are persistent over the long run. With the recent advancement in digital technology, therefore, innovative digital health education strategies should be developed to deliver digital media and digital products for stroke education. The current SAVE intervention is a community-based campaign, without targeting those at risk for stroke. Innovative strategies should also be developed to deliver interventions specific to those with higher risks of stroke.
This study has several strengths. First, this is the first empirical study in China to determine the implementation and downstream costs of educational campaigns in China, providing important evidence for future scaling up. The cost-effectiveness of mass media campaigns is a concern, and our study demonstrated that a well-designed educational campaign could reduce hospitalization costs and achieve potential cost-offsets. Second, our study also categorized implementation costs into conventional categories, which will aid in the planning of future implementation strategies. For example, mass media and personnel emerged as the primary cost drivers. These findings suggest that future research should explore the potential of digital health strategies to optimize resource allocation and reduce costs.
This study has some limitations. First, the study only collected data from one hospital; the findings may not be generalizable to other settings. Moreover, fair market valuation of donated media time may not accurately reflect real-world scalability if comparable donations are unavailable in other settings. Second, the shorten of LOS may contribute to other factors, e.g., the healthcare reforms that limit the length of hospitalization. Therefore, future studies could employ a stepped-wedge cluster trial design with a control group. Third, the personnel costs were estimated based on the average salary rather than the actual salary earned by physicians and project staff. Fourth, we cannot exclude the possibility of information bias, including changes in coding practices over time or the occurrence of regression to the mean. Fifth, patient-centered outcomes, e.g., functional status and quality of life, were not evaluated. It is important to note that a reduction in LOS does not inherently indicate an improvement in clinical outcomes. Sixth, the potential for residual confounding and the influence of secular trends, such as the optimization of stroke care pathways, national policy reforms, and hospital efficiency initiatives, could not be entirely ruled out. This is particularly relevant when extrapolating the study findings beyond the Xinzhuang region or when projecting potential cost savings at the national level. Given the limited statistical power in certain strata (e.g., patients with severe stroke) and the observational nature of the study design, the interpretation of the results should be made with caution to avoid overemphasizing subgroup differences.
Conclusions
Our findings suggest that the SAVE intervention was associated with potential cost-offsets through its effect on reducing LOS and hospitalization costs. To further enhance stroke awareness at the national level, innovative approaches to educational delivery, such as mobile-based platforms and AI-enhanced strategies, represent promising hypotheses that warrant future evaluation to confirm their cost-effectiveness and scalability.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
We acknowledge the community volunteers and the doctors of community hospital for promoting the Stroke 1-2-0 educational program in Minhang District.
Author contributions
JY: conceptualization, methodology, formal analysis, writing original draft, and review & editing manuscript; YW: conceptualization, methodology, formal analysis, writing original draft, and review & editing manuscript; YL: review & editing manuscript; ZKL: review & editing manuscript; RL: review & editing manuscript; JZ: conceptualization, methodology, supervision, funding acquisition, project administration, and review & editing manuscript.
Funding
Not applicable.
Data availability
The datasets used and/or analyzed during the present study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
This study was reviewed and approved by the ethics committees of the Institutional Review Boards of Minhang Hospital of Fudan University. The requirement for written informed consent from participants was waived by the Shanghai Ethics Committee for Clinical Research (Reference number: 2020-039-01 K).
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Jing Yuan, Yong Wang and Kevin Z. Lu contributed equally to this work.
Contributor Information
Jing Yuan, Email: jingyuan@um.edu.mo.
Jing Zhao, Email: zhao_jing@fudan.edu.cn.
References
- 1.Wu S, Wu B, Liu M, Chen Z, Wang W, Anderson CS, Sandercock P, Wang Y, Huang Y, Cui L, et al. Stroke in China: advances and challenges in epidemiology, prevention, and management. Lancet Neurol. 2019;18(4):394–405. [DOI] [PubMed] [Google Scholar]
- 2.Schroeder EB, Rosamond WD, Morris DL, Evenson KR, Hinn AR. Determinants of use of emergency medical services in a population with stroke symptoms: the second delay in accessing stroke healthcare (DASH II) study. Stroke. 2000;31(11):2591–6. [DOI] [PubMed] [Google Scholar]
- 3.De Keyser J, Gdovinová Z, Uyttenboogaart M, Vroomen PC, Luijckx GJ. Intravenous Alteplase for stroke: beyond the guidelines and in particular clinical situations. Stroke. 2007;38(9):2612–8. [DOI] [PubMed] [Google Scholar]
- 4.Goldstein LB. Acute ischemic stroke treatment in 2007. Circulation. 2007;116(13):1504–14. [DOI] [PubMed] [Google Scholar]
- 5.Centers for Disease Control and Prevention (CDC). Awareness of stroke warning symptoms–13 States and the district of Columbia, 2005. MMWR Morbidity Mortal Wkly Rep. 2008;57(18):481–5. [PubMed] [Google Scholar]
- 6.Moser DK, Kimble LP, Alberts MJ, Alonzo A, Croft JB, Dracup K, Evenson KR, Go AS, Hand MM, Kothari RU, et al. Reducing delay in seeking treatment by patients with acute coronary syndrome and stroke: a scientific statement from the American heart association Council on cardiovascular nursing and stroke Council. Circulation. 2006;114(2):168–82. [DOI] [PubMed] [Google Scholar]
- 7.Kleindorfer D, Kissela B, Schneider A, Woo D, Khoury J, Miller R, Alwell K, Gebel J, Szaflarski J, Pancioli A, et al. Eligibility for Recombinant tissue plasminogen activator in acute ischemic stroke: a population-based study. Stroke. 2004;35(2):e27–29. [DOI] [PubMed] [Google Scholar]
- 8.Vrudhula A, Zhao J, Liu R. Too young to have a stroke?-a global health crisis. Stroke Vasc Neurol. 2019;4(4):173–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Yuan J, Lu ZK, Xiong X, Li M, Liu Y, Wang LD, Liu R, Zhao J. Age and geographic disparities in acute ischaemic stroke prehospital delays in china: a cross-sectional study using National stroke registry data. Lancet Reg Health West Pac. 2023;33:100693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zachrison KS, Nielsen VM, de la Ossa NP, Madsen TE, Cash RE, Crowe RP, Odom EC, Jauch EC, Adeoye OM, Richards CT. Prehospital stroke care part 1: emergency medical services and the stroke systems of care. Stroke. 2023;54(4):1138–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Bouckaert M, Lemmens R, Thijs V. Reducing prehospital delay in acute stroke. Nat Rev Neurol. 2009;5(9):477–83. [DOI] [PubMed] [Google Scholar]
- 12.Fladt J, Meier N, Thilemann S, Polymeris A, Traenka C, Seiffge DJ, Sutter R, Peters N, Gensicke H, Flückiger B, et al. Reasons for prehospital delay in acute ischemic stroke. J Am Heart Assoc. 2019;8(20):e013101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wiyarta E, Hidayat R, Kurniawan M, Sinaga GHP, Budiman RA. Factors associated with prehospital and in-hospital delays in acute ischaemic stroke care in indonesia: a systematic review. BMJ Open. 2025;15(3):e095845. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhao J, Liu R. Stroke 1-2-0: a rapid response programme for stroke in China. Lancet Neurol. 2017;16(1):27–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Zhao J, Liu R. Calling for a rapid recognition and response program for stroke in China. Translational Perioperative Pain Med. 2016;1(4):1–4. [PMC free article] [PubMed] [Google Scholar]
- 16.Zhao J, Liu R. Stroke 1-2-0: The strategy and video release. Transl Perioper Pain Med. 2017;2(2):1–2. [PMC free article] [PubMed]
- 17.Zhao J, Li X, Liu X, Xu Y, Xu J, Xu A, Wang Y, Liu R. Changing the strategy and culture of stroke awareness education in China: implementing stroke 1-2-0. Stroke Vasc Neurol. 2020;5(4):374–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Tan J, Ramazanu S, Liaw SY, Chua WL. Effectiveness of public education campaigns for stroke symptom recognition and response in Non-Elderly adults: A systematic review and Meta-Analysis. J Stroke Cerebrovasc Dis. 2022;31(2):106207. [DOI] [PubMed] [Google Scholar]
- 19.Yuan J, Li M, Liu Y, Xiong X, Zhu Z, Liu F, Wang Y, Hu W, Lu ZK, Liu R, et al. Analysis of time to the hospital and ambulance use following a stroke community education intervention in China. JAMA Netw Open. 2022;5(5):e2212674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Flynn D, Ford GA, Rodgers H, Price C, Steen N, Thomson RG. A time series evaluation of the FAST National stroke awareness campaign in England. PLoS ONE. 2014;9(8):e104289. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453–7. [DOI] [PubMed] [Google Scholar]
- 22.Lyden P, Raman R, Liu L, Emr M, Warren M, Marler J. National institutes of health stroke scale certification is reliable across multiple venues. Stroke. 2009;40(7):2507–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Shoji A, Kudo K, Murashita K, Nakaji S, Igarashi A. Reduction in medical costs for cardiovascular diseases through innovative health awareness projects in a rural area in Japan. PLoS ONE. 2022;17(11):e0277600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Frick KD. Microcosting quantity data collection methods. Med Care. 2009;47(7 Suppl 1):S76–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.MacLeod KE, Chapel JM, McCurdy M, Minaya-Junca J, Wirth D, Onwuanyi A, Lane RI. The implementation cost of a safety-net hospital program addressing social needs in Atlanta. Health Serv Res. 2021;56(3):474–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ruger JP, Emmons KM, Kearney MH, Weinstein MC. Measuring the costs of outreach motivational interviewing for smoking cessation and relapse prevention among low-income pregnant women. BMC Pregnancy Childbirth. 2009;9:46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.National Burau of Statistics. [ https://data.stats.gov.cn/easyquery.htm?cn=C01 ].
- 28.Cao S, Liang W, Liang C, Lin H, Gao C, Yang L, Liu Y, Suo Y, Liu K, Chen Y, et al. Impact of China guidelines for Pharmacoeconomic evaluations on study quality: a systematic review of economic evaluations in China. Health Econ Rev. 2025;15(1):53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cao X, Li R, Tang W, Wang W, Ji J, Yin C, Niu L, Gao Y, Ma Q. How health risk factors affect inpatient costs among adults with stroke in china: the mediating role of length of stay. BMC Geriatr. 2024;24(1):131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.King G, Nielsen R. Why propensity scores should not be used for matching. Political Anal. 2019;27(4):435–54. [Google Scholar]
- 31.Stevens ER, Roberts E, Kuczynski HC, Boden-Albala B. Stroke warning information and faster treatment (SWIFT): Cost-Effectiveness of a stroke preparedness intervention. Value Health. 2019;22(11):1240–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Tunkl C, Paudel R, Thapa L, Tunkl P, Jalan P, Chandra A, Belson S, Prasad Gajurel B, Haji-Begli N, Bajaj S, et al. Are digital social media campaigns the key to Raise stroke awareness in low-and middle-income countries? A study of feasibility and cost-effectiveness in Nepal. PLoS ONE. 2023;18(9):e0291392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Wall HK, Beagan BM, O’Neill J, Foell KM, Boddie-Willis CL. Addressing stroke signs and symptoms through public education: the stroke heroes act FAST campaign. Prev Chronic Dis. 2008;5(2):A49. [PMC free article] [PubMed] [Google Scholar]
- 34.Morimoto A, Miyamatsu N, Okamura T, Nakayama H, Toyoda K, Suzuki K, Toyota A, Hata T, Yamaguchi T. Effects of intensive and moderate public education on knowledge of early stroke symptoms among a Japanese population: the acquisition of stroke knowledge study. Stroke. 2013;44(10):2829–34. [DOI] [PubMed] [Google Scholar]
- 35.Addo J, Ayis S, Leon J, Rudd AG, McKevitt C, Wolfe CD. Delay in presentation after an acute stroke in a multiethnic population in South London: the South London stroke register. J Am Heart Assoc. 2012;1(3):e001685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Oostema JA, Chassee T, Baer W, Edberg A, Reeves MJ. Brief educational intervention improves emergency medical services stroke recognition. Stroke. 2019;50(5):1193–200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Relaunch of the Act. FAST campaign to improve stroke outcomes. [ https://www.gov.uk/government/news/relaunch-of-the-act-fast-campaign-to-improve-stroke-outcomes ].
- 38.Kilbourne AM, Neumann MS, Pincus HA, Bauer MS, Stall R. Implementing evidence-based interventions in health care: application of the replicating effective programs framework. Implement Sci. 2007;2:42. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and/or analyzed during the present study are available from the corresponding author upon reasonable request.