Implementation of a Population-Based Cirrhosis Identification and Management System

Cirrhosis-related morbidity and mortality is potentially preventable. Antiviral treatment in patients with cirrhosis-related to hepatitis C virus (HCV) or hepatitis B virus can prevent complications.^1–3 Beta-blockers and endoscopic treatments of esophageal varices are effective in primary prophylaxis of variceal hemorrhage.⁴ Surveillance for hepatocellular cancer is associated with increased detection of early stage cancer and improved survival.⁵ However, many patients with cirrhosis are either not diagnosed in a primary care setting, or even when diagnosed, not seen or referred to specialty clinics to receive disease-specific care,⁶ and thus remain at high risk for complications.

Our goal was to implement a population-based cirrhosis identification and management system (P-CIMS) to allow identification of all patients with potential cirrhosis in the health care system and to facilitate their linkage to specialty liver care. We describe the implementation of P-CIMS at a large Veterans Health Administration (VHA) hospital and present initial results about its impact on patient care.

Population-Based Cirrhosis Identification and Management System Intervention

P-CIMS is a multicomponent intervention that includes a secure web-based tracking system, standardized communication templates, and care coordination protocols.

Web-Based Tracking System

An interdisciplinary team of clinicians, programmers, and informatics experts developed the P-CIMS software program by extending an existing comprehensive care tracking system.⁷ The P-CIMS program (referred to as cirrhosis tracker) extracts information from VHA’s national corporate data warehouse. VHA corporate data warehouse includes diagnosis codes, laboratory test results, vital status, and pharmacy data for each encounter in the VA since October 1999. We designed the cirrhosis tracker program to identify patients who had outpatient or inpatient encounters in the last 3 years with either ≥1 cirrhosis diagnosis (defined as any instance of previously validated International Classification of Diseases-9 and −10 codes)⁸; or possible cirrhosis (defined as either aspartate aminotransferase to platelet ratio index >2.0 or Fibrosis-4 >3.24 in patients with active HCV infection⁹ [defined based on positive HCV RNA or genotype test results]).

The user interface of the cirrhosis tracker is designed for easy patient lookup with live links to patient information extracted from the corporate data warehouse (recent laboratory test results, recent imaging studies, and appointments). The tracker also includes free-text fields that store follow-up information and alerting functions that remind the end user when to follow up with a patient. Supplementary Figure 1 shows screen-shots from the program.

We refined the program through an iterative process to ensure accuracy and completeness of data. Each data element (eg, cirrhosis diagnosis, laboratory tests, clinic appointments) was validated using the full electronic medical record as the reference standard; this process occurred over a period of 9 months. The program can run to update patient data on a daily basis.

Standardized Communication Templates and Care Coordination Protocols

Our interdisciplinary team created chart review note templates for use in the VHA electronic medical record to verify diagnosis of cirrhosis and to facilitate accurate communication with primary care providers (PCP) and other specialty clinicians. We also designed standard patient letters to communicate the recommendations with patients. We established protocols for initial clinical reviews, patient outreach, scheduling, and follow-ups. These care coordination protocols were modified in an iterative manner during the implementation phase of P-CIMS.

Setting and Patients

Michael E. DeBakey VA Medical Center (MEDVAMC) in Houston provides care to more than 111,000 Veterans, including more than 3800 patients with cirrhosis. At the time of P-CIMS implementation, there were 3 hepatologists and 4 advanced practice providers (APP) who provided liver-related care at the MEDVAMC.

The primary goal of the initial phase of implementation was to link patients with cirrhosis to regular liver-related care. Thus, the sample was limited to patients who did not have ongoing specialty care (ie, no liver clinic visits in the last 6 months, including patients who were never seen in liver clinics).

Implementation Strategy

We used implementation facilitation (IF), an evidence-based strategy, to implement P-CIMS.¹⁰ The IF team included facilitators (F.K., D.S.), local champions (S.M., K.H.), and technical support personnel (eg, tracker programmers). Core components of IF were leadership engagement, creation of and regular engagement with a local stakeholder group of clinicians, educational outreach to clinicians and support staff, and problem-solving. The IF activities took part in 2 phases: pre-implementation and implementation.

Preimplementation Phase

We interviewed key stakeholders to identify facilitators and barriers to P-CIMS implementation. One of the implementation facilitators (F.K.) obtained facility and clinical section’s leadership support, engaged key stake-holders, and devised a local implementation plan. Stakeholders included leadership in several disciplines: hepatology, infectious diseases, and primary care. We developed a map of clinical workflow processes to describe optimal integration of P-CIMS into existing workflow (Supplementary Figure 2).

Implementation Phase

The facilitators met regularly (bi-weekly for the first year) with the stakeholder group including local champions and clinical staff. One of the facilitators (D.S.) served as the liaison between the P-CIMS team (F.K., A.M., R.M., T.T.) and the clinic staff to ensure that no patients were getting missed and to follow through on patient referrals to care. The programmers troubleshot technical issues that arose, and both facilitators worked with clinical staff to modify workflow as needed. At the start of IF, the facilitator conducted an initial round of trainings through in-person training or with the use of screen-sharing software. The impact of P-CIMS on patient care was tracked and feedback was provided to clinical staff on a quarterly basis.

Implementation Results: Linkage to Liver Specialty Care

P-CIMS was successfully implemented at the MED-VAMC. Patient data were first extracted in October 2015 with 5 updates through March 2017. In total, 4 APP, 1 MD, and the facilitator used the cirrhosis tracker on regular basis. The clinical team (APP) conducted the initial review, triage, and outreach. It took on average 7 minutes (range, 2–20 minutes) for the initial review and outreach. The APPs entered each follow-up reminder in the tracker. For example, if they negotiated a liver clinic appointment with the patient, then they entered a reminder to follow up with the date by which this step (patient seen in liver clinic) should be completed. The tracker has a built-in alerting function. The implementation team was notified (via the tracker) when these tasks were due to ensure timely receipt of recommended care processes.

We identified 2065 patients who met the case definition of cirrhosis (diagnosed and potentially undiagnosed) and were not in regular liver care. Based on initial review, 1507 patients had an indication to be seen in the liver clinic. Among the remaining 558, the most common reasons for not requiring liver clinic follow-up were: being seen in other facilities (138 in other VHA and 51 in outside hospitals), followed in other specialty clinics (eg, liver transplant or infectious disease, n = 93), or absence of cirrhosis based on initial review (n = 165) (see Figure 1 for other reasons).

Figure 1.

Initial results of P-CIMS implementation. Identification of patients with cirrhosis and their linkage to specialty care. EMR, electronic medical record; HCC, hepatocellular carcinoma.

We used 2 different strategies to reach out to the patients. Of the 1507 patients, 597 were previously seen in the liver clinics but were lost to follow-up. These patients were contacted directly by the liver clinic staff. The other 910 patients with cirrhosis (of 1507) had never been seen in the ambulatory liver clinics (n = 559) or were seen more than 2 years before the implementation of cirrhosis tracker (n = 351). These patients were reached out through their PCPs. We used standard electronic medical record templates to request PCP’s assistance in reviewing patient’s records and submitting a liver consultation after they discuss the need for liver evaluation with the patient.

Of the 597 patients who were previously seen but lost to follow up, we successfully contacted 404 (67.7%) patients via telephone and/or letters (for latter, success was defined when patients called back); of these 335 (82.9%) patients had clinic appointments scheduled. In total, 208 (51.5% of 404; 34.8% of 597) patients were subsequently seen in the liver clinics during a median of 12-month follow-up. As shown in Figure 1, the most common reasons for inability to successfully link patients to the clinic were at the patient level, including no show, cancellation, and noninterest in seeking liver care. It took on average 1.5 attempts (range, 1–4) to link 214 patients to liver clinic.

Of the other 910 patients with cirrhosis, 686 (75.4%) were successfully contacted; and of these 365 (53.2%) patients had liver clinic appointments scheduled. In total, 225 (61.7% of 365; 24.7% of 910) patients were seen in the liver clinics during a median of 12-month follow-up. The reasons underlying inability to link patients to liver specialty clinics are listed in Figure 1 and included shortfalls at the PCP and the patient levels. It took on average 2.4 attempts (range, 1–5) to link 225 patients to liver clinic.

A total of 124 patients were initiated on direct-acting antiviral agents for HCV treatment and 18 new hepatocellular carcinoma cases were diagnosed as part of P-CIMS.

Discussion and Future Directions

We learned several lessons during this initiative. First, it was critical to allow time to iteratively revise the cirrhosis tracker program, with input from key stake-holders, including clinician end users. For example, based on feedback, the program was modified to exclude patients who had died or those who were seeking primary care at other VHA facilities. Second, merely having a program that accurately identifies patients with cirrhosis is not the same as knowing how to get organizations and providers to use it. We found that it was critical to involve local leadership and key stakeholders in the preimplementation phase to foster active owner-ship of P-CIMS and to encourage the rise of natural champions. Additionally, we focused on integrating P-CIMS in the existing workflow. We also had to be cognizant of the needs of patients, such as potential problems with communication relating to notification and appointments for evaluation. Third, several elements at the facility level played a key role in the successful implementation of P-CIMS, including the culture of the facility (commitment to quality improvement); leadership engagement; and perceived need for and relative priority of identifying and managing patients with cirrhosis, especially those with chronic HCV. We also had strong buy-in from the VHA National Program Office tasked with improving care for those with liver disease, which provided support for development of the cirrhosis tracker.

Overall, our early results show that ~30% of patients with cirrhosis without ongoing linkage to liver care were seen in the liver specialty clinics because of P-CIMS. This proportion should be interpreted in the context of the patient population and setting. Cirrhosis disproportionately affects vulnerable patients, including those who are impoverished, homeless, and with drug- and alcohol-related problems; a complex population who often have difficulty staying in care. Most patients in our sample had no linkage with specialty care. It is plausible that some patients with cirrhosis would have been seen in the liver clinics, regardless of P-CIMS. However, we expect this proportion would have been substantially lower than the 30% observed with P-CIMS.

We found several barriers to successful linkage and identified possible solutions. Our results suggest that a direct outreach to patients (without going through PCP) may result in fewer failures to linkage. In total, ~35% of patients who were contacted directly by the liver clinic met the endpoint compared with w25% of patients who were contacted via their PCP. Future iterations of P-CIMS will rely on direct outreach for most patients. We also found that many patients were unable to keep scheduled appointments; some of this was because of inability to come on specific days and times. Open-access clinics may be one way to accommodate these high-risk patients. Although a full cost-effectiveness analysis is beyond the scope of this report, annual cost of maintaining P-CIMS was less than $100,000 (facilitator and programming support), which is equivalent to antiviral treatment cost of 4 to 5 HCV patients, suggesting that P-CIMS (with ability to reach out to hundreds of patients) may indeed be cost-effective (if not cost-saving).

In summary, we built and successfully implemented a population-based health management system with a structured care coordination strategy to facilitate identification and linkage to care of patients with cirrhosis. Our initial results suggest modest success in managing a complex population who often have difficulty staying in care. The next steps include comparing the rates of linkage to specialty care with rates in comparable facilities that did not use the tracker; broadening the scope to ensure patients are retained in care and receive guideline-concordant care over time. We will share these results in a subsequent manuscript. To our knowledge, cirrhosis tracker is the first informatics tool that leverages data from the electronic medical records with other tools and strategies to improve quality of cirrhosis care. We believe that the lessons that we learned can also help inform efforts to design programs that encourage use of administrative data–based risk screeners to identify patients with other chronic conditions who are at risk for suboptimal outcomes.

Abbreviations used in this paper:

APP

advanced practice providers

HCV

hepatitis C virus

IF

implementation facilitation

MEDVAMC

Michael E. DeBakey VA Medical Center

P-CIMS

population-based cirrhosis identification and management system

PCP

primary care providers

VHA

Veterans Health Administration