ABSTRACT
Purpose: To measure the impact of physician engagement during the formulary decision-making (FDM) process on formulary adherence.
Methods: A retrospective review of formulary decisions made at an academic medical center during a three-year period was conducted. Decisions that could be evaluated for formulary adherence were assessed for whether pharmacists engaged the requesting physician during the FDM process, based on email contact or P&T committee meeting attendance. Comparison of formulary adherence, specifically medication utilization compliant with formulary restrictions, was made between decisions involving and not involving physician engagement. Nonformulary use was quantified for both groups. The relationship of factors and their potential influence on formulary adherence was examined. Acceptance by the P&T committee of the initial recommendation brought forward by a pharmacist-led FDM group was compared between engaged and unengaged cohorts.
Results: Of 139 decisions made effective during the evaluation period, 64 could be evaluated for adherence. Following exclusion of 22 nonformulary decisions and 13 decisions for low utilization, overall formulary adherence was 94%. Cumulative formulary adherence in the engaged and unengaged cohorts was 96% (n = 3,208 of 3,342) and 90% (n = 1,360 of 1,506), respectively. When compared based on individual decisions, no difference between adherence rates was detected and adherence between cohorts was not affected by assessed covariates. Initial formulary recommendations in the sample were more likely to be modified by the P&T committee if the physician was engaged (20% versus 11%, P = 0.48).
Conclusion: Physician engagement during FDM did not significantly affect formulary adherence at an academic medical center.
Keywords: formularies, hospital; pharmacy and therapeutics committee; interprofessional relations; guideline adherence; professional education
BACKGROUND
Formulary management in a hospital setting is a complicated process that presents many challenges in the modern era.1 Given that there are more than 1,500 new molecular entities2 on the market and upwards of 45 new medications approved each year,3 expertise in evidence-based medicine, an understanding of medication safety, and knowledge of pharmacoeconomic principles are required.
When considering a medication for addition to formulary, experts typically take into account clinical trial data detailing efficacy and safety, the potential advantages and disadvantages, the anticipated place in therapy, and the financial impact of the decision.4–6 These considerations align directly with the value of a formulary to an institution, as formularies exist to improve evidence-based decision-making, protect patients, and contain costs. Decisions, typically made by a multidisciplinary P&T committee, are given thorough consideration and based on the best efforts of the individuals to make the recommendation that is most beneficial for the patients, prescribers, and institution.4,5
During this process, there is an opportunity for pharmacists responsible for making formulary recommendations to work with physicians and gain their opinions on the evaluated therapeutic option;4 however, the actual utilization of this process is unknown.7 A dearth of literature exists on how physician engagement impacts the formulary process. One institution has shown that prospective academic detailing of prescribers, in comparison with nothing, led to a nonsignificant increase in use of formulary options (i.e., formulary adherence); however, authors suggested that the high formulary adherence at baseline likely mitigated the lack of statistical significance between intervention and control in this study.8 Another investigation suggested that group education, provided by government-sponsored experts, may have a positive impact on formulary adherence in a country (Sweden) with a national formulary.9
The initial objective of this research was to determine a rate of engagement with physicians requesting formulary medications at an academic medical center. Once determined, a comparison of formulary adherence was made between agents that had engagement by the requesting physician during formulary decision-making (FDM) and those that did not in order to identify whether engaging physicians during the FDM process increased adherence to the formulary.
METHODS
A retrospective data and chart review was conducted examining formulary decisions at the Medical University of South Carolina, an 800-bed academic medical center, from July 1, 2014, through June 30, 2017. The list of formulary changes and decisions was collected from staff education and formulary request databases maintained by the institution’s pharmacist-led FDM group. All decisions were further organized by formulary status, specifically formulary restricted, change in formulary restriction, or nonformulary. From this initial list, decisions were excluded if adherence could not be evaluated. These excluded decisions comprised of: 1) medications approved for formulary addition with no restrictions; 2) restriction-change requests for formulary-restricted medications when the change in restriction was not approved or the restriction was removed; 3) requests initiated internally by the FDM group without an identified physician champion; 4) temporary changes; 5) clarification of wording associated with restrictions; 6) decisions with a subsequent formulary change in the next six months; 7) medications not added to the formulary but with restricted purchasing requirements due to increased scrutiny related to adherence; and 8) medications with subjective restriction criteria (e.g., indication-based restrictions, failure of first-line therapies, patients unable to tolerate alternatives) making adherence indeterminable through data abstraction or chart review.
Physician engagement during the formulary process (Figure 1) was collected from P&T committee minutes and FDM team-member emails and recollection. Any mention of the requester speaking in the minutes was counted as attendance. A return email correspondence from the requester or confirmation of plan to attend the P&T meeting was documented as engagement. Decisions when requesters either were not contacted by the pharmacy liaison or did not return an introductory email were considered not to be engaged. This analysis formed two cohorts of decisions organized by whether physicians were engaged or unengaged. Service line of the requesting physician was also documented during this evaluation.
Figure 1
The primary endpoint for the comparison of physician-engaged versus physician-unengaged decisions was adherence rate, calculated as compliance with established formulary restrictions, specifically both new and changed restrictions. Utilization, as determined by medication charges collected from Epic prescribing data and chart review, was evaluated during the six months following the effective date of the formulary decision due to the likelihood that physician engagement during the FDM process has a limited time of impact. Medications with less than 10 total charges within the six months following the decision were excluded from this comparison. Formulary adherence for formulary-restricted medications was defined as the number of charges per the established restriction divided by the total number of charges for the drug. For instances when changes were made to an existing restriction, adherence rate was determined as the total number of charges conforming with the updated restriction divided by the total number of charges for the drug. These were analyzed from a cumulative perspective and as individual formulary decisions.
As a secondary endpoint, nonformulary orders were counted due to the lack of an acceptable denominator to establish adherence. In a secondary analysis, the relationship of the number of charges, number of restrictions, number of prescribers, orally or parenterally administered medication, restriction by outpatient setting, and adherence rate was evaluated. In order to determine the impact physician engagement might have on P&T committee acceptance of the initial recommendation made by the FDM group, the rates of initial recommendation acceptance versus final decision modification were compared between the engaged and unengaged cohorts. This study was deemed quality improvement by the institutional review board (IRB); thus, it did not require IRB approval.
Analysis
Descriptive statistics are presented as median (25th, 75th percentiles) and N (%) for continuous and categorical variables, respectively. To compare continuous and categorical variables with physician engagement, Wilcoxon rank-sum tests and Fisher’s exact tests were used. To assess the relationships between various factors, including number of charges, number of restrictions, number of prescribers, orally administered medication, restriction by outpatient setting, and adherence rate, linear models were constructed controlling for physician engagement. All factors associated with adherence rate were then entered into a full model to make final assessments. Statistical significance was assessed at α = 0.05. All statistical analyses were performed using SAS v9.4.
RESULTS
The total number of formulary decisions with an effective date between July 1, 2014, and June 30, 2017, was 139; Table 1 describes each formulary designation of these decisions. Figure 2 provides a flowchart of formulary decisions and application of exclusion criteria. The reasons for applying the first level of exclusion criteria are provided in Table 2; these caused 75 decisions to be excluded. Of the remaining 64 decisions, 28 were added to the formulary as restricted medications, 14 were changes to current formulary restrictions, and 22 were designated nonformulary. Of the nonformulary-designated medications, there were 166 charges for the 16 medications in the engaged cohort and zero charges for the six medications in the unengaged cohort.
| Table 1 Formulary Decisions by Category (n = 139) | |
|---|---|
| Formulary Decision | Number (%) |
| Added | 20 (14) |
| Added with restriction | 39 (28) |
| Change in restriction | 50 (36) |
| Not added (i.e., nonformulary) | 23 (17) |
| Not added with restriction | 5 (4) |
| Temporary additions | 2 (1) |
Figure 2
| Table 2 Formulary Decisions Excluded From Analysis (n = 75) | ||
|---|---|---|
| Reason for Exclusion | Number | Examples |
| Internal initiation | 22 | Dexmedetomidine (Precedex) Linezolid (Zyvox) |
| Added with no restrictions | 20 | Benzonatate (Tessalon) Pantoprazole (Protonix) |
| Clarification of restriction | 12 | Micafungin (Mycamine) Vancomycin (Vancocin) |
| Subjective restriction criteria | 9 | Palivizumab (Synagis) Palonosetron (Aloxi) |
| Nonformulary restricted designation | 5 | Alemtuzumab (Lemtrada) Blinatumomab (Blincyto) |
| Restrictions removed/change in restriction not approved | 3 | Lacosamide (Vimpat) Prothrombin concentrate (Kcentra) |
| Temporary changes | 2 |
Ceftazadime (Fortaz) Isradipine (Dynacirc) internal off-label assessment |
| Subsequent formulary decision within 6 months | 2 |
Alvimopan (Entereg) Dofetilide (Tikosyn) |
Requesting physicians were engaged during 70% (n = 45 of 64) of all decisions, which included 20 of the 29 (69%) formulary-restricted additions, nine of the 12 (75%) change-in-restriction decisions, and 16 of the 22 (73%) nonformulary decisions. Requesting physician service line for these 64 requests is presented in Table 3.
| Table 3 Requesting Physician Service Line (n = 64) | |
|---|---|
| Service Line | Number (%) |
| Hematology/oncology | 21 (33) |
| Infectious diseases | 9 (14) |
| Pulmonary, critical care, allergy, and sleep medicine | 8 (12) |
| Pediatrics–critical care | 3 (5) |
| Cardiology | 2 (3) |
| Neurosurgery | 2 (3) |
| Obstetrics-gynecology | 2 (3) |
| Otolaryngology–head and neck surgery | 2 (3) |
| Pediatrics–hematology/oncology | 2 (3) |
| Psychiatry | 2 (3) |
| Radiology | 2 (3) |
| Urology | 2 (3) |
| Anesthesia and perioperative medicine | 1 (2) |
| Dermatology | 1 (2) |
| Hepatology | 1 (2) |
| Nephrology | 1 (2) |
| Neurology | 1 (2) |
| Pediatric cardiology | 1 (2) |
| Rheumatology and immunology | 1 (2) |
In the 42 decisions related to formulary restrictions, cumulative adherence was 94% (n = 4,592 of 4,880). Excluding the 13 medications with less than 10 charges, adherence was 94% (n = 4,568 of 4,848). Comparing the engaged and unengaged cohorts for the ultimately included 29 formulary-restricted medications, cumulative formulary adherence rates were 96% (n = 3,208 of 3,342) and 90% (n = 1,360 of 1506), respectively. No statistically significant difference between rates of adherence for individual decisions was detected (Table 4).
| Table 4 Descriptive Statistics of Formulary Medications | |||
|---|---|---|---|
| Variable |
Not Engaged (n = 12) |
Engaged (n = 17) |
P |
| Percent adherent* | 98.3 (93.1, 100.0) | 92.9 (86.7, 100.0) | 0.27 |
| Number of prescriptions | 64.5 (28, 112.5) | 46.5 (20.0, 118.0) | 0.55 |
| Number of ordering providers | 4 (3, 16.5) | 15 (13, 35) | 0.09 |
| Number of restriction criteria | 0.39 | ||
| 1 | 3 (25%) | 3 (18%) | |
| 2 | 4 (33%) | 10 (59%) | |
| ≥ 3 | 5 (42%) | 4 (23%) | |
| Service restriction | 10 (83%) | 16 (94%) | 0.55 |
| Outpatient restriction | 8 (67%) | 5 (29%) | 0.07 |
| Oral | 10 (83%) | 9 (53%) | 0.13 |
| *Percent adherent is presented as median + interquartile range compared using Wilcoxon rank-sum test and categorical variables are compared using Fisher’s exact test. | |||
Assessing the impact of various factors on adherence rates for engaged and unengaged decisions, no differences in the total number of orders, number of restriction criteria, service restrictions, whether the medication was orally or parenterally administered, and number of prescribers were detected (Table 4). Though not statistically significant, the relative frequency of medications with outpatient restrictions was 29% for the engaged group and 67% for the not-engaged group. The linear regression analysis did not identify any factors associated with adherence rate.
Initial recommendations made by the pharmacist-led FDM group to the P&T committee were accepted in 53 of 64 (83%) decisions evaluated. For all decisions when the physician was engaged in the process, the initial recommendation was accepted 80% (n = 36 of 45) of the time; when the physician was not engaged, the initial recommendation was accepted at a higher rate (n = 17 of 19, 89%) that was not significant (P = 0.48). Further organized by type of decision, drugs with consistent nonformulary and formulary-restricted recommendations and designations involved physician engagement 75% (n = 15 of 20) and 66% (n = 21 of 33) of the time, respectively. For medications when recommendations were changed, the rates of physician engagement were 50% (n = 1 of 2) for nonformulary-designated drugs and 89% (n = 8 of 9) for formulary-restricted drugs.
DISCUSSION
Hospital formulary management has become more and more complicated as more medications are approved, billing and reimbursement procedures change, direct-to-consumer advertising boosts prescriptions,10 and safety and regulatory restrictions rise.1 This investigation, seemingly the first study to examine an association between physician engagement with FDM on subsequent medication utilization, attempted to determine whether working with physicians during FDM improved formulary adherence. While the results were not significant, there was a higher cumulative rate of adherence associated with formulary-restricted medication adherence when physicians were engaged in the decision-making process. Conversely, when reviewing adherence per medication by considering the median interquartile range of adherence proportions, physician engagement was less likely to indicate adherence, demonstrating that it is still unclear as to whether physician engagement during decision-making enhances formulary adherence.
Furthermore, there was more nonformulary use of medications when physicians were engaged in the process and no nonformulary use when physicians were not. These findings may be the result of the specific philosophy of the study institution that formulary status is dictated by available evidence (e.g., efficacy, safety, cost), with the understanding that nonformulary use may be necessary in a small number of patients. Working with physicians during the process helps get the nonformulary recommendation of certain medications confirmed while still giving the requesting physicians the comfort that they will be able to prescribe a nonformulary drug if they feel it is needed in a specific patient. It is also possible that physicians more engaged in the process felt strongly that the requested medication was needed and chose to prescribe it via a nonformulary order despite the decision going against their wishes. The fact that nonformulary use was not seen when unengaged physicians requested medications could indicate that these requesters were less passionate about the need for these medications for their patients and may have submitted a request for other reasons (e.g., pharmaceutical marketing requests). It is also possible that the unengaged requesters felt the formulary request process provided little value and therefore chose not to engage; however, this would not explain the lack of nonformulary prescribing of these medications.
The data suggest that working with a physician during the FDM process may cause a recommendation made by the FDM group to be changed by the P&T committee. This may seem counterintuitive, in that one may assume that the decision would be fully elucidated prior to the formal P&T committee presentation and that both the requesting physician and the decision-making group would be in concordance prior to the meeting. However, in this assessment, both returned email and P&T committee meeting attendance were defined as engagement, meaning that some initial recommendations may have been made without discussion between decision-maker and physician prior to the meeting, at which point the requesting physician attending the meeting could have swayed the P&T committee to his or her point of view and changed the final decision from the initial recommendation.
The only other identified evaluation of physician engagement and the impact on formulary adherence showed that educational meetings with physicians related to hospital formularies improved prescribing adherence in a cluster randomized controlled trial.8 Investigators organized 12 hospital wards into three groups: 1) control; 2) basic intervention; and 3) extended intervention. The four basic-intervention and four extended-intervention hospital wards were used as a venue for educational meetings involving prescribing physicians. These meetings described P&T committees, formularies, patient safety, and tools available to help physicians choose medications. Physicians practicing on the four extended-intervention wards received additional group detailing on drug utilization and instructions for how to improve adherence to the formulary at both two and four months after the initial education session. Results demonstrated that neither the basic nor the extended interventions significantly improved formulary drug use in comparison with the control group. Given the use of a randomized control group, this study had a more robust study design than the current report; however, the wards were fairly different at baseline despite randomization. Investigators concluded that the academic detailing strategy did not improve formulary adherence, attributing the lack of difference to high baseline formulary prescribing.
The current report was retrospective in nature, thus randomization was not an option. Additionally, the possibility for confounding exists in the current report, as medications were not organized by class, anticipated utilization, or cost prior to the unintentional randomization. To combat this, other identified covariates were evaluated for impact, but again, no significance was identified. This study was not sufficiently powered to detect significant differences, making the results of this study inconclusive; it cannot be definitively stated that engaging physicians during the decision-making process impacts formulary adherence, either positively or negatively. Finally, the single-center nature of the study means that the results are dependent upon the practices at this institution, and it may not be possible to extrapolate these findings to other organizations.
Further limitations of this study relate to the process that was used to determine physician engagement. At no point during the study were physicians asked about their attitudes toward the FDM process or the factors that influenced whether they chose to participate. Furthermore, the determination of physician engagement used in this study is a potentially weak indicator of this occurrence, particularly when combined with the occasions when pharmacy liaisons failed to initially email the requesting physician. Establishing reasons that physicians choose to engage in the formulary process and identifying incentives that encourage further engagement are important future areas of investigation. Initially collecting information from physicians on their opinions and beliefs about the formulary process in the hospital setting would help inform these future studies. Rather than speculating on motivations, developing a method for ensuring that all physicians have the opportunity to participate in the FDM process and allowing them to provide opinions about the formulary and formulary process and motivations for or against participating would yield valuable information for health care professionals responsible for making formulary decisions. Considering an educational initiative that would provide physicians with the reasons formularies exist, how decisions are made, and how their input is crucial would also be a useful initiative to undertake in order to improve physician engagement with the FDM process.
Because of these limitations and the lack of research on this important topic, a more thorough investigation of physician engagement during the FDM process in a prospective, randomized study would prove valuable to investigate more definitive conclusions. Regardless, it is important to encourage physicians to understand and engage in the formulary process, and formulary decision-makers should attempt to work with their colleagues when forming recommendations to the P&T committee. While the effect of direct physician engagement on formulary adherence is unclear, utilization of information technologies at the point of prescribing has been shown to reduce nonformulary use when automatic therapeutic substitution protocols have been presented to prescribing physicians.11 Ensuring that the formulary status of medications in computerized prescriber order-entry systems is accurate and using these systems to display formulary options and reasons for their inclusion may also be valuable in the pursuit of high formulary adherence.
CONCLUSION
Formulary adherence does not appear to be affected directly by physician engagement during the FDM process. Despite the unclear association, these results do not absolve formulary decision-makers from contacting physicians during the formulary request process. Cross-sectional studies to understand physicians’ attitudes toward FDM and sufficiently powered investigations to determine the value of physician–pharmacist collaboration during the formulary request and recommendation process should be conducted.
