Number needed to call in emergency care research: Postenrollment follow‐up data from a multicenter prospective syncope study

Emergency care research plays a crucial role in enhancing patient management and outcomes for individuals with acute medical conditions or injuries. ¹ Effective postenrollment follow‐up is essential for assessing intervention effectiveness and determining optimal management approaches. ² Insufficient follow‐ups can result in missing data and biases, while excessive attempts may lead to unnecessary resource allocation. ² Despite its significance, minimal research has focused on optimizing follow‐up procedures, leading to gaps in understanding the best approaches for obtaining patient follow‐up data.

Telephone follow‐up has emerged as a valuable tool in clinical research, offering high participation rates and precise data collection over multiple time points. ³ However, our understanding of telephone follow‐up characteristics and processes in clinical research remains limited. A previous study found that the response rate to telephone follow‐ups was significantly higher than those with mail and e‐mail follow‐ups. ⁴ Although the potential advantages of telephone follow‐up have been explored, no studies have examined the effectiveness of telephone follow‐up in emergency research, either in the United States or elsewhere. This study aims to address this gap by identifying the optimal number of phone surveys required to maximize effectiveness in capturing patient‐reported outcomes during telephone follow‐up in emergency care research.

We conducted a secondary analysis using an NIH‐funded, multicenter prospective cohort study, Practical Approaches to Care in Emergency Syncope (PACES, NCT04533425). The study cohort included all patients enrolled in the PACES study between August 2021 and August 2023. The PACES study included enrollment at six sites in the United States. The objective of the study was to externally validate two syncope risk stratification tools, the Canadian Syncope Risk Score and the FAINT score. ⁵ , ⁶ Eligible patients were adults 40 years or older who presented to the emergency department (ED) with syncope or presyncope without a serious ED diagnosis. The study was prepared and reported following the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. Ethical approvals were obtained from the institutional review boards at enrolling sites. Patient confidentiality was preserved by using anonymous health information. Enrolling EDs were affiliated with urban, academic medical centers located in the following states: New York, Tennessee, and California.

Data from the PACES database included patient demographics, participating sites, and telephone follow‐up survey information, including the total call attempts and the day and time the survey attempt was completed. We also assessed each participant's Area Deprivation Index (ADI), a composite measurement of neighborhood socioeconomic disadvantage. All telephone follow‐up surveys were performed centrally by research staff at the lead site at least 30 days after the index ED visits. Each participant was contacted twice daily for 7–14 days until contact was made or until 14 attempts were made. The telephone survey followed a prepared script (see Supplementary Files for details) and took 5 to 10 min to complete. At the completion of telephone follow‐up surveys, participants were given virtual gift cards of $25 via email or text message. The primary outcome was the number of telephone attempts required before participants completed the follow‐up surveys (between one and 14). Secondary outcomes included factors associated with participants who did not complete telephone follow‐up surveys.

All analyses were conducted using STATA MP, version 16 (StataCorp). Categorical variables were reported using frequencies and percentages, and continuous variables were reported using mean with standard deviation (SD) or median with interquartile range (IQR), depending on the data distribution. We reported cumulative incidence ratios with their corresponding 95% confidence intervals (CIs) for the primary outcome. We also displayed a Kaplan–Meier cumulative incidence curve for successful survey completion. The risk interval was described as the number of each telephone follow‐up call attempt. For the secondary outcome, we used univariable and multivariable logistic regression models to determine predictors associated with participants who did not complete telephone follow‐up calls. Adjusted odds ratios (aORs) and their corresponding 95% CIs were presented. All tests were two‐sided, and a p‐value of <0.05 was considered statistically significant.

A total of 758 participants were included. The mean ± SD age was 64.7 ± 12.8 years, 413 (54%) were female, and 331 (44%) were White. According to the ADI, most (43%) were ranked in the first national quintile (least disadvantaged). The median number of call attempts per participant was two (IQR 1–4), and 253 participants (33%, 95% CI 30%–37%) completed surveys on the first attempt. After 14 attempts, 661 (87%, 95% CI 85%–89%) completed the follow‐up calls. Contacts were made mostly on Monday, Saturday, and Sunday, and calls were attempted frequently from 12 p.m. to 2 p.m., 10 a.m. to 12 p.m., and 2 p.m. to 4 p.m. Table S1 demonstrates the characteristics and demographics of included participants. The cumulative incidence of completed telephone follow‐up calls is illustrated in Figure 1. A total of 491 patients (65%, 95% CI 61%–68%) completed surveys within three attempts, and 598 (79%, 95% CI 76%–82%) completed surveys within six attempts. Multivariable logistic regression found that age ≥ 80 years and participants from Vanderbilt University Medical Center were associated with unsuccessful follow‐up (aOR 2.21, 95% CI 1.01–4.88, p = 0.05; and aOR 3.87, 95% CI 2.05–7.30, p < 0.001, respectively). The details of factors associated with participants who did not complete the follow‐up calls after 14 attempts are described in Table S2.

FIGURE 1.

Kaplan–Meier cumulative incidence of completed survey responses. The numbers of patients at risk after zero, three, six, 12, and 14 survey attempts are shown below the cumulative curve.

In this study, we explored the characteristics of telephone follow‐up rates in emergency syncope research, utilizing data from the PACES study. Our findings provide insights into several key aspects of telephone follow‐up procedures and their effectiveness in capturing patient‐reported outcomes within this specific emergency care research setting.

One of the notable findings of our study was the high completion rate of telephone follow‐up surveys, with 87% of participants completing the follow‐up within the designated 14 attempts, with nearly one‐third of participants completing the survey on the first contact. This demonstrates the feasibility and acceptability of telephone follow‐up in this population. Also, our analysis revealed a point of diminishing returns, where subsequent call attempts yielded minimal increases in survey completion (65% completed within the first three attempts, 79% completed within the first six attempts, and slightly increased afterward). The completion rates observed in our study are consistent with findings from other clinical research domains, ³ , ⁴ indicating the robustness of telephone follow‐up as a data collection method across different medical specialties.

Loss to follow‐up is a critical concern in clinical research, as it can introduce bias and threaten study validity. ⁷ We proposed a threshold of six attempts may yield around 80% successful follow‐up surveys, providing a practical guideline for researchers. This information is crucial for optimizing call volume efficiency and resource allocation. However, to maximize efficiency, we recommend focusing efforts on the initial attempts and carefully considering the trade‐off between additional contact and resource allocation.

Correlated with previous studies, ⁸ , ⁹ participants aged 80 years and older were more likely to have unsuccessful follow‐up attempts, suggesting potential challenges in reaching and engaging older patient populations via telephone. Additionally, participants recruited from a specific external site were more likely to have unsuccessful follow‐up attempts than other participating sites. ¹⁰ We proposed that it would have been some actions that were not mentioned or documented in the follow‐up call protocol. For example, participants from University of California Davis may have been informed that the research coordinator from the central site will make a call for follow‐up in 7–14 days while those at Vanderbilt may have not. However, the exact explanation is intriguing and warrant further investigation. While the primary analysis suggests that the overall findings are robust, the specific influence of Vanderbilt on the results requires cautious interpretation. Further research is needed to fully understand the factors contributing to the observed differences in follow‐up rates across sites.

Our study benefits from a multicenter design, which enhances the generalizability of the findings. The standardized data collection protocols and central follow‐up procedures could minimize variability and improve data quality. However, our findings highlighted unexpected results: participants recruited from Vanderbilt University Medical Center had lower odds of successful telephone follow‐up. These finding underscores the importance of institutional differences and local practices in multicenter studies, despite the fact that standardized protocol was prepared and implementation. On the other hand, several limitations should be considered. First, our study focused specifically on patients presenting with syncope or presyncope to EDs, limiting the direct application of findings to other emergency care populations. In addition, while we identified factors associated with unsuccessful follow‐up attempts, the reasons for noncompletion (i.e., phone number accuracy, habitation status, severity of participants) were not systematically documented and warrant further investigations. Selection bias may have occurred if patients with certain characteristics were more likely to participate in the study or complete the follow‐up survey. For example, patients with milder symptoms or better overall health might be more likely to participate and follow up. This could potentially overestimate the completion rates. Information bias could arise from inaccurate or incomplete patient‐reported data, affecting the reliability of the collected information. Furthermore, using a single calling center might not reflect practices implemented at different research sites, and the incentive (i.e., a $25 gift card) offered for participation might not be generalizable to all studies. ¹⁰ Future research should explore the effectiveness of telephone follow‐up in other emergency care settings, comparing it with alternative follow‐up methods. Additionally, investigating cost‐effectiveness and participant burden associated with telephone follow‐up compared to other methods would be valuable.

In conclusion, our study provides valuable insights into emergency care research characteristics and telephone follow‐up data. Despite challenges with reaching certain patient demographics and institutional differences, telephone follow‐up is a feasible and effective method for collecting postenrollment data in this clinical setting. A guideline of six survey attempts may serve as a practical benchmark for future researchers. Future research should explore strategies to optimize telephone follow‐up protocols and enhance participant engagement to improve the validity and reliability of patient‐reported outcomes in emergency care research.

AUTHOR CONTRIBUTIONS

Wachira Wongtanasarasin, Daniel K. Nishijima, Nancy Wood, John DeAngelis, Alan Storrow, Jonathan Schimmel, Nataly Beltre, Dana Sacco, and Marc A. Probst conceptualized the study design and concept. Wachira Wongtanasarasin, Daniel K. Nishijima, Nataly Beltre, and Marc A. Probst contributed to the data acquisition. Wachira Wongtanasarasin, Daniel K. Nishijima, and Marc A. Probst contributed to the analysis and interpretation of the data. Wachira Wongtanasarasin drafted the manuscript. Daniel K. Nishijima and Marc A. Probst supervised, critically revised, and edited the manuscript. All authors contributed to its revision and approved the final manuscript.

FUNDING INFORMATION

The PACES study was supported by the National Heart, Lung, and Blood Institute (R01HL149680).

CONFLICT OF INTEREST STATEMENT

MP is currently supported by an R01 grant from the NIH/NHLBI (R01HL149680) and received a one‐time research donation from Roche Diagnostics in 2023. The other authors declare no conflicts of interest.

Supporting information

Data S1:

Wongtanasarasin W, Nishijima DK, Wood N, et al. Number needed to call in emergency care research: Postenrollment follow‐up data from a multicenter prospective syncope study. Acad Emerg Med. 2025;32:165‐168. doi: 10.1111/acem.15010