Effects of Home Self Management and Medication Adherence Evaluations by Telemedicine Interventions in Chronic Heart Failure

Veterans Administration Central California Health Care System

Summary

The goal of this program was to determine if telemonitoring of patients with congestive heart failure (CHF), a condition that uses more Medicare dollars than any other diagnosis, could significantly decrease all cause as well as CHF-related hospital admission rates among CHF patients served by the Veterans Administration Central California Health Care System (VACCHCS).

Using a randomized control trial (RCT) design, 113 CHF patients were assigned to telemonitoring care (intervention) or to usual care (control). Data were collected using the VA computerized patient records database for the 12 months preceding telemonitoring, and the 12 months during which telemonitoring was conducted. Results suggest that telemonitoring significantly reduced CHF-related hospital bed days for the intervention group using a pre- and post-use comparison of telemonitoring. However, there was no statistically significant improvement in CHF-related hospital bed days when comparison was made between the intervention and control groups.

Context

As the US population ages and the number of people living with a chronic disease rises, it becomes increasingly important for healthcare providers to provide patients with high quality but low-cost care. Previous studies have shown that use of telemonitoring is associated with improved clinical outcomes and self-management behaviors among patients with chronic conditions. Furthermore, telemonitoring fosters early detection of symptoms and a reduction in complications, which decrease the cost and burden of multiple emergency room (ER) visits and hospital admissions. Though there have been many recent studies of telemonitoring for CHF patients, a limited number have used a RCT design.

The VACCHCS patient population comes from a large geographical area that includes six partially rural, rural, and medically underserved counties in Central California. During FY 08, VACCHCS provided care to over 30,000 unique patients which accounted for approximately 276,000 outpatient visits and 3,000 inpatient admissions. Out of over 30,000 patients who received care at VACCHCS, approximately 2000 patients had a CHF diagnosis and over 700 of them had one or more admissions or ER visits during that time period.

Providing access to necessary medical services and assuring adherence to necessary medications for complex medical problems, such as CHF, in a predominantly elderly and rural population is extremely challenging. Effective home telemonitoring programs could meet some of these challenges with significantly lower operational costs compared to episodic care delivery at hospitals and clinic.

Methodology

Patients with a diagnosis of CHF who had > 1 hospital admission or ER visit during the 24 month period prior to the program start date were identified using the VACCHCS Heart Failure Registry. Patients who are unable to use telemonitoring equipment due to disability (e.g. blindness, severe conditions of hand deformity), non-ambulatory patients who depended on in-home caregivers, or patients not living at home (e.g., assisted home setting or nursing home) were excluded.

The VACCHCS randomly assigned 113 patients who had a formal diagnosis of CHF and had been admitted to the Fresno VA Hospital or had visited the ER in the past 24 months to either telemonitoring care using the Bosch Health Buddy™ as an adjunct to usual care (intervention group) or to solely usual care (control group). Both groups continued their routine medical care for CHF at VACCHCS. Once a day, patients entered information regarding symptoms, weight, and vital signs into the telemonitoring system and answered questions generated by the CHF Disease Management Protocol (DMP). This information was transferred via a secure network to a secure database. A Registered Nurse Care Coordinator viewed the information and then took appropriate action per standard VACCHCS protocols. Data were collected using the VA computerized patient records database for the 12 months preceding telemonitoring, and the 12 months during which telemonitoring was conducted. Admission rates, ER visits, total hospital bed-days used, 30-day re-admissions, and medication adherence, as measured by medication possession rate (MPR), were compared between the two groups. The data analysis was based on the 113 patients who had completed the 12-month follow-up. T test and Cox regression analysis were used to compare the two groups and a hazard ratio (HR) with 95% confidence interval was conducted. Intention-to-treat analysis was used to avoid dropout effects.

Results

The patients’ mean age was 73 years and 99% were male. At baseline, no significant differences between the two arms in demographics, co-morbidities, and medication profiles were noted, except for less use of ACE-I/ARB and Spironolactone in the control group. The overall attrition rate was 30%. This was due to technical problems, non-compliance, and discontinuation of telemonitoring as a result of relocation to a skilled nursing facility or residence of a family member. Patients who discontinued use of the Health Buddy were found to be much older than those who continued its use (83 years of age versus 72 years of age). Patient enrollment was constrained due to limitations of patients’ physical and cognitive capability, lack of residential telephone landlines, and a complicated consenting process. At the end of the 12-month follow-up, a paired t test for a within group, pre-post comparison showed significantly (P=0.024) reduced all cause 30-day admission rate in the control group, and significantly (P=0.025) reduced CHF related hospital bed days and marginally (P=0.059) reduced CHF related admissions in the intervention group. Comparison between the intervention and control groups, however, was not statistically significant different along all outcome measures.

Cox regression analysis showed that there was no statistically significant difference in all-cause mortality between the two groups after adjusting for confounders, including age, sex, co-morbid conditions, and use of certain medications. (HR by Cox Regression model: HR 1.004, 95% CI 0.688 – 1.466).

Study limitations include a predominantly male population, small sample size, enrollment constraints, and extended duration due to conducting a RCT. In addition, it is possible that weight changes and some of the pre-selected parameters for CHF monitoring may not be sensitive surrogates to represent early clinical de-compensation, and the algorithm with multiple questions may not provide timely instructions for patient self-care to prevent CHF exacerbation. Future studies should include a larger sample size, more diverse study populations, and modified parameters and care algorithms.

Lessons Learned:

RCT Design Can Limit Program Implementation and Success. A RCT design with strict Institutional Review Board (IRB) requirements for informed consent introduced several obstacles and artificialities to the care process and patient-provider relationship. First, patients were recruited via the VACCHCS research department rather than through referral by patients’ primary care physician. Second, the IRB for VACCHCS required a complicated informed consent process which hampered VACCHCS’s ability to meet their recruitment goals. The consent process included a 12-page consent form and necessitated a special in-person clinic visit for researchers to review the form with the patient. Originally intended as a one-year demonstration project that would evaluate a minor enhancement to a standard VA practice, this project became a difficult, drawn out, and artificial CHF quality improvement program.

Use of Opt-in rather than Opt-out Patient Recruitment Strategy. In addition to adding steps to the patient recruitment process, the RCT design prevented use of an “opt-out” patient recruitment strategy, a best practice utilized by leading telemonitoring providers. Opt-out recruitment provides a much larger pool of eligible patients and facilitates enrollment because it is assumed that patients are suitable and want to utilize telemonitoring until providers are told otherwise. In the opt-in scenario, as applied in this VACCHCS program, patients may perceive the telemonitoring program as an unusual or non-standard practice, and/or physicians must be aware and sufficiently interested in the program in order to actively enroll their patients. The opt-out approach removes administrative and other barriers to recruitment.