The Advisory Committee on Immunization Practices (ACIP) recommends influenza vaccination for all individuals ≥6 months-old [1]. Understanding influenza vaccine safety is important for annual vaccination programs and pandemic preparedness. Influenza vaccine is the only vaccine recommended annually, with formulations often changing yearly [2]. While seasonal vaccine safety patterns are usually similar, unexpected adverse events may arise. For example, the 2010–2011 trivalent inactivated influenza vaccine was associated with an increased febrile seizure risk in young children [3].

Although following a large number of vaccinated people prospectively for post-licensure vaccine-associated adverse events can be labor-intensive and expensive, we and others have successfully used text messaging- a scalable, low-cost method- for this purpose [4–8]. Currently, 95% of American adults have a cell phone [9]. Cell phone use is thought to be higher in harder-to-reach low-income populations [10] that may not be included in vaccine safety monitoring programs relying on information from managed care organizations [11]. Text messaging is particularly useful for monitoring of non-medically attended events, such as fever, since it allows patients or caregivers to report symptoms directly. This may be important in an influenza pandemic when information may need to be gathered rapidly and simultaneously from large numbers of vaccinated people nationally, but visits to medical facilities may be limited. Some important issues remain in the use of text messaging for monitoring vaccine-associated adverse events. First, adverse events may occur several weeks post-vaccination, but previous pediatric studies have focused primarily on the more immediate 7–10 days post-vaccination. Second, unlike in Australia [12], text messaging has not been used to date in the United States to assess vaccine safety outcomes at multiple sites, but multi-site vaccine safety monitoring would have value for pandemic and emergency vaccination program preparedness. Third, although wheezing is not considered a safety concern after IIV, accuracy of text messaging to capture respiratory events may be pertinent for monitoring safety of live attenuated influenza vaccination (LAIV) [13]. During our study period, ACIP recommended LAIV preferentially for healthy 2- through 8-year olds for the 2014–15 season; for 2015–16 season there was no preferential recommendation, but LAIV was an acceptable option for use in healthy children [14,15].

This study’s primary goal was to assess feasibility and acceptability of text messaging to assess wheezing prospectively from day 0 (vaccination day) through day 42 post-vaccination in children aged 2–11 years-old receiving quadrivalent live attenuated (LAIV4) or inactivated (IIV4) influenza vaccine. We hypothesized that (1) at least 80% of wheezing symptoms reported via text message would be verified by phone interview, in those able to be reached, and medical record review for those with a visit; (2) response rates to text message queries regarding wheezing symptoms would be higher on days 3 and 7 post-vaccination than on days 14–42; and (3) at least 80% of parents would have a high level of satisfaction with using text messaging to report wheezing occurring post-vaccination. A secondary objective was to assess feasibility of monitoring post-vaccination wheezing and fever using text messaging at multiple sites. We hypothesized that text message response rates would not differ by more than 10 percentage points between sites.

This prospective observational study was conducted during the 2014–2015 (Year 1) and 2015–2016 (Year 2) influenza seasons, in collaboration with the Centers for Disease Control and Prevention (CDC). During the 2014–2015 season, children were recruited from three community-based clinics affiliated with New York-Presbyterian Hospital/Columbia University Medical Center (CUMC) in New York City. These sites serve a primarily Latino, publicly-insured population and share an electronic health record (EHR) system with the hospital. During the 2014–2015 and 2015–2016 seasons, children were recruited from a private general pediatric practice in Fall River, Massachusetts affiliated with Boston Medical Center (BMC) that has its own EHR system. Each child also had a parent or guardian participate in the study. All text messages for both sites were sent from a centralized program to the parent (or guardian) participating in the study. Vaccination decisions were made by the patients’ health care provider and caregivers, and all influenza vaccines administered were the quadrivalent forms. The influenza vaccine strain composition differed in the two seasons for both an A and B strain. CUMC and BMC’s Institutional Review Boards (IRB) approved the study; CDC relied on the CUMC approval.

Children were eligible for enrollment if they (1) were 2–11 years-old, (2) were receiving their first or second influenza dose of that season, (3) their parent had a cell phone with text messaging capabilities, and (4) their parent spoke English or Spanish at CUMC sites or English at the BMC site. Exclusion criteria included (1) any chronic medical condition considered a contraindication or precaution for LAIV (with the exception of asthma/wheezing history) [14], (2) current/recent (<2 weeks) asthma exacerbation, (3) oral or other systemic steroid use in the preceding 2 weeks, (4) temperature (T) ≥100.4 °F at vaccination, (5) antipyretic administration within 6-hours pre-vaccination or stated intent to use prophylactic antipyretics, (6) parental inability to read or send text messages, and (7) sibling or child already enrolled in either season. Receipt of other vaccines was not an exclusion criterion.

Two-hundred and sixty-eight children and their parent/guardian were enrolled: 168 in 2014–2015 and 100 in 2015–2016. Two participants were excluded from analysis (Fig. 1). Most children (72.9%) received IIV4 (Table 1). Demographic characteristics differed by recruitment site (Supplemental Table 1). The proportion of children with parent-reported history of asthma/RAD/recurrent wheezing was 32.5% in 2014–2015 (enrollment began in December) and 69.0% in 2015–2016 (enrollment began in October).

Chart adjudication identified 131 (49.2%) children with a history of asthma/RAD/recurrent wheezing. Nearly all children with an asthma history received IIV4. Most (105/112; 93.8%) children with a chart-adjudicated history of asthma/RAD/recurrent wheezing in the last 12 months also had parent-reported asthma medication use in the last 12 months. Most (105/111; 94.6%) children with parent-reported asthma medication use in the last 12 months also had a recent chart history, and the remaining had a remote chart history.

Three (1.1%) participants requested to stop messages and two (0.8%) lost text message or phone service during the 42 days observation period. Half (53.0%) returned the diary in Year 1 and 78.0% in Year 2. Individual text message response rates were above 70% on all days across all sites (Fig. 2) though rates varied by day. Mean response rate for those who had text messages delivered was 88% ± 25% overall and was similar between the study period d3–7 versus d14–42 (90% ± 25% vs. 88% ± 27%; p = .26). Mean response rate was significantly higher at the BMC-affiliated site (95% ± 15% vs. 79% ± 32%; p <.001). Due to demographic differences by sites, results were stratified. When stratified by site, there were no differences in response rates by age, gender, preferred language, education level, text message plan or utilization of the respondent, or age, sex, race, ethnicity, insurance status of the child. In addition, across both sites combined, all demographic sub-groups had response rates of ≥80% with the exception of those who did not have unlimited texting (76% ± 36%) or did not text daily at baseline (74% ± 31%).

We demonstrated in this novel study that text messaging is a reliable method to identify post-vaccination respiratory symptoms both in the more immediate seven days post-vaccination as well as through 42 days post-vaccination with most episodes- both medically attended and non-medically attended- being captured by text message. It also showed acceptability with high response rates and parental satisfaction. Finally, it was a successful demonstration of a central text messaging program with inclusion of a geographically distant health facility with no previous experience with text message monitoring. Response rates were high across sub-populations, which may be important for planning future pandemic or emergency vaccination safety programs.

Wheezing is considered to be an important outcome of interest in vaccine safety assessments [19], and there was a specific concern about wheezing after LAIV [17,20,21]. As noted above, LAIV was in use during the seasons our study was conducted 2014–15 and 2015–16). Subsequently, the ACIP recommended that LAIV should not be used in the U.S. during 2016–17 and 2017–18 influenza seasons, due to concerns about its effectiveness [1,22]. However, LAIV continues to be recommended and used in some other countries such as Canada and United Kingdom [23,24]. We saw no difference in wheezing/chest tightness by vaccine type in children with a recent asthma/RAD/recurrent wheeze history, but very few of those children received LAIV since there was a precaution against the use of LAIV in children with asthma or recurrent wheezing. Other studies have shown that IIV does not significantly increase asthma exacerbations immediately after vaccination in children over three years of age [25,26], and our proportion of wheezing/chest tightness episodes were similar to published data [25]. In our study, the majority of parent-reported wheezing and/or chest tightness episodes did not result in medically-attended visits, highlighting the importance of methods to collect data directly from participants as a complement to EHR monitoring. While these events may be considered as medically less serious since they do not require medical care, they may affect parental perceptions of vaccine safety [27–31]. Being able to assess a fuller complement of both medically-attended and non-medically-attended adverse events would be advantageous when evaluating a new vaccine.

While there could have been a potential concern that having parents directly report vaccine safety data may make them feel less safe about a vaccine, we found the opposite was true. This is similar to our findings in pregnant women for whom taking part in text message monitoring led to a perceived increase in positive views about vaccine safety [32]. Two potential reasons for this positive effect are that, first, parents are more aware that vaccines are carefully monitored [33,34], and, second, that as parents pay more attention in the post-vaccination period they realize their child did not experience any health issues.

A number of lessons were also learned about using text messaging for vaccine adverse event monitoring. First, while response rates drop off when texting daily for a number of days in a row, as seen in our previous studies [4,5], sending messages weekly for up to 6 weeks appears to be well-tolerated with sustained response rates. Second, it is helpful to build into the text messaging error messages that ask for clarification to cut down on erroneous responses, as well as to have unanswered text messages re-sent after 20 minutes to help prompt a response. For example, in response to the first message of the temperature cascade sent on day 0, an automated message was triggered 6.4% of the time to clarify a response; 88.2% resulted in correction of the initial response. In addition, a repeat message was triggered 27.1% of the time due to lack of response; 91.7% responded with data. Finally, specific to screening for asthma, we found that parent report of asthma medication use within the last 12 months was a fairly accurate indicator (>90%) of a chart-adjudicated recent asthma/RAD/recurrent wheeze history, which may be helpful for future studies.

This observational study had a number of limitations. While there was potential for underreporting, most episodes of wheezing and/or chest tightness were reported via text message. We could not reach all parents by phone to confirm symptoms in year one, and respiratory symptoms were confirmed by phone but not by medical exam. However, if a system such as this were used in a pandemic most episodes would also not be verified by a provider examination. There is previous evidence that caregiver report of lower respiratory symptoms in young children is sensitive, particularly when reporting wheeze [35]. Of note, if text messaging were to be used in a large scale post-licensure study, a text message report could be used to trigger a study visit, if desired. An additional limitation is that limiting the denominator by data reporting levels may have increased fever and wheezing rates, but overall rates were low. Finally, while the study being conducted at two geographically distant and demographically distinct sites is a strength, it may not be generalizable to other locations.

This study demonstrates the feasibility and acceptability of using text messaging to assess pediatric respiratory symptoms after influenza vaccination through 6 weeks post-vaccination. The results also illustrate the capacity to monitor post-vaccination fever and wheezing across more than one site, which could be helpful during an influenza pandemic or other emergency vaccination program.