Enhancing COVID-19 booster vaccination among the elderly through text message reminders

ABSTRACT

The BOOST (Booster promotion for older outpatients using SMS text reminders) program at Taipei Veterans General Hospital assessed the effectiveness of text message reminders in enhancing COVID-19 booster vaccination rates among the elderly, guided by the Health Belief Model (HBM). Targeting patients aged 65 and above, eligible yet unvaccinated for a COVID-19 booster, this cohort study sent personalized reminders a week prior to their scheduled appointments between April 18, 2022, and May 12, 2022, acting as cues to action to enhance vaccination uptake by overcoming perceived barriers and raising awareness of benefits. Over 5 weeks, the study observed a 38% increase in vaccination rate among 3,500 eligible patients, markedly surpassing the concurrent national rate increase of 4% for the same demographic. The majority of vaccinations occurred within two weeks after the reminder, illustrating the effectiveness of the strategy. Cox regression analysis identified age and time since last vaccination as significant predictors of responsiveness, with those aged 65–74 and 75–84 showing higher uptake, particularly when reminders were sent within 4 months after the last dose. A single reminder proved to be effective. The findings of this study demonstrate the potential of SMS reminders to promote COVID-19 vaccination among the elderly through the strategic use of HBM principles, suggesting a feasible and effective approach to public health communication.

GRAPHICAL ABSTRACT

Introduction

The use of text messaging as a tool to promote COVID-19 vaccination is a crucial component of public health efforts. Short Message Service (SMS) reminders have been considered a cost-effective method for improving vaccine uptake^1–3 with response rates varying across different settings. Previous studies have demonstrated that SMS reminders can boost vaccination rates by 2% to 11% in various vaccine campaigns.^2,4–10 In addition, text reminders sent by trusted medical institutions with doctor endorsement prior to a primary care appointment also lead to higher response rates.^7,9,11 However, using SMS to promote vaccination in the elderly population is a subject of debate. Limited technological access,^3,8 skepticism of information received via text messages, and the complexity of health-related SMS content may impede the reach of SMS vaccination promotions.^12–15 Therefore, real-world evidence is required to explore the effectiveness of SMS in promoting vaccination among this population.

The aged population is particularly vulnerable to severe illness and mortality from SARS-CoV-2,^16,17 making vaccination critical to mitigate these risks.^18,19 However, the coverage of booster vaccines among senior citizens in Taiwan during the first half of 2022 was suboptimal. In response to this challenge, Taipei Veterans General Hospital (TVGH) launched the BOOST program (Booster promotion for older outpatients using SMS text reminders). By utilizing the National Immunization Information System (NIIS), the program identified eligible patients, updated vaccination records, and sent customized text reminders. The primary objective was to increase the uptake of COVID-19 booster shots in individuals over 65, guided by insights from the Health Belief Model (HBM) – a theoretical framework that emphasizes the importance of perceptions in health-related decision-making.²⁰ This study evaluated the impact of the BOOST program and the factors influencing vaccine uptake, and provided the basis for discussing how HBM components such as perceived barriers, benefits, and cues to action inform our approach.

Materials and methods

Data source, and ethical concerns

This cohort study was approved by TVGH’s institutional review board (IRB No. 202212005AC) and exempted the need for informed consent from participating patients, as all analyzed data were de-identified by TVGH’s Big Data Center. Furthermore, no personally identifiable information or human biospecimens were utilized. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for observational studies.

Theoretical framework

The design of the intervention was informed by the Health Belief Model (HBM), which posits that health-related actions are influenced by an individual’s perception of the severity and susceptibility to a health issue, the benefits of taking action, the barriers to such action, and the presence of a cue to action.²⁰ In the context of our study, the SMS reminders acted as cues to action, aimed to mitigate a variety of perceived barriers – not only by simplifying the booking process but also by overcoming forgetfulness, motivational hurdles, informational gaps and source credibility – while simultaneously reinforcing the critical benefits of enhanced protection against COVID-19 through personalized, direct communication and the convenience of scheduling vaccination with existing appointments. This framework guided both the design of the reminder messages and the analysis of factors influencing vaccine uptake.

Study design and participants

The BOOST program aimed to assess the effectiveness of text reminders as an intervention to promote COVID-19 booster vaccinations among the elderly population. All data were obtained from TVGH’s electronic medical record system and appointment system. The study included individuals aged 65 and above with scheduled outpatient appointments at TVGH between April 18, 2022 and May 12, 2022. To be eligible, participants must have completed the primary COVID-19 vaccine series at least three months before the study period, and the primary series received must be approved by Taiwan’s Food and Drug Administration. (TFDA), such as ChAdOx1 48 (AZD1222), mRNA1273 (Spikevax), BNT162b2, MVC-COV1901, or vaccines approved by the WHO Emergency Use Listing (EUL). We excluded patients who met any of the following conditions:

1. Received a primary series consisting of a single dose rather than the regular 2-dose series.

2. Received a primary series not approved for Emergency Use Authorization (EUA) in Taiwan.

3. Had already received a booster before we sent the text reminder, but the vaccination record had not been uploaded to NIIS when we checked eligibility.

4. Had record errors.

The data flow diagram is shown in Figure 1.

Figure 1.

Study flow.

Abbreviations: OPD, outpatient department; EUA, emergency use authorization; SMS, short message service.

The first box of the flowchart described the patients included in the study. The second box summarized the exclusion criteria of the study. The participants who met the requirements were divided into responders and non-responders.

Exposure

A reminder message stating “Dear (patient name), your upcoming outpatient appointment at TVGH in the (specific department) is on (date). You are eligible for a COVID-19 booster vaccine. Please make an appointment soon to enhance your protection.” was sent to eligible patients one week before their scheduled visits. The TVGH online booking link was also provided, enabling them to opt for receiving the booster on the same day as their planned appointments. Moreover, patients with multiple appointments during our research period would receive a repeat text message before each appointment if they were still unvaccinated.

Outcomes

The hospital’s information system would automatically monitor patients’ vaccination records on the NIIS for 5 weeks: one week before the scheduled visit and four weeks afterward. The primary outcomes were the vaccination rate and response rate to SMS reminders within 5 weeks. After 35 days from the date of the text reminder, patients’ vaccination records were no longer tracked, and any subsequent vaccinations would not be attributed to the effectiveness of the SMS. Responders were defined as those who received the booster within 5 weeks of the text reminder. Conversely, individuals who received the booster on the same day as the text reminder or did not receive it within 5 weeks were classified as non-responders. Additionally, we documented the cumulative response rate on a weekly basis. We also compared our results with the national increase in vaccination rates for the corresponding age group over our study period. The national database, COVID-19 Dashboard,²¹ is updated weekly for vaccination statistics and is managed collaboratively by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University and the Taiwan Centers for Disease Control, Ministry of Health and Welfare.

Covariates

In our analysis of the impact of text reminders on COVID-19 vaccination, we examined various factors that could potentially influence the outcomes, categorizing them into three primary domains: demographics, vaccine-related factor, and SMS-related variables. Demographic considerations included sex (male or female) and age groups (65 to 74, 75 to 84, and over 85). Vaccine-related factor comprised the brand of the previous COVID-19 vaccine (ChAdOx1, mRNA1273, BNT162b2, and MVC-COV1901). SMS-related variables were the number of text messages received (1,2, or ≥ 3) and the interval between the last vaccination and the date of the SMS reminder (within 4 months, 4 to 6 months, 6 to 8 months, and over 8 months).

Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics, version 26, (I.B.M. Corp., Armonk, NY, USA). Univariate analysis was performed using the Chi-square test to evaluate similarities between SMS responders and non-responders. Vaccination rate is the percentage of participants who were vaccinated within 5 weeks after receiving text reminders. To assess the effectiveness of SMS, response rates were calculated based on the number of SMS recipients who completed booster vaccination within a 35-day period after receiving the first text message, divided by the total person-days observed during the same period, and expressed as per 100 person-days. Furthermore, to further evaluate responses after the first text reminder, response rates per 100 person-days were calculated using the life table method every week following the initial text reminder. A cumulative response rate since the first text reminder was also estimated using the Kaplan-Meier curve.

A Cox regression model was fitted and the adjusted hazard ratio (HR) was used to assess the factors associated with the responder, including sex, age, brand of the previous COVID-19 vaccine, number of text messages received and the interval between the last vaccination and the date of the SMS reminder. To quantify the effect of the level of each factor, the average marginal effect (AME) of the factors was calculated, which is the average predicted probability that would be observed if the entire study population were at the same level of each factor, while keeping everything else in the data unchanged. The AMEs were expressed as percentages and interpreted as the average predicted probability of being a responder if individuals were at the average level for those factors. A two-tailed level of 0.05 was considered statistically significant.

Results

From April 18, 2022, to May 12, 2022, a total of 3,872 elderly patients with appointments at TVGH were eligible for the COVID-19 booster vaccination. They received one or more text message reminders a week before their appointments, encouraging them to schedule a booster shot. Overall, 3,500 eligible participants were enrolled in the study and 4817 SMS reminders were sent. Among them, 1984 (57%) were female, 1918 (54.8%) were aged 65–74, 1003 (28.7%) were aged 75–84 and 579 (16.5%) were over 85. A total of 1,318 outpatients completed a booster vaccination within 35 days after receiving the text reminder, yielding a remarkable vaccination rate of 38%. This vaccination uptake is especially significant compared to the 4% increase in the national vaccination rate for the same age group during the same period.²¹ Furthermore, the younger group responded more positively (57.1% vs 53.4% in the 65–74 group, 29.4% vs 28.2% in the 75–84 group, and 13.4% vs 18.4% in the over 85 group), as shown in Table 1.

Table 1.

Factors associated with responders and non-responders.

	Respondersa	Non-responders
Factors	Count(%)	Count(%)	P-value
Overall	1,318(100.0)	2,182(100.0)
Sex			.33
Female	761(57.7)	1,223(56.0)
Male	557(42.3)	959(44.0)
Age group			<.001
65–74	753(57.1)	1,165(53.4)
75–84	388(29.4)	615(28.2)
≥85	177(13.4)	402(18.4)
Brand of previous COVID-19 vaccine			.58
ChAdOx1	330(25.0)	591(27.1)
mRNA1273	835(63.4)	1,352(62.0)
BNT162b2	137(10.4)	216(9.9)
MVC-COV1901	16(1.2)	23(1.1)
Interval between last vaccine and the date of SMS reminder			<.001
≤4 months	110(8.3)	121(5.5)
4–6 months	431(32.7)	628(28.8)
6–8 months	756(57.4)	1,382(63.3)
≥8 months	21(1.6)	51(2.3)
Cumulative count of SMS reminders			<.001
1 message	1,061(80.5)	1,604(73.5)
2 messages	185(14.0)	409(18.7)
≥3 messages	72(5.5)	169(7.7)

Abbreviations: SMS, short message service.

^aUnivariate analysis was performed using the Chi-square test to evaluate similarities between SMS responders and non-responders.

We also observed that participants were more likely to receive a booster vaccination if the text reminder was sent closer to the date of their last vaccination. (8.8% vs 5.5% within 4 months, 32.7% vs 28.8% between 4–6 months, 57.4% vs 63.3% between 6–8 months, 1.6% vs 2.3% over 8 months). Additionally, a single message appeared to be effective, as 2 or more messages did not significantly increase the response rate (80.5% vs. 73.5% with one message, 14% vs. 18.7% with 2 messages, 5.5% vs. 7.7% with more than 3 messages).

Response rates by days

The cumulative response rate to the booster vaccine over time following the initial text reminder is illustrated in Figure 2(a). Notably, responders showed a prominent increase within the first two weeks after receiving the text reminder, and the growth slowed as time passed. Most responders (416 out of 1318) received the booster shot within 7 days of the SMS reminder. The response rate per 100 person-days was highest from day 8 to day 14 (1.90%; 95% CI, 1.72–2.09), followed by day 1 to day 7 (1.55%; 95% CI, 1.41–1.71), and then gradually declined from day 15 to day 35 after the first text reminder was sent, as presented in Figure 2(b).

Figure 2.

Response rate by days.

(a)

Cumulative response rates and remaining number of non-responders to the COVID-19 booster vaccine by days since the first text reminder for COVID-19 booster. A cumulative response rate was estimated using the Kaplan-Meier curve.

(b) Response rates for COVID-19 booster vaccine by days after the date of the first text reminder.

Response rates were calculated as the number of responses to text reminders during a 35-day period after the first text message, divided by the total person-days observed during the same period, and expressed as per 100 person-days.

Influential factors

The factors influencing the response rate are presented in Table 2. The Cox regression analysis revealed that sex was not a remarkable factor (HR 1.02; 95% CI, 0.91–1.13, p = .789), while age was a significant independent variable. Patients aged 65 to 74 and 75 to 84 were more likely to receive booster shots than those aged 85 and over (HR, 1.41; 95% CI, 1.15–1.74, and HR, 1.42; 95% CI, 1.16–1.72, respectively, both p < .001). If the text message reminder was sent within 4 months of the last vaccination, SMS recipients were more likely to receive a booster (HR, 2.18; 95% CI, 1.36–3.49, p < .001). A single reminder message was effective (HR, 1.57; 95% CI, 1.23–1.99, p < .001), while two or more reminders did not enhance the effect.

Table 2.

Influential factors and adjusted hazard ratios (HR) for responders.

Factors	Adjusteda hazard ratio (95% CI)	P-value	Significanceb
Sex
Female	1.02 (0.91–1.13)	.789	-
Male	-ref-	NA	NA
Age group
65–74	1.41 (1.15–1.74)	<.001	***
75–84	1.42 (1.16–1.72)	<.001	***
≥85	-ref-	NA	NA
Brand of previous COVID-19 vaccine
ChAdOx1	1.30 (1.03–1.65)	.030	*
mRNA1273	1.13 (0.93–1.38)	.209	-
BNT162b2	-ref-	NA	NA
MVC-COV1901	1.45 (0.86–2.47)	.166	-
Interval between previous vaccine and date of reminder
≤4 months	2.18 (1.36–3.49)	.001	***
4–6 months	1.65 (1.06–2.56)	.027	*
6–8 months	1.36 (0.88–2.11)	.164	-
≥8 months	-ref-	NA	NA
Cumulative count of text reminders
1 message	1.57 (1.23–1.99)	<.001	***
2 messages	1.07 (0.82–1.41)	.606	-
≥3 messages	-ref-	NA	NA

Abbreviation: CI, confidence interval; -ref-, reference; NA, not applicable.

^aA Cox regression model was fitted using factors associated with the responder, including sex, age, brand of the previous COVID-19 vaccine, number of text messages received and the interval between the last vaccination and the date of the SMS reminder.

^bSignificance.

-, insignificant.

*, p < .05.

**, p < .01.

***, p < .001.

Average marginal effects

To further quantify the effects of factors associated with the response to SMS reminders, average marginal effects (AME) were calculated. These effects can be interpreted as the maximum impact of each factor on the likelihood of responding, as presented in Figure 3. The AME was markedly higher in patients who received our SMS reminder within 4 months of their last vaccination (AME, 2.68; 95% CI, 0.11–5.25, p = .041). This indicates that a shorter interval between the last vaccination and our text message reminder is the most crucial factor for potential vaccine recipients. Moreover, a single reminder message could increase the probability of responding to the text reminder by 33% (95% CI, 0.23–2.44, p = .018), while multiple messages did not statistically provide additional benefit. Regarding age, older adults aged 65–74 and 75–84 years also exhibited a 4% and 5% increase, respectively, in the probability of receiving a booster dose (AME, 1.04; 95% CI, 0.10–1.99, p = .030) and (AME, 1.05; 95% CI, 0.14–1.96, p = .024).

Figure 3.

Average marginal effects for response to SMS reminders among elderly outpatients in TVGH.

Abbreviations: AME, average marginal effects; CI, confidence interval; -ref-, reference; SMS, short message service.

AMEs were calculated using a Cox regression model adjusted for the listed factors. The most common level of each factor was used as the base value (red dotted reference line at zero).

Discussion

Our findings indicate that SMS reminders serve as effective cues to action and significantly enhance COVID-19 booster vaccination rates among the elderly. This result directly aligns with the Health Belief Model (HBM) by addressing key components of perceived benefits and barriers. Specifically, they streamline the booking process, bridge information gaps, and increase message trustworthiness, while reinforcing the vital benefits of enhanced protection against COVID-19.

Notably, our research achieved a very impressive 38% vaccination rate with our text reminders, which exceeded the rates observed in other vaccination programs.^{2,5–8,10,22,23} In contrast, the national vaccination rate for the same age group increased by only around 4% points, from 66.4% to 70.1%, during the same period.²¹ This disparity underscores the potential of targeted text message reminders as an effective tool in boosting vaccination rates within specific populations.

Senior citizens often face an information gap, particularly with the complexity of the COVID-19 vaccine series, leaving the elderly uncertain about when to receive the next dose. Our personalized reminders informed patients of their eligibility and offered a link for easy scheduling. This allowed the convenience of scheduling vaccination alongside existing appointments, which altogether reduced barriers to vaccination. Previous research has shown that tailored messages targeting specific demographic groups and ensuring access to vaccines are more effective than generic messages.^6,24–27 In addition, trust in the information is enhanced when it is provided by reputable healthcare institutions and endorsed by physicians.^7,11 Our reminders, sent from the hospital before scheduled visits, not only enhanced message credibility but also enabled patients to consult their doctors about COVID vaccination during their upcoming appointments, potentially increasing the perceived benefits of vaccination.

Despite the promising results, the cost-effectiveness of upscaling this intervention warrants further investigation. During our study period, the cost of sending a text message in Taiwan was approximately 1 New Taiwan Dollar (NTD), equivalent to 0.034 United States Dollars (USD).²⁸ The estimated total cost of the SMS intervention, based on our collected data, amounted to 4,817 NTD (approximately 163.8 USD). Given that no additional staff were needed to manage the SMS system, making it a potentially low-cost intervention, detailed cost-benefit analyses are essential for broader application. Besides, we sent the text reminder using contact information provided by patients who consented to receive health-related communications. Although all patient data were de-identified by our Big Data Center prior to analysis, safeguards such as strict access control, periodical consent renewal and regular privacy audits could be applied to minimize privacy risks in future studies. Measures to prevent unauthorized access and misuse of data are also imperative to maintain trust and compliance with data protection laws.

SMS reminders have proven to be an effective method for informing patients of their vaccine eligibility, complementing traditional communication methods such as phone calls, websites, television, and flyers.^1,29,30 This approach has notably boosted vaccination rates among the elderly in Taiwan, and the findings are consistent with previous studies that emphasize the significance of mobile phone coverage in enhancing SMS-based vaccination campaigns.^2,8,31–33 A key factor in our study’s success is Taiwan’s high rate of mobile phone ownership, especially among seniors. Standing out for its technological access and digital literacy, Taiwan ranked first in the world in mobile broadband subscribers and third in smartphone ownership in 2021,³⁴ with an impressive 87.8% of people aged 65 and older owning a smartphone.³⁵ This widespread technological engagement among the elderly significantly contributed to the effectiveness of our SMS-based intervention. However, despite the high overall response rate within the elderly population, we found that the effects are particularly evident in relatively younger individuals aged 65–84, implying that additional or alternative interventions may be required for those over 85 to achieve similar levels of vaccine uptake.

Furthermore, we noticed a wear-out effect in our study. This phenomenon suggests that the initial drive and high response rates may gradually diminish over time.^36,37 In our study, most vaccinations took place in the first two weeks following the receipt of the SMS reminder. Recognizing this pattern, it becomes crucial to implement proactive measures early on, such as ensuring vaccination services are readily accessible during this initial period. This observation highlights the need for future SMS-based vaccination promotion campaigns to focus their efforts within this critical timeframe of early engagement and to urgently develop alternative strategies. Such strategies are essential to maintain momentum for immunization advocacy beyond the initial surge, ensuring sustained engagement and uptake.

It’s also important to note that a shorter interval between vaccine doses is crucial for increasing vaccination rates. Therefore, timely SMS reminders should be sent as soon as patients become eligible for their next dose. However, our research observed that repeated SMS reminders with similar content have little additional effect. Thus, caution is advised when sending repeated reminders to avoid negative outcomes such as message fatigue, recipient irritation, increased opt-outs, and resource wastage. These issues can diminish the overall effectiveness of the messages over time. To maintain the effectiveness of SMS-based vaccination promotion, it is essential to carefully manage the frequency and content of repeated messages.

Our SMS intervention, which proved to be inexpensive and convenient, suggests that similar strategies could be beneficial in other disease control campaigns. For programs with less precise registries, approaches such as data cleansing and integration may be helpful. These efforts could aim to remove inaccuracies and outdated information by linking diverse healthcare databases. In our case, the BOOST program combined data from TVGH’s electronic medical record and the National Immunization Information System to identify eligible patients. It’s also essential to establish feedback mechanisms that allow participants to directly verify and update their information, as well as provide incentives for maintaining accurate records. In addition, conducting pilot studies before full implementation will provide valuable insights. Real-world data is instrumental in evaluating the effectiveness of the SMS strategy across various demographic groups and determining if the positive outcomes can be sustained in different public health contexts.

Limitations

While our study provides valuable insights, it is limited in several ways. One limitation of our study is potential confounding bias. The study design lacks a control group and randomization, making it challenging to solely attribute changes in vaccination rates to the text message reminders. Numerous factors, including rising disease prevalence, fear of severe illness, encouragement from friends or family, and exposure to vaccine-related information, contribute to motivating individuals to get vaccinated.³⁸ Vaccine supply, public opinions, incentives, and governmental strategies also influence vaccination willingness.^39,40 Although our national statistics do not meet the criteria for a traditional control group due to the absence of detailed information other than age, the comparison remains instructive. It indicates that our intervention achieved significantly higher vaccination rates among SMS recipients (38%) compared to their same-age peers nationwide (4%), despite these similar domestic influencing factors.” This suggests that the SMS strategy effectively enhanced vaccination uptake among the targeted group. Second, our study is a single-center study conducted at Taipei Veterans General Hospital, which introduces a representative bias. Therefore, the generalizability of our findings to other cities and medical institutions in Taiwan or other countries may be limited. Moreover, it’s important to note that our study findings are specific to senior outpatients and their response to COVID-19 booster shot promotions, potentially leading to selection bias. It remains uncertain whether the observed effects can be extrapolated to the entire population and extended to other types of vaccines. However, it’s worth mentioning that the effectiveness of SMS reminders has been reported in various studies,^2,10,33,41 which may mitigate this limitation to some extent. We did not conduct an in-depth cost-effectiveness analysis for this study. In future research, comprehensive cost-benefit analyses should be performed to more rigorously evaluate the economic impact and scalability of this intervention. Another limitation is the potential for censoring bias, as we stopped tracking participants’ vaccine records 35 days after our SMS reminders. While some data may have been missed or delayed due to institutions not updating vaccination records promptly, we believe this issue is minimal given that our national database is updated daily. Although our study did not directly measure the Health Belief Model’s components through questionnaires, the SMS reminders were designed with the HBM’s principles in mind, aiming to act as cues to action while addressing perceived barriers and enhancing perceived benefits related to COVID-19 booster vaccination. The significant increase in vaccination rates following these reminders suggests that they effectively influenced participants’ health-related behaviors in a manner consistent with the HBM. Addressing these limitations and incorporating direct measurements of HBM constructs in future research could further validate the practical effectiveness of text message reminders in vaccination promotion and elucidate this intervention’s impact on individuals’ health beliefs.

Conclusion

Our study highlights the potential of text message reminders as a powerful tool in health communication, effectively increasing the COVID-19 booster vaccination rate by up to 40% among the elderly, particularly in the relatively younger segment. These findings demonstrate that a single, well-timed text message reminder designed on the basis of the Health Belief Model (HBM) and sent within an appropriate timeframe, can increase vaccination rates. By acting as cues to action that address perceived barriers and enhance the perceived benefits of vaccination, our SMS approach aligns with key HBM constructs, and motivates older adults to engage in health-promoting behaviors. These insights provide valuable directions for improving vaccination strategies targeting senior populations worldwide, and underscore the importance of applying behavioral health theories such as HBM to influence positive health outcomes.

Funding Statement

This research received no external funding. This work was supported by the Taipei Veterans General Hospital under intramural Grants [V111E-002-1, V112E-001-1, V113E-002-1]. We would like to acknowledge the support provided by the Hsu Chin-De Memorial Foundation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Author contributions

YC Lee, WM Chen, and YC Chen conceived the design. YC Lee, BH Lee, YH Lin, and YC Chen collected, analyzed and interpreted the data. YC Lee drafted the manuscript. BJ Wu, TJ Chen and WM Chen provided administrative and technical support. The study was supervised by TJ Chen, WM Chen, and YC Chen, who also contributed to manuscript revision. All authors reviewed and approved the final manuscript.

Data availability statement

The nationwide data that support the findings of this study are openly available on the National Center for High-performance Computing (NCHC) COVID-19 platform, available at https://covid-19.nchc.org.tw/2023_terms.php. The data of participants used in this study are not available for public sharing due to hospital policy.