Search

Objective: Using Failure Mode and Effects Analysis (FMEA) as an example quality improvement approach, our objective was to evaluate whether secondary use of orders, forms, and notes recorded by the electronic health record (EHR) during daily practice can enhance the accuracy of process maps used to guide improvement. We examined discrepancies between expected and observed activities and individuals involved in a high-risk process and devised diagnostic measures for understanding discrepancies that may be used to inform quality improvement planning.Methods: Inpatient cardiology unit staff developed a process map of discharge from the unit. We matched activities and providers identified on the process map to EHR data. Using four diagnostic measures, we analyzed discrepancies between expectation and observation.Results: EHR data showed that 35% of activities were completed by unexpected providers, including providers from 12 categories not identified as part of the discharge workflow. The EHR also revealed sub-components of process activities not identified on the process map. Additional information from the EHR was used to revise the process map and show differences between expectation and observation.Conclusion: Findings suggest EHR data may reveal gaps in process maps used for quality improvement and identify characteristics about workflow activities that can identify perspectives for inclusion in an FMEA. Organizations with access to EHR data may be able to leverage clinical documentation to enhance process maps used for quality improvement. While focused on FMEA protocols, findings from this study may be applicable to other quality activities that require process maps.

Background: Multidisciplinary, collaborative care improves outcomes and reduces costs but may unintentionally increase the operational overhead to manage the complexity of increasingly sophisticated teamwork. Objective: To quantify complexity of a standardized discharge planning process by closely examining team continuity, composition, and experience for both explicitly defined, deliberately organized teams and implicitly defined, organically assembled teams. Methods: We examined discharge planning team membership for inpatient cardiology encounters with a length of stay (LOS) greater than 48 hours for a three-year period from 2012-2014 in a large academic medical center. By constructing a co-affiliation graph from transactional EHR data, we determined the frequency of team occurrence and size. We then calculated the shared experience of teammates using a pairwise similarity metric and identified modularity, groups of healthcare personnel with dense connections, within the graph by applying a community detection algorithm. Results: We identified 52,254 transactions for 3,213 encounters with an average LOS of 8.7 days for the time period. The standard discharge planning team accounted for 41,101 (79%) transactions, consisted of 7 team member types, and comprised 709 individual providers performing 36 activities. We identified 569 additional discharge planning team members, consisting of 22 additional provider types. When constrained to only explicitly defined teams, 404 unique teams with an average size of 5.9 members occurred with a mean of 8 (min=1, max=118) shared encounters. When unconstrained, organically assembled teams with no explicit definition occurred 3,209 times (mean=1; min=1, max=3) with an average size of 7.3 members. Single-encounters accounted for 21,107 (62%) of all provider pairs. However, over 50% of all discharges had at least 1 pair sharing over 99 encounters (median=4 pairs). Community detection found 9 modules (range: 12-275 members) among pairs with more than 5 shared encounters. Conclusions: We have shown by deconstructing digital interactions via the EHR, a core group of providers, defined by role and activity, anchor most discharge planning teams. However, the EHR can also identify identify the 20% of teams with constantly recombining membership due to situational care; this can impose overhead when targeting team-wide process improvements, communication strategies, or educational initiatives.

Background: The vulnerable period for hospital readmission begins upon hospital discharge, a complex, interdisciplinary process that necessitates close teamwork for accurate execution of the discharge plan. Thus, better understanding of the quality of teamwork throughout the discharge process may inform strategies to reduce hospital readmissions rates. Novel methods using a network analytics approach to quantify teamwork may better characterize this critical clinical process, facilitate quality improvement (QI), and become an important tool in learning healthcare systems. Methods: We extracted three years of patient, provider, and activity data related to discharge planning for an inpatient cardiology unit from Northwestern Medicine’s Enterprise Data Warehouse. We then created a provider-patient network to identify providers who shared patients and calculated readmissions rates for provider pairs. Using these data, we calculated a novel parameter, the Shared Positive Outcome Ratio (SPOR), an objective composite measure that quantifies the concentration of positive outcomes over a set of shared patients. To identify significant low-readmission and high-readmission collaborative relationships, we compared the observed network to 1000 sample networks containing randomized readmission values. Results: We identified 133,927 actions distributed among 38 discharge activity types performed during 13,720 patient encounters. The collaborative network was composed of 1,542 providers including 503 nurses, 432 residents, 207 physicians, 111 physical and occupational therapists, 59 medical students, 32 dieticians, and other medical and administrative staff. The average encounter involved 4 providers performing 10 discharge-related actions. After pruning the network to include only provider pairs with 6 or more shared patients, we found that 6% of collaborative interactions had a significantly low SPOR, indicating lower than expected readmission rates. Conversely, 12% of collaborative interactions had a significantly high SPOR, indicating higher than expected readmission rates. We identified 21 providers who had a low SPOR for a significant percentage of their collaborations, indicating potential top performers in the teamwork domain. Conclusions: Readmission rates appear to vary across different teams of providers. Team collaboration can be quantified using a composite measure of collaboration across provider pairs. Tracking provider pair outcomes over a sufficient set of shared encounters may inform various QI strategies, such as optimizing team staffing, identifying high-performing teams who can share their best practices, and redesigning discharge care processes. Ongoing work on this model is focused on accurately risk-adjusting outcomes, which will increase the robustness of this method.

Background: Multidisciplinary transitional care teams represent a model for reducing heart failure readmissions. Within this context, early identification of patients hospitalized with acute decompensated heart failure (ADHF) permits meaningful transitional care plan development. Improving the efficiency of early identification of the higher risk ADHF patient represents an area not well studied in hospitalized heart failure (HF). Objective: To validate the sensitivity and specificity of an enterprise data warehouse (EDW)-based strategy for early identification of patients with ADHF. Methods: An EDW query was constructed to identify patients with ADHF based on clinical and diagnosis-related parameters, including BNP level and administration of intravenous diuretics. The EDW query was run daily; expert clinicians verified the diagnosis of ADHF based on comprehensive chart review. This classification was used to determine specificity of the query for ADHF. We computed the sensitivity of the EDW-based approach by matching query results to heart failure diagnosis related group (DRG) data and primary discharge diagnosis data from separate hospital systems. Results: During the study period of 70 days, a total of 2354 charts were screened (33.6 charts per day). A total of 410 patients were identified by chart review as having heart failure requiring active management, for a specificity of 17.4%. Sensitivity was computed using both heart failure DRG data and primary discharge diagnosis data. Of the 114 patients discharged with a heart failure DRG (291, 292, or 293), all 114 were detected a priori by the admission EDW screen, for a sensitivity of 100%. A similar analysis conducted using HF principal diagnoses, which includes cardiac surgery-related admissions, yielded a sensitivity of 97.2%. Conclusions: EDW-based screening of patients based on simple clinical parameters early in the hospitalization is highly sensitive for detection of ADHF hospitalizations, but specificity is low. Brief chart review by expert clinicians is rapid, and identifies a specific cohort of patients that can be targeted for multidisciplinary HF transitional care. A better delineation of risk has broad outpatient workflow implications. Ongoing process improvements will demonstrate if early identification of at-risk patients yields significant reduction in HF readmissions.

Objectives: Failure Mode and Effects Analysis (FMEA) is a frequently-used approach for prospective risk assessment and quality improvement in healthcare, particularly for high-risk care processes such as hospital discharge planning. Our goal was to evaluate whether secondary use of metadata collected by the electronic health record (EHR) during daily practice can inform assembly of a comprehensive FMEA team by showing: 1) discrepancies between expected and observed process activities and individuals involved, and 2) the presence of individuals who may be appropriate to include in an FMEA based on their variable familiarity with a process. Methods: We extracted discharge planning data for an inpatient cardiology unit from the Enterprise Data Warehouse (EDW) and compared it to a hand-drawn map (HDM) indicating clinicians’ understanding of discharge activities and providers expected to complete each activity. We assessed the presence of providers highly experienced in the process, the diversity of involved disciplines, and the accuracy of the HDM compared to observation from EDW data. Findings: Over 500 providers completed nearly 35,000 discharge-related activities across 18 activity types over 2,000 encounters. Experience was skewed such that 90% (510 of 569) of providers completed between 0 and 99 activities while the remaining 10% (59 of 569) performed up to 1,200 activities. Frequent performers completed similar activities to their peers, but did so as many as 12 times more frequently than average for their discipline. Expectation of who performed an activity closely matched observation for 11 discipline-specific activities, such as case management assessment. However, providers from up to 10 different disciplines performed the remaining 7 activities, such as scheduling a follow-up visit or ordering a therapy consult. Overall, 35% (12,183 of 34,939) of activities were performed by an unexpected provider. Conclusions: Analyzing metadata from EHRs is a novel method to inform FMEA of high-risk processes. This study provides a framework for assessing process activities and the providers involved. In the discharge planning process, there appears to be significant discrepancy between clinicians’ understanding and the actual discharge process and team, which suggests the presence of providers who could be overlooked during typical FMEA team construction. This methodology can empirically enrich the FMEA team and highlight quality improvement target areas.

Background: Heart failure (HF) readmissions remain a major driver of cost and health care utilization. Timely follow-up of patients post-discharge represents an evidence-based intervention proven to reduce readmission rates. A previously unexplored characteristic of hospital discharges is variability in discharge caseload. This variability thwarts the timeliness of follow-up, negates the benefit of transition care planning and may lead to a higher risk of HF readmissions. Queuing theory is the mathematical study of waiting times. We opted to use queuing theory to determine if caseload can be determined more precisely in a manner that sufficiently accommodates HF discharge variability. Objective: To analyze the impact of hospital discharge rate variability on outpatient clinic capacity needs using HF hospitalization discharge data and operations management approaches. Methods: Higher risk hospitalizations requiring active transitional care heart failure management were detected using an enterprise data warehouse-supported process over the study period. Queuing theory approaches were used to model the impact of HF discharge clinic capacity on wait time to an appointment. Discharge clinic was modeled as a single 7-day follow-up appointment, with an acceptable scheduling window of 5 to 9 days post-discharge. Results: During the study period of 100 days, 566 HF discharges were made, for a median of 5.66 discharges daily, or 39.6 discharges weekly. The distribution of daily discharges was skewed rightward (mode = 3, range = 0 to 18, standard deviation = 3.3, coefficient of variation = 0.58). Current clinic design: Providing one discharge slot for every hospital discharge (100% utilization) leads to an average wait of 18.3 days prior to an appointment, with only 31.9% of appointments scheduled within 7 days, and 38.9% of appointments scheduled within 9 days. Clinic re-design (queuing theory): Providing five extra discharge appointment slots per week (88% utilization or 13.6% excess capacity) reduces the expected waiting period to 1.1 days, with 99.8% of patients seen within 7 days, and virtually all patients seen within 9 days of discharge. Conclusions: Deployment of queuing theory allows for a more precise quantification of needed clinical capacity to accomplish appropriate HF follow-up with a reasonable degree of certainty. Our simplified model demonstrates that variability in hospital discharge rates leads to excessive clinic wait times in the absence of a modest capacity buffer and consequently exposes patients to a higher risk of HF readmission. We show using single center HF discharge data that a 10-15% increase in capacity is needed to ensure an adequate follow-up service level. Ongoing process of care work will demonstrate if optimization of clinic load yields a significant reduction in HF readmissions.