Increasing COVID-19 Vaccination Rates for Children With Sickle Cell Disease

This study was conducted at Boston Children’s Hospital, a tertiary academic pediatric center in the United States. Our SCD clinic provides longitudinal care to 300 to 400 children with SCD. During the time of this initiative, the COVID-19 vaccine was offered at Boston Children’s Hospital’s centralized vaccine center but was not directly administered in the hematology clinic.

Our project followed the Model for Improvement, and the SQUIRE 2.0 guidelines were used to develop this manuscript.²⁶  The project team consisted of hematology physicians, nurses, a physician assistant, the vaccine clinic coordinator, and QI specialists. The team met monthly to solicit and evaluate contextual elements that may have contributed to the project’s success, failure, efficiency, and cost.

A Key Driver Diagram was developed at the start of the project (Fig 1). Major drivers included eligible patient identification, increasing patient’s vaccine knowledge, eliminating logistical barriers to vaccination, and increasing provider awareness of their patients’ vaccination statuses. Based on these drivers 3 plan-do-study-act (PDSA) cycles were designed. A SMART aim of vaccinating 70% of patients by June 2022 was chosen. We hypothesized that by June 2022, 70% of children in Massachusetts aged 5 to 19 years would have received ≥2 vaccine doses and felt that achieving a similar vaccine uptake as a similarly aged cohort of children without SCD in Massachusetts would be an important milestone.

PDSA Cycle #1 (January 2022–March 2022): An enterprise data warehouse (EDW) aggregates and stores an organization’s electronic data. An EDW filter was created to identify patients with SCD ≥5 years of age and to determine their vaccination statuses. This list was manually reviewed to remove ineligible patients. Using that report, targeted information encouraging vaccination was sent via text message to the families of children who had not received ≥2 COVID-19 vaccine doses. Text messages were sent via Solutionreach, an automated messaging platform, at a single time to all eligible patients independent of their hematology visit timing. In addition, hematology providers were e-mailed monthly reports with their patients’ vaccination statuses and encouraged to discuss vaccination with those patients either via ad hoc phone calls or at their next clinic visit.

PDSA Cycle #2 (March 2022–May 2022): An automated weekly report of patients with a clinic visit in the next 2 weeks and their COVID-19 vaccine status was sent via e-mail to clinic teams. Patients who were not fully vaccinated were contacted via phone by a vaccination clinic nurse 2 to 4 weeks before their hematology appointment and offered a vaccination appointment on the same day as their hematology appointment. Before this project the vaccination clinic was not contacting patients overdue for their COVID-19 vaccination.

PDSA Cycle #3 (May 2022–June 2022): A promotional video recorded in English was sent to all patients who were not fully vaccinated via text message. The video features a hematologist, the mother of a patient with SCD who had initial skepticism toward the vaccine but decided to vaccinate her family after contracting COVID-19, and a vaccine clinic nurse. The video highlights the importance of the COVID-19 vaccine for children with SCD and reviews the logistics of booking a vaccine. A QR code linking to the video was also posted in our clinic.

The primary outcome measure was the cumulative proportion of patients with SCD who received ≥2 doses of any COVID-19 vaccine (booster dose not required for the purpose of this project). A secondary outcome was the proportion of unvaccinated eligible patients at the start of each 2-week interval that received a vaccine during that interval. Vaccine adherence was assessed over the 5 months before project initiation and then biweekly during the project. The number and dates of vaccinations were obtained from our EDW. Vaccination status was determined based on the number of vaccine doses in our EDW at each 2-week interval. Vaccine data travel to the EDW in 3 ways: (1) vaccines administered at our institution are automatically captured, (2) vaccines administered at other Massachusetts centers can be reconciled by providers via the Massachusetts Immunization Information System, and (3) vaccines administered out of state can be manually added by providers. Vaccination data can be viewed in our electronic health records by all care providers, and patients with overdue COVID-19 vaccinations were flagged by clinical decision support as such. Vaccination status was manually verified by a nurse before a patient’s hematology appointment. As a process measure, we evaluated the effectiveness of vaccine scheduling calls. The categories were vaccine booked, vaccine refused, received a vaccine that was not captured, barrier, or concern to vaccination, and unable to reach. If a patient was classified as “received a vaccine that was not captured,” the doses were then reconciled via the Massachusetts Immunization Information System or manually added. These doses would then be reflected in the EDW at the start of the next period and the patient’s vaccination status would be amended going forward. Vaccination statuses were not amended retrospectively for previous months based on the reconciliation of outside vaccinations.

The number and proportion of missed hematology clinic appointments was used as a balancing measure. We hypothesized that patients who were against the vaccine may cancel clinic visits if vaccination outreach efforts made them feel disconnected from our clinic. The proportion of missed appointments was determined by dividing the number of visits missed by the total number of visits scheduled each month. The number and proportion of missed appointments was assessed over a 5-month period before project initiation and then biweekly during the project.

The number of patients who viewed the educational video was assessed 2 months after the video launched by reviewing the view count on YouTube.

Vaccine adherence over time was assessed using line graphs and statistical process control charts. We applied standard control chart rules to denote special cause variation.²⁷  We also performed an interrupted time series analysis by estimating a logistic regression model with number of patients with ≥2 COVID-19 vaccine doses as the dependent variable and time (measured biweekly), and time-by-study period interaction as the independent variables.²⁸  This model compared the pre- and postimplementation rates while accounting for preexisting trends in the outcome. All analyses were performed using RStudio, version 1.4.1717, and Microsoft Excel.²⁹  The SQUIRE 2.0 guidelines were used to develop this manuscript.²⁶ 

This project was conducted between September 2021 and June 2022. As a QI initiative, this project was exempt from institutional review board review.

Our primary outcome was to identify changes in the cumulative proportion of patients who had received ≥2 COVID-19 vaccinations. During the preintervention period (September 2021–January 2022) the cumulative proportion of patients who had received 1 COVID-19 vaccine dose increased from 6.0% to 8.5%, and the cumulative proportion who had received ≥2 COVID-19 vaccine doses increased from 33% to 41%. During the intervention period (February–June 2022), the cumulative proportion of patients who received ≥2 COVID-19 vaccine doses increased from 41% to 64% (Fig 2). The cumulative proportion of patients ≥12 who received a COVID-19 booster increased from 6% to 23% over the intervention period.

Our secondary outcome was to identify the proportion of unvaccinated patients who were vaccinated during each 2-week period. The mean vaccination rate increased from 2.1% to 7.2%, with evidence of special cause variation starting around January 2022. Additional special cause variation was noted in May 2022 with a vaccination rate above the outer control limit (Fig 3).

Interrupted time series modeling showed that the achieved vaccination rate for children with SCD exceeded the predicted rate by 11% (64% vs. 53% predicted), whereas the achieved vaccination rate for the cohort of Massachusetts children without SCD was 23% below the predicted rate (64% vs. 87% predicted) (Fig 4).

Of the 84 patients who were contacted, 10 (13.5%) booked a vaccine, 6 (8.1%) had received a vaccine that was not captured, 19 (25.7%) articulated a barrier or concern to vaccination, 11 (14.8%) refused the vaccine, and 28 (37.8%) could not be reached. In the 1 month following the release of our promotional video, it was watched by approximately 40% of the patients who received the video.

There was no change in the proportion of missed clinic appointments in the preintervention versus intervention period (8.7% [n = 18] vs. 11.0% [n = 20]). The proportion of missed appointments was also unchanged when analyzing only the proportion of unvaccinated patients who missed their visit in the preintervention (8.2%, n = 11) and intervention (10.3%, n = 10) periods. There was also no difference in the proportion of missed visits among those who refused the vaccine during the patient outreach calls (9.1%, n = 1) versus those who did not (11.1%, n = 19).

In this QI initiative, we increased the COVID-19 vaccination rate for children with SCD from 41% to 64% over 5 months. During the same period the vaccination rate for children 5 to 19 years of age in Massachusetts increased from 55% to 63%.³⁰  Although we did not achieve our SMART aim of vaccinating 70% of patients, there was evidence of special cause variation and of a meaningful shift in the mean (Fig 3). In any given period, the increase in vaccination was modest; however, over the course of the intervention, this resulted in a greater than 2-fold increase in the proportion of vaccinated patients resulting in a clinically meaningful cumulative improvement in vaccination.

Previous QI projects focused on increasing vaccination rates in high-risk pediatric populations have found that eligible patient identification is a crucial first step.³¹  Our automated data pipeline allowed us to readily identify patients without manual chart review. A systematic review of effective interventions to increase COVID-19 vaccination rates demonstrated that 1 successful strategy is to send personalized communication to patients.¹⁹  Our data pipeline allowed us to target eligible patients with personalized text messages.

During our second PDSA cycle, the vaccine clinic phoned patients who were under vaccinated to offer a vaccination. As a result, 13.5% of patients booked a vaccine appointment and 25.7% articulated a barrier or concern that they were interested in discussing at their upcoming visit. Flagging patients interested in discussing the vaccine allowed providers to prioritize nuanced conversations regarding the vaccine with these patients. Patients who were willing to book a vaccine or discuss the vaccine can be characterized as vaccine hesitant as opposed to antivaccine.³²  Individualized outreach to vaccine-hesitant individuals has been shown to be an effective way to improve vaccination rates among vaccine-hesitant individuals.^24,33,34 

Nearly 15% of patients declined the vaccine when phoned. It is important that our clinic continues to maintain a strong therapeutic relationship with these patients for their ongoing SCD care. Our balancing measure of visit cancellations showed that we did not disrupt willingness to continue care despite our vaccine outreach. During monthly meetings that included individuals from the vaccine clinic, hematology clinic, and hospital leadership, we did not identify any adverse effects of this work on clinic efficiency, provider satisfaction, or resource utilization.

To our knowledge, our video is the only freely available YouTube video that specifically discusses the importance of the COVID-19 vaccination for children with SCD. Our informational video was watched by 40% of the eligible patients. Although a higher viewership was anticipated, our views greatly exceed that of studies of similar interventions in adults.²⁴  Special cause variation was noted in vaccination rates 1 week after the distribution of the video, which may indicate that the video was a significant motivator for patients to get vaccinated.

This study has several important limitations. The first is that the vaccination rate among children with SCD increased during the preintervention period and therefore may have continued to increase during the intervention period in the absence of this project. This issue is complicated by the fact that although a cumulative percentage of patients with SCD who had been fully vaccinated was felt to be the most clinically meaningful measure, the denominators of eligible patients from 1 period to the next are not independent and there is likely autocorrelation. We addressed this issue in 2 ways: first, we present a noncumulative percentage of patients with SCD vaccinated against COVID-19 for each period. Although these increases look modest, we demonstrate a greater than 2-fold increase in vaccination rates over the intervention period with an increase in the mean from 2.1% to 7.2% on February 1. Although special cause variation was also noted on May 15, we chose not to shift the mean line because the increase in vaccination rates was not sustained in the periods that followed. Second, we compared the vaccination rate achieved to the counterfactual prediction of the vaccination rate for children with SCD and exceeded the prediction. This suggests that our interventions provided additional momentum in the vaccination effort above and beyond what would have occurred in the absence of this project. This becomes even more apparent when we compare the vaccination trend to that of a cohort of comparable aged children in Massachusetts. The Massachusetts pediatric cohort also had an increase in vaccination rates during the preintervention period but saw a large drop in vaccine momentum during the intervention period. The SCD cohort may have seen a similar drop in momentum in the absence of this work. In addition, during the preintervention period hematology, providers were aware of this QI project and may have been more cognisant to discuss COVID-19 vaccination with patients contributing to the increasing rate during this period. The second limitation is that some patients with SCD may have waited longer than patients without SCD to get the vaccine delaying the rise in the proportion of patients with SCD that are vaccinated relative to the patients without SCD. This is unlikely to be the main contributor to the increase in vaccination observed in this study given the clear temporal increase in vaccination related to the start of each PDSA cycle. The third limitation is that our study may have underestimated the total number of patients with SCD who received a vaccine. Vaccines administered at other centers and out of state need to be reconciled or manually added by a provider. During patient outreach calls, 8.1% of patients indicated that they had received a vaccine that our system had not captured. In addition, our outreach efforts were exclusively provided in English despite a significant number of our patients requiring an interpreter. We also provided educational materials via text messages and QR codes, and patients without mobile devices may not have been able to view these materials.

Future work will focus on targeting populations with low English proficiency by translating our educational materials into Spanish. We are working on engaging with community partners to offer patients with SCD the COVID-19 vaccine at community-based events. We also are expanding this work to include COVID-19 booster shots, patients between 6 months and 5 years of age, and patients in other programs that are at a high risk of COVID-19. This work can be applied in other contexts because our interventions are generalizable, do not require specialized resources, and are not cost prohibitive. In addition, the use of automated reporting tools and clear communication pipelines will allow us to maintain a sustainable program focused on improving COVID-19 vaccination rates. In summary, increasing vaccine access can improve COVID-19 vaccination rates.

The authors thank Dr Anne Stack for her guidance, mentorship, and expertise.

EDW

enterprise data warehouse

PDSA

plan-do-study-act

QI

quality improvement

SCD

sickle cell disease