Improving Pneumococcal Vaccination Rates in High-risk Children in Specialty Clinics

CMKC is a large tertiary pediatric hospital located in the Midwest of the United States. CMKC has a population health management subsidiary, which partners with and integrates data from commercial and Medicaid payers, CMKC, and nearly 40 community-based practices to advance value-based care strategies and interventions.^26,27  During one of the subsidiary’s monthly meetings with community pediatricians, the team learned that a vast majority of primary care physicians did not feel accountable for administering PPSV23. Some of the reasons identified included an inability to stock low volumes of the vaccine, lack of insight into qualifying diagnoses, lack of comfort in administering the vaccine to patients with high-risk conditions, and a general assumption of specialist accountability. The team recognized the value of the subsidiary’s comprehensive population health data infrastructure, inclusive of electronic medical record (EMR) data, CMKC scheduling data, payer claims, and state immunization data, to inform, monitor, and sustain pneumococcal vaccination improvement efforts.

The CMKC Institutional Review Board evaluated the project and determined that the proposed activity did not involve research.

The population health management subsidiary partnered with a technology vendor and an infectious diseases physician to develop and program a high-risk pneumococcal report, starting in July 2017 to help assess baseline performance across specialty divisions. Specialty divisions were recruited on the basis of established QI efforts in this area, volume of patients at risk for IPD, and interest. These initial steps led to the formation of our multidisciplinary project team, which included QI champions from the divisions of rheumatology (1 physician), infectious diseases (2 physicians), and pulmonology (1 clinical pharmacist), in addition to leaders from CMKC’s population health management subsidiary (1 physician and 2 population health management team members). The rheumatology division had been working on QI efforts to increase pneumococcal vaccination since 2016, starting with systemic lupus erythematosus patients and expanding to other immunosuppressed patients as well. The team began meeting on a quarterly basis starting in May 2018. We created a cause and effect diagram to identify reasons why patients did not receive recommended pneumococcal vaccination (Fig 1) and designed a key driver diagram to highlight interventions that could lead to improvement (Fig 2). These methods helped us identify interventions with the highest impact.

One early project step focused on an in-depth review of patient inclusion criteria per recommendations for immunization by the Centers for Disease Control and Prevention.¹  Although the population health management vendor provided diagnostic and other applicable codes, we recognized the importance of reviewing the qualifying codes (a “value set”) for conditions managed by the participating specialty divisions. Not only was this review critical for building trust in the data and measure algorithm, but it also ensured the accuracy and effectiveness of the automated measurement. After we discussed and reviewed the coding logic, we validated and finalized the report in summer 2017 by manually reviewing >10% of charts.

The first intervention to directly impact patient care occurred in May 2018 when each division ensured that medication stations at all clinics were adequately stocked with PCV13 and PPSV23.

The team recognized that one of the most significant barriers to administrating PPSV23 was the additional time needed to review immunization history to evaluate whether and when the patient had received PCV13 and/or PPSV23. Using the organization’s broad population health management data infrastructure, the team designed and deployed an automated previsit planning report (Supplemental Information). The report contained all eligible patients who would be seen in the subsequent 2 weeks in the participating division’s clinic, the date of next visit, location and clinic, provider, attributed division, pneumococcal status (“needs PCV13,” “needs PPSV23,” or “PPSV23 received”), PPSV23 qualifier (diagnostic code making them an eligible patient), medical record number, first and last name, and date of birth. The report also included the specific dates patients received PCV13 and PPSV23, if applicable. This report was sent weekly via secure e-mail starting in August 2018 to each division’s designated contacts.

In addition to team interventions, each participating division developed its own process to review and integrate the report in clinic workflow. All divisions provided education to providers and staff, in addition to reviewing progress at division meetings. Different clinical team members reviewed the report in each division (administrative assistant for rheumatology, clinic nurses for infectious diseases, and clinical pharmacist for pulmonology). This was followed by direct communication with providers seeing the patients regarding pneumococcal vaccine eligibility, and the process incorporated a review of the chart to confirm eligibility. Rheumatology and pulmonology incorporated pneumococcal vaccination review in preclinic huddles; pulmonology also included this information in a preclinic planning form. All patients required consent before vaccination, per our hospital policy, and received Centers for Disease Control and Prevention handouts on the vaccine provided.

The multidisciplinary team met quarterly to review performances at an individual and collaborative level, share current practices, discuss successes and challenges, and adapt changes to the different clinics. For example, the automated previsit planning report contained a lot of information, including 33 columns of data, so a new “summary” tab that included only essential information (name, date of birth, next scheduled visit, clinic location, provider, pneumococcal status, and PPSV23 qualifier) was created to more efficiently review and integrate within each clinic’s workflow.

Our outcome measure was the percent of patients eligible and attributed to the 3 clinics who received PPSV23 in their lifetime. Our numerator was the number of patients with a high-risk diagnosis or on immunosuppressive therapy who have received PPSV23 ever; the denominator was the number of patients with a high-risk diagnosis or on immunosuppressive therapy that have had at least 2 encounters for a high-risk diagnosis in the past 1095 days that were attributed to rheumatology, infectious diseases, or pulmonology. We evaluated this measure as an aggregate and using division-specific data. Attributed providers were defined by using a “most frequent, most recent” algorithm over a 2-year period. The algorithm attributes each patient to the specialty provider seen most frequently for a high-risk diagnosis. If 2 providers had the same number of encounters, the provider who had seen the patient most recently was attributed. High-risk diagnoses included in our measures were systemic lupus erythematosus and other conditions associated with immunosuppressive therapy (rheumatology), HIV (infectious diseases), and cystic fibrosis (pulmonology). Inclusion criteria were patients 2 to 21 years old with at least 2 occurrences of an outpatient, inpatient, or emergency department visit with any high-risk diagnosis in the last 3 years. Diagnoses associated with laboratory tests and imaging studies were not evaluated for inclusion because of potential for inaccurate coding. Exclusion criteria included patients ever in hospice and patients attributed to a nonparticipating division.

Our process measure was the number of patients each month who received a PPSV23 in any of the 3 clinics. As a balancing measure, we evaluated whether any patient who received PPSV23 throughout the intervention period had an adverse vaccination diagnosis (anaphylactic reaction due to vaccination, postimmunization acute necrotizing hemorrhagic encephalopathy, and adverse effect of vaccines) based on claims data within 14 days from receiving the PPSV23.

We gathered PPSV23 vaccination rates from an automated report generated from the measure definition. We created time series graphs for our outcome measure and run charts for our process measure to monitor for improvement over time. Data were analyzed by division and as a combined rate of all 3 clinics. Our data collection started in July 2017, which is when the report with the desired value sets was retrospectively activated. The interventions that directly impacted patient care began in May 2018 and, therefore, we used the 10 data points from July 2017 to April 2018 for our baseline. We used 2 standard run chart rules to analyze for nonrandom signals of change for our process measure: (1) shift: ≥6 points in a row above or below the center line; and (2) trend: 5 consecutive points increasing or decreasing.²⁸ 

Project steps and interventions are annotated on Fig 3. The combined PPSV23 vaccination rate included an average of 564 eligible patients for each measurement period, including 158 in rheumatology, 67 in infectious diseases, and 339 in pulmonology. As shown in Fig 3, overall PPSV23 vaccination rates for applicable high-risk patients across the 3 specialty clinics increased from ∼20% at the start of the efforts in July 2017 to ∼60% by the end of 2019. The improved vaccination rate was sustained for the last several months measured.

The PPSV23 vaccination rate in the rheumatology division began at ∼35% and increased to ∼57% (Fig 4A). The infectious diseases division’s PPSV23 vaccination rate started at ∼47% and increased to ∼60% (Fig 4B). The initial vaccination rate in the pulmonology clinic was at ∼12% and increased to ∼63% (Fig 4C).

There were some improvements in vaccination rates seen before initiation of our patient-impacting interventions (Fig 3) that were mostly driven by rheumatology (Fig 4A).

Our process measure of the number of patients receiving PPSV23 in 1 of the 3 clinics was tracked monthly. Our baseline median was 5.5 patients, and this increased during the first year of our project to 28 after a shift. We had a downward shift in our data starting in January 2019, for a final median of 9 patients per month (Fig 5). For the balancing measure of any adverse events related to PPSV23 vaccination, our claims-based report did not identify any patients with pertinent billing codes.

By standardizing evaluation of pneumococcal vaccination and eligibility, creating an automated report, getting buy-in from stakeholders, developing a previsit workflow process, and stocking vaccines in specialty clinics, we improved pneumococcal vaccination rates of high-risk patients in our 3 specialty clinics by 40% over an 18-month period with sustainability in improvement. The improvement seen before patient-impacting interventions were mostly driven by the rheumatology clinic. It is possible this improvement was due to the previous QI project initiated in rheumatology or heightened awareness because of discussion of the project before its initiation.

This was a multidisciplinary, multidivision effort at a large children’s hospital to align resources and efforts to target many high-risk patients. Our project core was our hospital’s population health management subsidiary that can access claims-based and EMR data, which was essential, given limitations in our statewide immunization registries. The automated report with eligibility reasons, vaccination dates, and recommendations allowed for previsit planning and bypassing manual chart reviews. Although this project included necessary standardization (automated report), it was flexible, allowing each division to use the report as best suited for them, given variable clinic workflow. In addition, our project highlights the sustainability of our improvements, especially as new eligible patients were added each month.

Some specialists expressed concern regarding their responsibility and accountability to administer the PPSV23. The Infectious Diseases Society of America’s clinical practice guideline states that vaccination of immunocompromised patients is a shared responsibility with the primary care provider and specialist.⁷  To build off of this premise, our report identified eligible patients regardless of qualifying condition (eg, a patient with diabetes being seen in rheumatology clinic), prompting participating clinics to vaccinate any eligible child. This project shows that any provider can and should take every clinical opportunity to vaccinate a child to prevent infectious complications.

There have been reports of improving pneumococcal vaccination in a variety of chronic diseases both in children and adults.^23,29  Most reports, however, have been focused on manual screening, which is not a highly reliable process, especially if the screening is dependent on an individual person.^19,21,22,30  Our validated automated report bypasses this time-consuming step and creates a more sustainable process that is not dependent on a single individual and is exportable to other specialties. When Garg et al²¹  surveyed adult rheumatologists at 1 site to better understand barriers to pneumococcal vaccination for high-risk patients, they found forgetfulness and insufficient time to be the most significant barriers. Our automated report eliminates both of these potential barriers. In addition, the report provides dates of previous pneumococcal vaccinations, thereby decreasing the risk of incorrect vaccine timing. Previsit planning is another common successful intervention in many vaccine QI projects^19,21,22 ; we found that to be useful in our project as well, particularly with the preclinic huddles. The most striking increase we had in our project was seen in the pulmonology clinic; we suspect that is due to the focus on a single disease (cystic fibrosis) that has a dedicated multidisciplinary team that includes our project champion. In addition, their team met weekly and incorporated patient pneumococcal vaccination status in preclinic huddles and checklists.

We plan to expand our project to other divisions within our institution as the population health management subsidiary collects data on all patients eligible for pneumococcal vaccination. Our work could also be generalized to other clinics and hospitals that have access to a database capable of generating a report similar to ours (Supplemental Information). In addition, other clinics could substitute an EMR report, especially if linked to a reliable state immunization registry, to replicate this work.

Our QI project has limitations related to the automated report. There was variability in identifying certain patients at risk due to immunosuppressive medication use because of limited payer medication claims data and the encounter-based attribution model. This limitation led to a relatively low number of eligible rheumatology patients appearing on the report because their eligibility was dependent on a medication (except for systemic lupus erythematosus), rather than a specific diagnosis. The attribution model of the report also had some limitations because we came across examples of patients with >1 chronic disease that led to attribution to a clinic that was not primarily managing their eligible diagnosis. In addition, we did not capture improvements in vaccinations of patients attributed to other divisions. There were instances when the report did not capture the patients’ complete vaccination status, particularly for vaccines administered outside the catchment area of the population health data set. In these circumstances, we discussed any potential discrepancies with the parent or guardian, and shared decisions were made.

Although the report captures the receipt of 2 PPSV23 vaccinations (if the second one was given 5 years after the first one), our outcome measure reported patients up to date if they had ever received a PPSV23 in their lifetime. Although this limitation is unlikely to have affected our numbers greatly because only patients with asplenia or immunocompromising conditions would need 2 doses, it could lead to inflation of our vaccination rate. In addition, although the report prompted individuals to administer PCV13, if needed, before PPSV23, we did not have a separate PCV13 measure.

Our process measure saw 2 shifts, 1 positive and 1 negative; however, we suspect that the bulk of our eligible patients received this vaccine within our project’s first year, leading to a presumed drop in eligible patients after that. Our final median continued to be higher than our baseline as clinics continued to administer this vaccine to new and established high-risk patients.

On the basis of changes that our project team made to eligible patient criteria, patient denominator numbers changed, and the report could collect accurate data only retrospectively starting in July 2017. Our outcome measure data, therefore, started when project discussions began, but improvements in vaccination rates may have started before our project’s initial clinical plan-do-study-act cycles because of heightened awareness of this clinical need.

Finally, there were some improvements seen before our QI project initiation, so there is a possibility those improvements could have continued even without our interventions, although the degree of improvement seen after our QI project was initiated would argue against this theory.

Using QI methodology, we increased PPSV23 vaccination rates in 3 pediatric specialty clinics, and this improvement was sustained. One of the next steps moving forward will include collecting reasons for missed vaccination opportunities, so those could be addressed appropriately to increase our vaccination rates further. We will continue to identify best practices and are actively recruiting additional specialty divisions. This initiative will have an opportunity to reach >9000 high-risk pediatric patients at our institution who need protection from IPD.

We acknowledge all rheumatology, infectious diseases, and pulmonology providers, nurses, pharmacists, and staff. A special thanks to Annie Wirtz, PharmD, BCPPS, who helped with vaccine stock, and Amy Ivy, BA, BS, in rheumatology for her assistance. We thank the medical writing center at CMKC for editing this article.

CMKC

Children’s Mercy Kansas City

EMR

electronic medical record

IPD

invasive pneumococcal disease

PCV13

13-valent pneumococcal conjugate vaccine

PPSV23

23-valent pneumococcal polysaccharide vaccine

QI

quality improvement