Authors

  1. Swedlund, Matthew MD, MBA
  2. Kamnetz, Sandra MD
  3. Birstler, Jen MSc
  4. Trowbridge, Elizabeth MD
  5. Arndt, Brian MD
  6. Micek, Mark MD
  7. Lochner, Jennifer MD
  8. Pandhi, Nancy MD, PhD, MPH

Abstract

With a goal of improving efficiency and reducing workload outside of visits, we sought to examine a primary care redesign process aimed at reducing refill requests made outside of office visits. Data on the number of refill encounters per panel member were collected at 17 clinics before, during, and after the implementation of a redesign process. There was an initial reduction in the number of medication refill encounters, and the rate of refill encounters continued to decline following implementation. Variation across clinic contexts suggests that redesign processes may need to be tailored for different settings to optimize effectiveness.

 

Article Content

HEALTH CARE ORGANIZATIONS are increasingly recognizing the need to change processes of care to meet evolving patient needs and demands. Care is increasingly delivered through phone calls, patient portals, and virtual visits rather than office visits (Goldzweig et al., 2013; McDonald et al., 2014; Murphy et al., 2012; Sinsky et al., 2016; Zheng et al., 2015). The management of medication refills outside of office visits represents significant work (Shipman & Sinsky, 2013; Sinsky & Sinsky, 2012), typically occurring through telephone calls, patient portal messages, or fax communications requiring medical record review and reconciliation. Multiple steps are necessary to complete every refill encounter, and these clerical tasks performed within the electronic health record (EHR) system impact the number of patients that physicians and their clinical support staff can care for in a day, thus reducing office visit access (Bae & Encinosa, 2016; Cheriff et al., 2010; Fleming et al., 2014; Lam et al., 2016; Samaan et al., 2009). Given the impact of asynchronous work on burnout and provider joy in practice, efforts to reduce this type of work are warranted to support the quadruple aim (Bodenheimer & Sinsky, 2014): enhancing patient experience, improving population health, reducing costs, and improving the work life of health care providers. Previous studies have analyzed medication refill workflows in pharmacist-based refill clinics (Cassidy et al., 1996; McKinnon & Jorgenson, 2009; Riege, 2005) but have not evaluated workflow changes within primary care clinics.

 

To address many of these changing demands of health care, UW Health began a primary care redesign (PCR) process in 2013. Under a team-based care model, care team members including medical assistants and registered nurses deliver increasingly complex care when trained and empowered to do so with clear patient care policies and protocols. A key component of the redesign for in-office care included a change to the medical assistant rooming workflow. In this new workflow, medical assistants perform a comprehensive medication reconciliation and pend for clinician review a refill of any needed medications

 

This study sought to examine this redesign effort within primary care clinics to improve the efficiency of refilling patients' medications during office visits. We studied the effect of this change to evaluate its impact on rates of requests for medication refills outside of office visits across our family medicine clinics. We hypothesized that this change would lead to fewer refill requests outside of office visits.

 

METHODS

Data were collected from 17 family medicine clinics (137 131 empaneled patients) within a large Midwest academic health system. Clinics included 13 community sites and 4 residency clinics. Community clinics primarily focus on patient care, with some medical student education and no resident education, and include suburban, urban, and rural locations. Residency clinics are the primary care sites for resident physicians and the residency teaching faculty.

 

Data from January 2012 through December 2016 were extracted from the institution's Epic Clarity Database (Epic Systems Corp, Verona, Wisconsin). PCR implementation occurred in a rolling fashion between 2013 and 2015. The number of refill encounters per active panel member per month was assessed before and after the implementation of PCR. Refill encounters were defined as any such encounter documented in the EHR. Active panel members were defined as any patient who was assigned to a family physician working at one of the study clinics seen within the health system over the past 3 years.

 

Our intervention was the implementation of PCR workflows. The workflow of interest was a standard rooming process for pediatric and adult patients with the addition of medication review including identification of medications that would require refill within the next 6 months. Those medications requiring refill are pended for signature by the clinicians when they conduct their portion of the visit.

 

Statistical analysis was conducted in 2 stages. In the first stage, the effects of the redesign on medication refill encounters were estimated for each clinic separately. We took the approach of fitting separate models to each clinic, analogous to treating each clinic as a separate study on workflow redesign. For each clinic, an autoregressive time series linear model, with order 1, was used to predict the number of refill encounters per 1000 patients in that clinic's patient panel per month. Models estimated the effects of redesign on an instantaneous difference in refill encounters (level) and difference in month-to-month changes of refill encounters (slope). Seasonal and linear time trends were accounted for in the models. After an initial model fit, observations with residuals greater than 3 standard deviations (SDs) from the mean were excluded and models were refit. Wald tests were used to assess significance of change both in level and in slope. The second stage utilized these individual clinic estimates in an across-clinic estimate, pooled by clinic context (community or residency). Because of heterogeneity between clinics as discussed in results, we did not feel that a single model comprising data from all clinics, even perhaps adjusting for clinic, would be adequate to capture this between-clinic heterogeneity. By analogy to meta-analysis, we combined the individual clinic models across all of the clinics and by clinic context to obtain effects pooled by clinic context on level and on slope, and jointly on the 2 parameters simultaneously. Statistical analysis was performed with R 3.6.1 using the prais package 1.1-1 (https://CRAN.R-project.org/package=prais) to fit autoregressive models.

 

RESULTS

Across all clinics, the total active patient panel averaged 137 131 patients (ranging from 128 416 to 148 087 due to monthly variability), with a range of 2562 patients in the smallest clinic and 17 650 in the largest clinic. During the entire study period, 779 057 medication refill encounters occurred (mean = 95.9 refill encounters per 1000 patients per month). The mean number of refill encounters per 1000 patients per month was 117 (SD = 42.0) prior to redesign and 82.1 (SD = 31.6) afterward. Both the amount and month-to-month variability of refill encounters were quite heterogeneous across clinics. When estimates were pooled across all clinics, before redesign, refill encounters did not significantly change month to month (0.21 difference per 1000 patients per month; 95% CI, -0.10 to 0.52). After redesign, though, there was a difference in refill encounters both in amount and in month-to-month change (Wald P < .001; Figure 1, Table). The expected monthly number of refill encounters decreased by 24.55 per 1000 patients (95% CI, 13.21 to 35.89). Refill encounters continued to drop an estimated 0.46 per 1000 patients (95% CI, 0.04 to 0.89) with each additional month, relative to the rate before redesign.

  
Figure 1 - Click to enlarge in new windowFigure 1. Model results pooled across all clinics for before redesign and after. After the redesign, refill encounters decreased by 25 per 1000 patients (95% CI, 13.30 to 36.80) (drop from pre to post redesign) and continued to decrease each month by 0.51 per 1000 patients (95% CI, 0.07 to 0.94) (difference in slopes between pre to post redesign).
 
Table. Results of Ti... - Click to enlarge in new windowTable. Results of Time Series Modeling of Refill Changes After PCR

In a subgroup analysis by clinic context, redesign was associated jointly with a difference in amount and month-to-month change of refill encounters for community clinics (Wald P = .001) but not for residency clinics (P = .436; Figure 2, Table). The expected monthly number of refill encounters decreased by 28.00 per 1000 patients (95% CI, 14.49 to 41.51) in community clinic sites. No statistically significant change was observed in residency clinics, with a drop observed of 13.71 per 1000 patients (95% CI, -10.03 to 37.46). With each additional month after redesign, the expected number of refill encounters continued to decrease relative to the rate before redesign by 0.56 per 1000 patients (95% CI, 0.02 to 1.10) for community clinic sites and no significant change at residency clinic sites, with 0.27 per 1000 patients decrease each month (95% CI, -0.73 to 1.26).

  
Figure 2 - Click to enlarge in new windowFigure 2. Model results pooled by clinic context before redesign and after divided by community and residency sites. Thick lines are the pooled results; thin lines represent individual clinic results.

DISCUSSION

Our analysis demonstrated that a workflow change in which medical assistants addressed medication refills during office visits was associated with reduced rates of subsequent refill encounters in the EHR. We found both an initial reduction in refill encounters following implementation of the PCR workflows and a significant downward trend in the subsequent rate of refill encounters. These results became more nuanced when clinics were grouped by context. Community clinics demonstrated post-redesign reductions in refill encounters as well as the subsequent rate of change in refill encounters after implementation. Residency clinics did not have any significant reductions in refill encounters after redesign or in the subsequent rate of change.

 

We hypothesize that the absence of change in the residency clinics may be related to the relatively fewer clinic sessions per week for each resident and faculty member than for their community clinic colleagues. Less patient care time may lead to more difficulty in uniformly implementing systematic changes and to more inconsistent staff relationships. In contrast, the community clinics have physicians with a higher clinical time allocation on average, allowing for smaller and more consistent team structures with physicians and staff. The variation by clinic contexts in this study suggests that different approaches to implementing care redesign are needed on the basis of clinic characteristics. While the organization was striving to create uniformity in the patient experience, different clinic contexts may require variation in the approach to adopt new workflows. In particular, residency clinics may benefit from strategies that focus on the creation and support of a larger team. Such a team can promote practice-level continuity rather than continuity with individual clinicians who may be in the office less often and have increased variation in staff with whom they are paired while in clinic. Further study would help to better understand characteristics of residency clinic sites that may have led to lack of significant changes and can support a more tailored approach for those contexts.

 

Our study had several limitations. The clinics included, though variable in size and location, were all part of the same Midwest academic health care system, thus reducing generalizability. In addition, our study design was observational, with no randomization or control group limiting our ability to assess for causation. Finally, refill encounter frequency was evaluating a specific encounter type; however, medications can be refilled through other types of encounters, introducing the possibility of unanticipated impacts of workflows shifting refill work to different encounter types not evaluated in this study. Future work should focus on a more rigorous experimental design when implementing operational changes designed to improve clinic efficiency and clinician and patient satisfaction.

 

Despite these limitations, the results we present are important in that they demonstrate a measurable reduction in the burden of asynchronous work, which can be a driver of physician dissatisfaction and burnout. In addition, redesigning the prescription refill process offers a more patient-centric process through reducing the number of contacts required and potentially reducing delays in prescription refill authorizations. Other clinics could consider studying these workflow redesign processes in order to understand what tailored strategies are required for effective care that achieves the quadruple aim in their environments.

 

REFERENCES

 

Bae J., Encinosa W. E. (2016). National estimates of the impact of electronic health records on the workload of primary care physicians. BMC Health Services Research, 16, 172. doi:10.1186/s12913-016-1422-6 [Context Link]

 

Bodenheimer T., Sinsky C. (2014, November-December). From triple to quadruple aim: Care of the patient requires care of the provider. Annals of Family Medicine, 12(6), 573-576. doi:10.1370/afm.1713 [Context Link]

 

Cassidy I. B., Keith M. R., Coffey E. L., Noyes M. A. (1996, July-August). Impact of pharmacist-operated general medicine chronic care refill clinics on practitioner time and quality of care. Annals of Pharmacotherapy, 30(7-8), 745-751. doi:10.1177/106002809603000707 [Context Link]

 

Cheriff A. D., Kapur A. G., Qiu M., Cole C. L. (2010, July). Physician productivity and the ambulatory EHR in a large academic multi-specialty physician group. International Journal of Medical Informatics, 79(7), 492-500. doi:10.1016/j.ijmedinf.2010.04.006 [Context Link]

 

Fleming N. S., Becker E. R., Culler S. D., Cheng D., McCorkle R., da Graca B., Ballard D. J. (2014, February). The impact of electronic health records on workflow and financial measures in primary care practices. Health Services Research, 49(1, Pt. 2), 405-420. doi:10.1111/1475-6773.12133 [Context Link]

 

Goldzweig C. L., Orshansky G., Paige N. M., Towfigh A. A., Haggstrom D. A., Miake-Lye I., Shekelle P. G. (2013, November 19). Electronic patient portals: Evidence on health outcomes, satisfaction, efficiency, and attitudes: A systematic review. Annals of Internal Medicine, 159(10), 677-687. doi:10.7326/0003-4819-159-10-201311190-00006 [Context Link]

 

Lam J. G., Lee B. S., Chen P. P. (2016, January 13). The effect of electronic health records adoption on patient visit volume at an academic ophthalmology department. BMC Health Services Research, 16, 7. doi:10.1186/s12913-015-1255-8 [Context Link]

 

McDonald C. J., Callaghan F. M., Weissman A., Goodwin R. M., Mundkur M., Kuhn T. (2014, November). Use of internist's free time by ambulatory care electronic medical record systems. JAMA Internal Medicine, 174(11), 1860-1863. doi:10.1001/jamainternmed.2014.4506 [Context Link]

 

McKinnon A., Jorgenson D. (2009, December). Pharmacist and physician collaborative prescribing: For medication renewals within a primary health centre. Canadian Family Physician, 55(12), e86-e91. [Context Link]

 

Murphy D. R., Reis B., Sittig D. F., Singh H. (2012, February). Notifications received by primary care practitioners in electronic health records: A taxonomy and time analysis. American Journal of Medicine, 125(2), 209.e1-e7. doi:10.1016/j.amjmed.2011.07.029 [Context Link]

 

Riege V. (2005). A patient safety program and research evaluation of U.S. Navy pharmacy refill clinics. In Henriksen K., Battles J. B., Marks E. S., Lewin D. I. (Eds.), Advances in patient safety: From research to implementation (Vol. 1, pp. 213-224). Rockville, MD: Agency for Healthcare Research and Quality. [Context Link]

 

Samaan Z. M., Klein M. D., Mansour M. E., DeWitt T. G. (2009). The impact of the electronic health record on an academic pediatric primary care center. The Journal of Ambulatory Care Management, 32(3), 180-187. doi:10.1097/JAC.0b013e3181ac9667 [Context Link]

 

Shipman S. A., Sinsky C. A. (2013). Expanding primary care capacity by reducing waste and improving the efficiency of care. Health Affairs (Project Hope), 32(11), 1990-1997. doi:10.1377/hlthaff.2013.0539 [Context Link]

 

Sinsky C., Colligan L., Li L., Prgomet M., Reynolds S, Goeders L., Blike G. (2016). Allocation of physician time in ambulatory practice: A time and motion study in 4 specialties. Annals of Internal Medicine, 165(11), 753-760. doi:10.7326/M16-0961 [Context Link]

 

Sinsky T. A., Sinsky C. A. (2012). A streamlined approach to prescription management. Family Practice Management, 19(6), 11-13. [Context Link]

 

Zheng K., Ciemins E., Lanham H., Lindberg C. (2015). Examining the relationship between health IT and ambulatory care workflow redesign (Prepared by Billings Clinic under Contract No. 290-10-0019I-1) [AHRQ Publication No. 15-0058-EF]. Rockville, MD: Agency for Healthcare Research and Quality. [Context Link]

 

medication refills; primary care; primary care redesign