Abstract
Introduction
Patients nonadherent with oral antipsychotics are frequently readmitted to inpatient psychiatry. Reducing readmissions through use of long-acting injectable (LAI) medications can help lower costs and improve patient care. Evidence-based research supports the use of LAIs to decrease admissions/remissions, that patients can be identified via electronic medical records (EMRs), and that nursing’s collaboration in medical decision making can improve care (Andrews et al., 2017; Merrick, Fry, & Duffield, 2014). This quality improvement project offered in-service education to nursing staff on an inpatient psychiatric unit to help increase nurse advocacy and prescriber collaboration for frequently readmitted patients. Chart auditing was completed for the 90 days before in-service teaching to measure the scope of the problem and 90 days post in-service to measure effect. This project aimed to reduce frequent readmissions by 25% and increase LAI prescriptions by 7%. The clinical site was a teaching hospital located in Chicago, Illinois.
Problem and Significance
Nonadherence with medications is a key factor for frequent hospitalizations of patients with psychiatric disorders and costly to institutions (Andrews et al., 2017). The Centers for Medicare & Medicaid Services (CMS) define readmission as an unplanned return to an acute-care facility or hospital in under 30 days with the same index diagnosis (Boccuti & Casillas, 2017). Patients readmitted to a hospital with the same diagnosis in less than 30 days will prohibit the admitting institution from billing CMS for services rendered. Patients who require readmission after recent discharge constitute lost revenue for the hospital and demonstrate the need to improve care.
A cost analysis published by the Agency for Healthcare Quality and Research (Hines, Barrett, Jiang, & Steiner, 2014) found in 2011 that for Medicaid patients alone the highest 30-day readmission rates of all causes, mood disorders (41,600) and schizophrenia (35,800) were the two most frequent, with diabetes (23,700) in third. In another statistical brief by the Healthcare Cost and Utilization Project, Heslin and Weiss (2015), found that mental health disorders require frequent hospitalization and that the problem has increased between 2003 and 2011. Also, in 2012, of the approximately 383,000 hospitalizations for schizophrenia, 9% of them were readmitted within 30 days.
Available Knowledge
For identifying individuals at the clinical site, evidence supports recognition of nonadherence with medication through the EMR. Andrews et al. (2017) conducted a retrospective study of patients with psychiatric disorders in outpatient clinics. They discovered that the only variable that increased relative risk of readmission (adjusted relative risk = 7.19, p < .001) was medication nonadherence. They concluded that recording medication compliance in the EMR is important for identifying at-risk patients. Electronic tracking was echoed by Roque, Findlay, Okoli, and El-Mallakh (2017), who also reported that patients with chronic mental health issues need clinical screening for risk of readmission to reduce 30-day readmission rates.
Tiihonen et al. (2017) studied the effectiveness of various antipsychotic formulations (i.e., LAIs vs. oral medications). Through prospective analysis of the Swedish national databases of schizophrenic patients, those considered to have failed treatment (i.e., readmission, suicide attempt, discontinued or altered medication regimen, and death) were studied. Of the 29,823 patients, 71% had treatment failure and 43% were readmitted to the hospital. The lowest rate of treatment failure was from those on LAIs, whereas those on oral medications were considered to have the highest risk (an increase of 20% to 30%). In the United States, this is mirrored by MacEwan et al. (2016), who monitored readmission rates of patients on LAIs; 15,556 patients in multistate Medicaid databases were included. The use of LAIs showed a significantly reduced probability of readmission versus patients taking oral antipsychotics. They estimated that patients on LAIs have a 5% reduction in the absolute probability of readmission.
Researchers have also identified knowledge gaps. Barrio et al. (2013) reported that LAIs might improve adherence to medications, reduce readmissions, and improve psychosocial functioning. They studied clinical remissions between LAIs and oral medications. This was a case-control study (n = 26) but failed to demonstrate significance of LAI use over oral medications on readmission rates, and that efficacy of LAI treatment requires further research. Similarly, Kane, Kishimoto, and Correll (2013), in a review of meta-analyses, reported that randomized control trials might not be the best form of testing as they tend to increase adherence by their nature and reduce accurate adherence estimation. Their main argument against effectiveness on specific formulations of antipsychotics was that study designs lead to skewed results. Where LAIs and oral antipsychotics have similar adherence rates in random control trials, mirror-image and large cohort studies favor LAI use. It should be noted that while failures of significance and potentially flawed study designs have been identified, the use of LAIs is still supported in these studies.
While the information favors LAI use even in skeptical research, how that information is used by nursing requires better application. According to Merrick et al. (2014), nursing’s influence on medical decision making through collaboration is required to meet rising health care demands. Collaboration through interprofessional relationships is considered to improve medical decision making, and trusting relationships are important for nurses while they advocate for best practices of their patients (Merrick et al., 2014).
Collaboration and best practices were pillars of this initiative, and while the intervention was not credentialed continuing education, according to Dickerson (2016), the American Nurses Credentialing Center’s outline for educating staff should include identifying practice gaps and improve care through evidence. Nurses have a voice that should be represented in patient care decisions, and a continuing education framework should be considered when delivering information to staff nurses. McKnight (2013) furthers that professional practice should be guided by continuing education that meets the needs of the nurse and allows them to impart this into best practices.
The practice implications demonstrated by Tiihonen et al. (2017) and Camacho, Ng, Galangue, and Feifel (2008), for the use of LAIs to reduce 30-day readmission rates and increase adherence should be considered when making medical decisions by reflecting the research data. Andrews et al. (2017) also demonstrated the importance of electronically tracking at-risk individuals. The research presented provided evidence for using this platform to use LAIs at the clinical site in frequently readmitted patients. While some studies reported mixed results on the effectiveness of LAIs over oral antipsychotics, Barrio et al. (2013) and Kane et al. (2013) reported better results with LAIs, though not always statistically significant (Barrio et al., 2013). When compared with the financial disease burden, adapting LAI use in this population would likely benefit patients and the clinical site. This is supported in a study by Phan and VandenBerg (2012), wherein an outpatient LAI clinic was able to avoid up to $9000.00 of losses from non-reimbursable readmission charges per patient enrolled.
Education and nurse advocacy for patients should also ally with current best practices and the educational needs of nursing staff (Dickerson, 2016; McKnight, 2013). The evidence supports identifying individuals who are frequently readmitted to inpatient psychiatry through the EMR and initiate them on LAI therapy. Improving quality of patient care based on current evidence is the focus and should be followed.
Rationale
For this initiative, the transitional care model (Hirschman, Shaid, McCauley, Pauly, & Naylor, 2015) was applied. This is a nurse-led, collaborative approach; screening, collaborating, and engaging patients and caregivers are primary concepts of this model. The theory underpinning this initiative is that nursing has a voice that can be used for medical decision making (Merrick et al., 2014). This model was originally designed for elderly patients who have chronic conditions and are at risk for frequent readmission (Hirschman et al., 2015). The transitional care model was designed at the University of Pennsylvania and has been refined through scholarly rigor for two decades (Hirschman et al., 2015). An interdisciplinary staff, under coordination by an advanced practice registered nurse has been shown under this model to improve patient outcomes and satisfaction, while reducing both cost and readmission rates.
The nine core components of the transitional care model are screening, staffing, maintaining relationships, engaging patients and caregivers, assessing/managing risks and symptoms, educating/promoting self-management, collaborating, promoting continuity, and fostering coordination. The relative components of the model used include screening, maintaining relationships, engaging patients and caregivers, assessing/managing risks and symptoms, and collaborating. This model is designed for patients with chronic conditions that lead to frequent admissions and has had a thorough vetting process which demonstrates a sufficient evidence base to improve care in the target population. This model is supported by Hanrahan, Solomon, and Hurford (2014), through application on an inpatient psychiatric unit, which yielded positive results.
Specific Aims
The objectives of this project were focused on identifying frequently admitted patients’ nonadherence with oral antipsychotic medications and improving nursing advocacy by making information available through graphical statistical data. As nurse advocacy is only one dimension of medical decision making, a modest reduction was expected. The benchmark was a 25% reduction in frequent readmissions and a 7% increase in LAI prescriptions.
Design and Methods
Chart reviews for frequently readmitted patients were completed pre- and post-LAI in-service education to nurses on an inpatient psychiatric unit. A final chart audit of identified patients was conducted from the associated clinical site outpatient center. These first two audits were considered a measure of admission rates for oral antipsychotic nonadherence. The incorporation of hospital data with LAI in-service teaching was provided to nursing staff via a PowerPoint presentation. This project had two major sections, one was data collection through chart auditing, the other was nurse in-service education with a following questionnaire on willingness to suggest LAIs for nonadherent patients. In-service education to nursing staff was voluntary as was participation with the questionnaire. At no time did the project director interact with patients; The University of Illinois at Chicago, Office for the Protection of Research Subjects (OPRS) also determined that this activity did not meet the definition of human subject research as defined by 45 CFR 46.102(f), (Protocol # 2017-1206). Specifically, the OPRS determined that voluntary nursing in-service education and a subsequent questionnaire did not meet the criteria for human subjects.
Sample and Setting
The project site was located at an urban teaching hospital in Chicago, Illinois, inpatient psychiatry, for both the chart auditing and in-service teaching. Inclusion criteria included patients with frequent readmissions, a schizophrenia spectrum diagnosis (ICD-10 code F20-29.XX), and nonadherence with oral antipsychotics. Given that this was a quality improvement project, deidentified variables such as age and sex were not coded.
Schizophrenic spectrum patients were considered to have frequent readmissions if they had two admissions with the same admitting diagnosis within a 30-day window. Patients with frequent readmissions and nonadherent with off-label oral antipsychotics (e.g., bipolar with psychotic features) were excluded as they may not be appropriate for LAIs.
Method
The EMR used by the clinical site gathers data on admission dates, admitting diagnoses, adherence, and medications prescribed. The first chart audit spanned the 90 days prior to staff in-service education (August 15, 2017 to November 15, 2017) and included all admissions. Patients admitted and meeting the inclusion criteria had their deidentified information included in the results. Aggregated data were presented with the institutional financial burden, and generalized patient prognoses when nonadherent, at the in-service teaching for nurses. As there are multiple medications and formulations, information on mechanisms of action and pharmacokinetics was kept to a minimum to maintain brevity. A questionnaire was then provided to staff post in-service to assess willingness to suggest LAIs. Participation by nursing staff with both in-service education and the subsequent questionnaire was voluntary.
Ninety days post in-service (December 7, 2017 to March 7, 2018), admission rates using the same criteria above was revisited to measure the effect. The second chart review 90 days post in-service was conducted to assess admission rates using the same exclusion/inclusion criteria. The resulting data from the second chart audit was presented to administration and providers as well. The final chart audit was conducted to find which of the frequently admitted patients had been transitioned to LAIs at the outpatient clinic.
Results
The first chart audit revealed 27 patients meeting criteria, with 60 total admissions over the 90 days, 32 of which were a readmission in under 30 days. The questionnaire provided to nurses allowed for perceived barriers to be expressed. Via survey results barriers reported by nursing post in-service included the cost of LAIs, MD pushback from RN collaboration, and homelessness/social factors. Ninety days post RN education, 17 patients met criteria and amassed 39 admissions, 17 of which were in under 30 days.
Over the entire 180 days, there were 49 frequent readmissions, from 44 patients, totaling 253 inpatient days which would be considered non-reimbursable to CMS. The range of admission days was 1 to 15, and the average length of stay was 5.2 days. The host site reported that total cost of having a patient admitted to psychiatry was $3200.00/day, constituting a potential revenue loss of $1.6 million/year. Results from the initial two audits were presented to providers during their monthly staff meeting.
Chart auditing from the outpatient clinic revealed that 4 of the 44 patients had been started on LAIs post in-service (an increase of 9%) and overall frequent admissions by the same criteria above were reduced by 53% post in-service.
Discussion
During the needs assessment, unit administration reported seasonal variation in admission rates given that admissions tend to increase during winter months. The reduction in admissions for this specific population coincides both with what became a mild winter and in-service education. Because of this, the major reduction in admissions was not attributed to the intervention alone. Also, homelessness/social factors reported by staff as contributing aspects makes the reduction in admission rates challenging to qualify.
The increase of LAI prescriptions found in the final chart audit can be attributed to the intervention. Nursing staff verbalized pushback from MDs accepting RN suggestions for medications. While the interactions between RNs and MDs were reported as having a negative connotation, they did collaborate and nursing did advocate successfully for their patients. While the outcome of a 9% increase in LAI prescriptions surpassed the benchmark of 7%, this still seemed low compared with the cost of readmission. Therefore, an additional presentation was delivered to the site psychiatric providers (APRNs/MDs). Providers acknowledged that LAIs need to be used with more frequency. The Department of Psychiatry was unaware that their readmission rate was as high as reported for schizophrenia alone. Also, at the provider staff meeting, discussions regarding how to integrate LAIs into care with regard to cost, and how to include social work, nursing, and ancillary staff was also considered.
Conclusions
Patients nonadherent to oral antipsychotics are frequently readmitted to inpatient psychiatry. Reducing remissions through use of LAIs can help reduce costs and improve patient care. Evidence within the literature supports the use of EMR to identify patients who may benefit from LAIs, that use of LAIs reduce admissions/remissions, and that nursing’s collaboration as part of medical decision making can improve care.
At completion of this quality improvement project, the admission rate had reduced by 53% and LAI prescriptions had gone up 9%. The sharp drop in admissions, however, was not wholly attributable to this intervention given seasonal variation in admission rates and an unusually mild winter. This improvement project supported nurse advocacy and collaboration with providers. This quality initiative highlights the need to audit charts for patterns of frequent readmissions and an increase nursing advocacy through evidence-based practice education. Despite the lack of a direct correlation between nursing education and an increase with LAI use, a glaring need for improved practice was identified through a focused audit alone. Though, as advocacy remains a fundamental tenet of nursing, analyzing pertinent data for improved outcomes falls within that practice of support. The next steps would involve collaboration with social work and ancillary medical staff in an effort to assure improved patient outcomes and satisfaction, while reducing both cost and readmission rates.
Footnotes
Author Roles
Bernard Vonderhaar was the project director, primary author, and researcher. Marsha Snyder was the clinical mentor, provided expertise with regard to design and application, and reviewed/edited the final manuscript for publication.
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: The primary author is an employee of the health system, but not of the host site and was not compensated for time. No grants were given for this initiative.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
