Background Influenza vaccination is the most effective method for preventing influenza virus infection. Adult hospitalised patients form a particularly high-risk group for severe influenza given their advanced age and comorbidities. We sought to improve the influenza vaccination rates of hospitalised patients at the White River Junction Veterans Affairs Medical Center.
Methods The improvement effort started in 2007 when our baseline vaccination rate was about 60%. An interprofessional team analysed the influenza vaccination process for hospitalised patients. During the course of six influenza seasons, eight Plan-Do-Study-Act cycles were used including a hospital-wide flu campaign, embedded orders in the electronic medical record (EMR) to facilitate ordering vaccinations by providers, daily reminders from ward clerks and standing orders for influenza vaccination on discharge. The measure was the monthly percentage of patients discharged from the hospital with an up-to-date influenza vaccination.
Results The percentage of veterans discharged with an up-to-date influenza vaccination increased to over 80% in February 2009 and has remained high.
Conclusions Although we are confident that our local efforts helped to improve the vaccination rate, external factors such as the 2009 H1N1 pandemic and universal vaccination may have primed our system to respond more readily to the implemented changes. Understanding all of the relevant factors that lead to vaccination uptake can be applied to future hospital influenza vaccination campaigns. In addition, our work demonstrates that an interprofessional approach is still required to apply the functionality of the EMR effectively.
- Quality improvement
- Hospital medicine
- Infection control
Statistics from Altmetric.com
Influenza vaccination is a proven method for preventing influenza virus infection and its potentially severe complications. The WHO recommends annual vaccination for nursing home residents, people with chronic medical conditions, elderly individuals and other groups such as pregnant women, healthcare workers and young children.1 Adult hospitalised patients form a particularly high-risk group for severe influenza given their advanced age and comorbidities. At the start of the influenza season in 2007, the performance data at our Veterans Affairs (VA) medical centre showed that about 60% of hospitalised patients were discharged with an up-to-date influenza vaccination. This rate of influenza vaccination coverage was comparable with that which was collected in the 2008 USA National Health Interview Survey, where coverage of high-risk adults was found to be around 40% and that of those aged ≥65 years was around 67%.2 The performance goal at our VA was to discharge 83% of hospitalised patients with an up-to-date vaccination. This is consistent with the US Healthy People 2020 initiative annual influenza vaccination target of 80% for adults and 90% for high-risk adults, including those aged ≥65 years.2 A systematic review that examined the efficacy of interventions applied to patients aged ≥60 years to increase influenza vaccination rates found no randomised controlled trials that involved hospitalised patients.3
In 2007, one of the four internal medicine inpatient resident physician teams at our medical centre was asked by their supervising attending physician to consider the importance of the influenza vaccination for hospitalised patients and to focus on their vaccination performance. During the course of one month, this individual team was able to improve the influenza vaccination rate from 40% to 100% for their patients. The team used their team room whiteboard for data feedback, which also served as a reminder to vaccinate. However, this improvement was isolated to one team, was not sustained and did not spread to the other medical teams.4 The resident on this team became a faculty member at the medical centre in 2008 and was assigned to an interprofessional team to improve this performance measure for all inpatient medical centre services.
The goal of our improvement work was to increase the rate of influenza vaccination for hospitalised patients from about 60% to 90% over successive influenza seasons.
The White River Junction VA Medical Center is a 60-bed teaching hospital serving USA veterans (adults aged ≥18 years) in Vermont, New Hampshire, and Northwestern Massachusetts. There are 42 medicine and surgical beds, 8 intensive care unit beds and 10 mental health beds. The Computerised Patient Record System (CPRS) is used for order entry and all patient data are stored in the VA health record database, the Veterans Integrated Systems Technology Architecture. Resident physicians from Dartmouth-Hitchcock Medical Center rotate through the medicine, surgery and psychiatry services and are responsible for writing most of the hospitalised patient care orders.
One component of CPRS is the automated clinical reminder system. The system provides each healthcare provider with automated reminders to perform health maintenance at the point of care. We focused on the clinical reminder for influenza vaccination. The clinical reminder system for influenza vaccination is activated annually at the start of influenza season for all patients aged ≥18 years. If the clinical reminder for influenza is ‘due’, the reminder is completed by the provider by completing a note that is linked to the reminder. If the patient received the vaccination elsewhere or refuses the vaccination, this must be documented to complete the reminder. For influenza, the note is completed by the nursing staff that actually administers the vaccination. This is to ensure documentation of the anatomical site of the vaccination administration, the dose administered and the lot number for quality control purposes.
The impetus for this work was Veterans Health Administration performance data on influenza vaccination which is collected through the External Peer Review Program (EPRP). The EPRP is a manual patient chart review process that provides medical centres with specific quality of care information. EPRP provides cross-sectional, continuous data for analysis, internal and external comparison and a set of performance goals for each measure.5 In this study, we used the clinical reminder system to generate the vaccination coverage reports and not the EPRP data, to include all patients discharged from our medical centre during influenza season.
Planning the intervention
The improvement effort started with a detailed analysis of the process of influenza vaccination administration to inpatients. The process analysis identified the steps involved in ordering and administering the vaccination as well as several barriers to vaccination.
An interprofessional team was formed, consisting of an inpatient physician, the nursing informatics coordinator, the infection control nurse and the medical centre's clinical applications coordinator. The chart review was performed monthly to identify those patients with outstanding influenza clinical reminders for influenza vaccination on discharge from the hospital and identify barriers to the administration of the influenza vaccination. The barriers were then used as the basis for the interventions for system improvement. We used the model of Langley et al6 for improvement with iterative ‘Plan-Do-Study-Act (PDSA)’ cycles.
During the study period, eight PDSA cycles were carried out (table 1). Each PDSA was preceded by a system analysis that identified a specific barrier that was inhibiting the successful delivery of the vaccination.
To promote influenza vaccination awareness on the wards, the infection control nurse launched an influenza vaccination campaign (PDSA cycle 1). The campaign included posters that were hung throughout the wards and in the resident team rooms. In addition, the infection control nurse made brief announcements at morbidity and mortality and resident report conferences during influenza season. These announcements focused on the central role of the inpatient provider in vaccination of their patients.
The interprofessional team examined provider's conceptions of the performance measure and how to attain that goal by informal questioning of several inpatient providers. It was noted that the discharge instructions template contained misleading text that led the provider to believe they were ordering a vaccination on discharge. This led to several PDSA cycles wherein the discharge templates were modified. During PDSA cycle 2, the influenza vaccination order was added to the discharge order menu. During PDSA cycle 3, the date of the most recent vaccination was pulled into the discharge template. Ultimately, during PDSA cycle 4, the actual vaccination order was programmed to pop up automatically when the discharge order was written.
The interprofessional team examined the influenza vaccination reports on a monthly basis. It was observed that vaccinations were noted to be ‘missed’ by the clinical reminder system when they had actually been given. On chart review, it became apparent that although the provider ordered the vaccination, the administration of the vaccination by nursing staff had been improperly documented and therefore not captured by the clinical reminder system. The documentation system was reviewed by the infection control nurse with the nurse managers. The nurse managers suggested changes to the electronic medical record (EMR; the nursing ‘flu shot’ note template and the daily nursing progress note) to ensure systematic documentation of vaccination. In addition, the nurse managers educated their staff during nursing report on a regular basis during influenza season. Together, these interventions to improve nursing documentation were denoted as PDSA cycle 5.
The initial improvement efforts were leading the medical centre closer to the performance goal. However, the H1N1 (‘swine flu’) epidemic in 2009–2010 added another performance measure in addition to the seasonal influenza vaccination: the H1N1 vaccination. The interprofessional team was asked to ensure that inpatients received both vaccinations before discharge. The team felt that given this complexity, the sustainability of systems previously put into place was at risk. It was then that the ward clerks were asked by the interprofessional team to remind inpatient providers to vaccinate their patients before discharge. The clerks used the clinical reminder system to print out reports of those inpatients who were due for both vaccinations (PDSA cycle 6).
In the following influenza season, 2010–2011, vaccination awareness appeared to be waning, given an observed decrease in the percentage of patients discharged from the hospital with an up-to-date vaccination. The discharge process had been previously examined (PDSA cycles 2–4), and thus attention was turned to the admission process. The date of the most recent vaccination was pulled into the resident physician admission note to help promote provider awareness of the vaccination status of their patients (PDSA cycle 7).
The implementation of the standing order for influenza vaccination to be administered by nursing staff occurred at the medical centre leadership level at the start of influenza season 2012–2013 (PDSA cycle 8). The focus of the performance measure is on nursing staff and no longer on inpatient providers. The medical centre remains at goal for this performance measure, and therefore, there have been no PDSA cycles undertaken to improve the measure since the 2012–2013 influenza season.
Analysis of the intervention
Our main outcome measure was the percentage of patients discharged from the hospital for whom the influenza vaccination was up to date. All eligible patients who were admitted to the medicine, surgical and psychiatry service during each flu season (from 1 October of 1 year to 30 March of the following year) from 2007 to 2013 were included. Eligible patients were defined as those who had not yet received the influenza vaccination and did not have documented contraindications to the vaccination. The percentage of completed influenza vaccinations was extracted electronically from the EMR by reviewing clinical reminder completion rates for influenza vaccination. We used statistical process control to analyse the data and determine the impact of the interventions.7 Individuals and moving range (xMR) charts were used for continuous data such as the number of discharges per month and p-charts were used for the dichotomous data such as the influenza completion rate.7
From October 2007 to November 2010, the volume of discharges from our hospital showed variation that is typical for the process being measured, this is referred to as ‘common cause variation’ (figure 1).7 From December 2010 to November 2012, there are 12 points above the average during influenza vaccination season. This indicates a ‘special cause signal’ which is unlikely to be observed unless there has been a change in the process.7 In this case, there was an increase in our monthly discharges during that timeframe.
The percentage of patients who were discharged with an up-to-date influenza vaccination increased during the 6 years of our work (figure 2A). The initial increase occurred from January 2008 to January 2009 and spanned two influenza seasons. Starting in February 2009, there are eight consecutive points above the average which indicates a special cause variation signal. After continued performance at this level, the control limits were split to reflect the new level of function in the system (figure 2B). This split is at the start of the following influenza season in 2009. Since October 2009, we have observed only common cause variation, with the exception of October 2012 when the vaccination coverage rate decreased to around 75%. October 2012 was the start of the influenza season when standing orders were implemented at our medical centre.
The goal of our improvement work was to identify barriers to the administration of the influenza vaccination, to improve the vaccination rates of hospitalised patients at our VA medical centre and to assess the vaccination rates over time. The barriers included inattention to the vaccination, confusing discharge order templates and sustaining the changes that were implemented. PDSA cycles to address each of these included a hospital-wide campaign to educate staff, patients, families and visitors about the importance of the vaccination; multiple cycles to modify the discharge template orders, including ward clerks to sustain gains; and finally, the implementation of standing orders for influenza vaccination.
The limitations of our improvement work include the fact that the data were collected during six different flu seasons, including the 2009 H1N1 pandemic. The H1N1 pandemic raised national awareness of the importance of influenza vaccination.8 The estimated seasonal influenza vaccination rates for this season were higher overall.9 Moreover, it is well known that contact with healthcare providers increases rate of vaccinations in general.10 Therefore, being hospitalised during the peripandemic period could have further bolstered higher vaccination rates. The turning point that was observed in February 2009 was conceivably because of the adaptations of the discharge instructions template. The template had been misleading providers to believe they were vaccinating their patients on discharge, when in fact they were not. After the performance goal had been met and sustained over a significant amount of time, the control limits were split to reflect a new level of function in the system.
Another major change that occurred during the studied timeframe which could have influenced our results was that in February 2010, the Advisory Committee on Immunization Practices (ACIP) voted to expand the influenza vaccination recommendations to include all adults.11 Before 2010, the influenza vaccination was recommended in persons aged 18–49 years with high-risk conditions and persons ≥50 years of age. Universal vaccination, which would appear to be a simpler protocol to follow, could conceivably lead to higher vaccination rates. However, influenza vaccination coverage statistics from the 2010 to 2011 season reveal similar rates to the season prior.12 Universal recommendation likely contributed to the sustainability of the higher rates of vaccination at our medical centre (see figure 2B).
Overall, influenza vaccination rates in the US have steadily increased during the studied timeframe.13 For persons aged 18–64 years with high-risk conditions, the coverage rate has increased to around 45% and for those aged ≥65 years of age, the rate has remained around 70%.13 These rates are much lower than the Healthy People 2020 target goal for these groups (90%).2 These statistics support the need for ongoing efforts to increase vaccination rates and especially in high-risk groups, such as hospitalised patients.
There have been many publications on increasing vaccination rates; most of these are in primary care but there are some that specifically address inpatients.3 Most of these studies used computer-decision support models to increase vaccination compliance. In these models, standing orders have been shown to be the most effective.14 The ACIP has long encouraged standing orders to be used in inpatient settings.15 When our improvement work started, standing orders were not the policy at our institution. This required a cultural change that took place with the implementation of new nursing leadership in 2011. During the 2012–2013 influenza season at our institution, influenza vaccinations became standing orders to be given by nursing unless contraindicated by the physician. The implementation of standing orders for vaccination led to a decrease in vaccination coverage rates. With standing orders, the interventions during PDSA cycles 2, 3, 4 and 7 became obsolete because these were provider dependent, not nursing dependent. The special cause signals that we observed after PDSA 6, led to a new level of function in the system which is demonstrated in the split p-chart (figure 2B). During subsequent influenza seasons, what we learnt during PDSA 6 (to include the ward clerks in our approach) could be aimed at nursing staff.
Unlike other studies reported in the literature,16 the foundation for this improvement work was an already fully functional EMR. From 2008 to 2013, the percentage of US hospitals using EMRs increased from 9% to 80%.17 This makes our work applicable to most hospitals today. Our work demonstrates that an interprofessional approach is still required to apply standing orders and the functionality of the EMR effectively.
In conclusion, external factors such as the H1N1 pandemic and universal vaccination may have primed our system to respond more readily to the implemented changes. The impact of external factors has been described before in other improvement campaigns such as hand hygiene in hospitals.18 Understanding all of the relevant factors that lead to vaccination uptake can be applied to future hospital influenza vaccination campaigns. We have sustained high vaccination rates for our hospitalised patients over six consecutive influenza seasons, which may indicate that the systems we implemented are adaptable to ongoing changes in local and external factors.
This work was supported with the use of materials and facilities from the White River Junction VA in White River Junction, VT. The views expressed in this manuscript are those of the authors and do not represent the views or policy of the Department of Veterans Affairs or the US Government.
Contributors Each of the authors contributed to the conception and design of the work, in addition to the acquisition, analysis and interpretation of data. Each of the authors revised the manuscript critically for important intellectual context. Each of the authors gave final approval of the version that has been submitted. Each of the authors agrees to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. In addition to meeting the requirements for authorship, each author had a specific focus area during the work. The authors’ initials with each of their focus areas is listed below: EC planned, conducted and reported the work. GO supervised the planning, conduct and reporting of the work. TT supervised the conduct and reporting of the work. CB adapted the functionality of the electronic medical record as needed for the improvements and collected the data. JG supervised the overall planning, conduct and reporting of the work.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.