Improving Post-Injury Follow-up Survey Response: Incorporating Automated Modalities

Background Incorporating post-discharge data into trauma registries would allow for better research on patient outcomes, including disparities in outcomes. This pilot study tested a follow-up data collection process to be incorporated into existing trauma care systems, prioritizing low-cost automated response modalities. Methods This investigation was part of a larger study that consisted of two protocols with two distinct cohorts of participants who experienced traumatic injury. Participants in both protocols were asked to provide phone, email, text, and mail contact information to complete follow-up surveys assessing patient-reported outcomes six months after injury. To increase follow-up response rates between protocol 1 and protocol 2, the study team modified the contact procedures for the protocol 2 cohort. Frequency distributions were utilized to report the frequency of follow-up response modalities and overall response rates in both protocols. Results A total of 178 individuals responded to the 6-month follow-up survey: 88 in protocol 1 and 90 in protocol 2. After implementing new follow-up contact procedures in protocol 2 that relied more heavily on the use of automated modalities (e.g., email and text messages), the response rate increased by 17.9 percentage points. The primary response modality shifted from phone (72.7%) in protocol 1 to the combination of email (47.8%) and text (14.4%) in protocol 2. Conclusions Results from this investigation suggest that follow-up data can feasibly be collected from trauma patients. Use of automated follow-up methods holds promise to expand longitudinal data in the national trauma registry and broaden the understanding of disparities in patient experiences.


Background
Trauma patient follow-up in the United States (US) has challenged researchers and trauma care teams for decades. 1 Data from US trauma patients' medical records are abstracted into local, state, and national trauma registries to guide intervention and prevention efforts, direct quality improvement, track patient health trajectories and outcomes, and examine the ways in which provider and systemic factors may contribute to disparities in outcomes. 2,3However, current gaps in this system limit researchers' ability to identify groups at highest risk for disparities in functional recovery.5][6] Despite this critical need, and the successful implementation of follow-up procedures in many other countries, 7,8 post-discharge follow-up in the US has remained a considerable challenge.Moreover, signi cant disparities exist in trauma care follow-up, with persons of color and those with higher poverty and lower education levels consistently presenting lower follow-up rates. 9Given these gaps in data collection, there is a signi cant need for a feasible, cost-effective, and culturally resonant follow-up data collection process to be incorporated into existing trauma care systems.
The COVID-19 pandemic has presented additional challenges to reaching patients after discharge.With pandemic-related restrictions on in-person contact and post-discharge follow-up, data collection has been and continues to be adapted to virtual modalities including text and email.Prior research with hard-toreach trauma populations, including patients with increased environmental instability and/or co-occurring mental health and substance use disorders, has consisted of persistent and intensive follow-up outreach efforts, with limited success. 10These efforts are not sustainable long-term or feasible in the context of the COVID-19 pandemic, and there is a need for follow-up methods that reduce the burden on research and hospital staff.National trauma care experts have expressed an interest and need to nd a feasible, cost-effective, and low-burden method to follow-up with trauma patients. 4Moreover, trauma registry experts have noted that due to limited resources, calling patients individually is not feasible for long-term follow-up, and an automated system is needed. 4e current study assessed post-injury survey response rates across different outreach modalities including mail, telephone, text, and email to determine feasibility of collecting patient-reported follow-up data from trauma patients 6 months after hospital admission, as well as most e cient data collection modalities.

Design
This study was part of a larger prospective cohort study conducted at a level-1 trauma center in the Paci c Northwest.Interviews were conducted with n = 245 racially and ethnically diverse trauma patients to develop a culturally resonant data collection system for equity-related measures.Six months after enrollment, study participants were contacted to complete the PROMIS-29, 11 a follow-up health-related quality of life survey measure.To test the feasibility of an automated system, researchers used Twilio, a low-cost ($0.007) module within the Research Electronic Data Capture (REDCap) survey platform 12,13 that automatically sends text messages or emails at speci ed intervals.All participants were given a $10 gift card upon completion of the interview.Study procedures were approved by the University Institutional Review Board (IRB).The STROBE guidelines were used to ensure proper reporting of methods, results, and discussion.

Eligibility
Potentially eligible patients were identi ed via electronic medical record review.Inclusion criteria were: 18 years or older, admitted for at least 24 hours with a physical injury (preliminary ICD-10 codes S00-T88 or V00-Y99), and had the ability to consent and interview in English or Spanish.Patients were excluded if they had a burn injury, were under law enforcement supervision, were under care for a secondary complication to a prior injury sustained more than 2 weeks prior to screening, or were unable to consent to participate in research as judged by nursing staff (e.g., those with cognitive impairments).

Recruitment and Procedure
We used a purposive sampling strategy to ensure a racially and ethnically diverse sample.The recruitment process was modi ed because of changes to study site procedures due to the COVID-19 pandemic (Table 1).Protocol 1. Patients in protocol 1 were enrolled from 6/24/19 to 3/6/2020.These patients were approached by the bedside, provided with the purpose of the study, consented, and completed a 25-40-minute audio-recorded interview (SDC).Participants (N = 136) were asked to provide their phone, email, and mailing address, as well as additional contact information for two friends or relatives, their work, and their health clinic.Six months later, patients were contacted up to 10 times per week via multiple simultaneous outreach methods including phone, text, e-mail, and mail.They were asked to participate in a 10-minute interview during which they completed the PROMIS-29, a short form assessment containing one pain intensity question and four items from seven domains: depression, anxiety, physical function, pain interference, fatigue, sleep disturbance, and ability to participate in social roles and activities. 11Participants were eligible to complete the follow-up survey from 5-7 months after initial interview.In an effort to yield high follow-up response rates, 10 research staff prioritized contacting patients for follow-up in as many ways possible and engaged in phone, text message, and email outreach concurrently (Fig. 1).Paper surveys with return postage were sent to participants who had not responded during second week of months 6 and 7 after interview.Protocol 2. Due to restrictions on in-person contact stemming from the COVID-19 pandemic, recruitment for protocol 2 patients was conducted remotely from 4/15/2020-8/30/2020. Potentially eligible patients were called while they were admitted to the hospital and offered the option to participate in the interview then or be contacted again post-discharge.
Potentially eligible patients who had already been discharged from the hospital were contacted by phone up to 10 times in the two weeks after discharge.Once they had consented to participation and completed the interview, participants were asked to provide the same contact information as protocol 1 participants.2) Mailed survey week 2 of months 6 and 7 1) Letter sent 2 weeks before eligibility 2) Initially offered only one automated response modality (text or email, based on participant preference) for the rst 1-2 weeks of eligibility (Fig. 1) 3) Mailed survey week 2 of months 6 and 7 To increase follow-up survey response rates and further test the feasibility of low-cost automated contact options (i.e., text and email sent through Twilio), the study team implemented a new 6-month follow-up contact procedure during the COVID-19 pandemic (Fig. 1).Prior research demonstrates that sending a letter before asking participants to complete follow-up surveys increased telephone survey response rate by 11%. 14In line with this evidence, the investigative team sent a letter by mail to participants (N = 109) two weeks prior to them being eligible for 6-month follow-up; participants were eligible to complete the survey 5-7 months after initial interview.Additionally, in line with prior literature demonstrating increased response rates when offering only one response modality at a time, 15 participants were only offered one modality (either text or email) for the rst 1-2 weeks of eligibility (Fig. 1).Paper surveys with return postage were sent to participants who had not responded during second week of months 6 and 7 after interview.

Data Analysis
Once participants completed the follow-up survey, research staff recorded the modality by which the participant responded.Descriptive statistics were used to summarize participant demographics, report the frequency of follow-up response by modality, and overall response rates.

Results
A total of 178 individuals out of 245 eligible participants responded to the 6-month follow-up survey: 88 using protocol 1 and 90 using protocol 2 (Table 2).Overall response rate was 64.7% for protocol 1 and 82.6% for protocol 2. Most (n = 124; 69.6%) self-identi ed as persons of color.Overall, after implementing new follow-up contact procedures, the response rate by any modality increased by 17.9 percentage points from protocol 1 (64.7%) to protocol 2 (82.6%).In protocol 1, the primary response modality was by phone (72.7%), and email and text only made up a combined 12.5% of responses.In contrast, the combination of email (47.8%) and text (14.4%) made up more than half of the responses during use of protocol 2 (62.2%), with phone accounting for only 35.6%.Mail responses also decreased between the two phases (14.8-2.2%).
The 90 participants from protocol 2 who responded to the 6-month follow-up survey were asked if they had received the letter indicating they would be soon contacted to complete the follow-up.Among these, 53.3% (n = 48) indicated they recalled receiving the letter.These individuals were additionally asked how receiving the letter in uenced their willingness to respond to the survey; 47.9% (n = 23) indicated the letter made them more likely to respond, 47.9% (n = 23) indicated the letter did not in uence their likelihood of responding, and 4.2% (n = 2) indicated the letter made them less likely to respond.

Discussion
The current study is an innovative test of the feasibility of collecting follow-up data from racially and ethnically diverse patients in the US who have experienced injury via cost-e cient and low-burden automated modalities including text and e-mail.Findings from this investigation reveal that the implementation of a follow-up outreach procedure that included sending a letter two weeks prior to requests for follow-up completion and reducing the number of options for initial contact was associated with an increase in response rate.Furthermore, implementation of this purposeful modi cation in outreach procedure changed the most common response modality from phone, a higher cost and higher administrator burden modality, to a combination of email and text message, which may be automated and therefore more cost-e cient alternatives.Taken together, our results suggest that follow-up data can feasibly be collected from trauma patients by implementing an automated and cost-e cient data collection process.
This investigation identi es several areas for future research.First, this investigation is US centric and does not consider successful post-discharge follow-up procedures that have been implemented in other countries. 7,8Future research should be informed by lessons learned from successful post-discharge follow-up efforts implemented in other countries.While this study did not assess patient preference for follow-up outreach modality, recent evidence suggests that patients who speak a language other than English may have speci c preferences for non-phone outreach. 16Assessing patient preference should be a priority in future studies.Furthermore, recent concerns have been raised regarding cybersecurity speci cally with Twilio and REDCap. 17Although the current investigation does not have the means to address these concerns, cybersecurity remains a critical issue and must be addressed in future research.
While this study offers important insights into follow-up with hard-to-reach trauma populations, limitations must be noted.Results from this investigation could be impacted by selection bias.Due to COVID-19, recruitment procedures were solely remote for some of the participants.With remote recruitment, these participants had to answer the phone in order to be enrolled in the study, whereas protocol 1 participants were approached and enrolled by the bedside.This difference in recruitment strategies could suggest fundamental differences between cohorts.There were also several changes made between the two protocols and investigators cannot be sure whether improvement in response rate is due to one change or a combination of the changes.Additionally, future research should include a detailed cost analysis to ensure that costs of implementing this follow-up protocol are sustainable.Finally, all study participants were recruited from one level-1 trauma center located in the northwest.Further research is needed to ensure ndings are generalizable on a national scale.
Recent scholarship has established expert consensus on outcome measures for data collection from trauma patients. 6However, there is a demonstrated and signi cant need for the implementation of a feasible and cost-effective follow-up data collection process to be incorporated within existing trauma care systems.Trauma registry experts have identi ed that this system should prioritize automated data collection processes to increase willingness of trauma centers to implement procedures for long-term follow-up with trauma patients. 4This investigation provides evidence that the implementation of a coste cient and automated data collection process may be both feasible and effective for following up with US trauma patients and ultimately expanding the national trauma registry to include data crucial for promoting long-term positive outcomes.

Supplementary Files
This is a list of supplementary les associated with this preprint.Click to download. STROBEchecklistcrosssectionalResponseModality.docx

Table 1
Similarities and Differences Between Protocol 1 and Protocol 2

Table 2 .
Demographics of participants who responded to 6-month follow-up survey according to self- *Race categories are not mutually exclusive and some participants self-identi ed as more than one race ^Cell sizes <5 not shown for participant con dentiality