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 the 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 the 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 the 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 stakeholders, 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 ownership 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 about 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, about 35% of patients who were contacted directly by the liver clinic met the endpoint compared with about 25% 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 four to five 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.