The materials needed for the TM model included a home monitor attached to an integral scale stand, a data-receiving computer, and a network server connection. Other than the UIC network server, all hardware and software were developed by and purchased from AvidCare, formerly Advanced Medical Devices (Milwaukee, WI). All devices used were extensively validated prior to this study and subsequently approved for use by the Federal Drug Administration (FDA) in 1996.

The home monitor (HM) is a small, lightweight device (64mm x 152mm x 89mm, 680 grams), that automatically notifies patients to obtain their vital signs. The display on the device has a readability equivalent to the 5th grade reading level. The device shares patients’ existing telephone lines and does not cause interruption of existing phone service, except during data transmissions. The HM was programmed for this study to measure blood pressure (BP), heart rate (HR), arterial oxygen saturation (O2 sat) and weight. Each of these variables is measured by a clinically proven, FDA-approved monitoring module incorporated into the HM. The rationale for selecting the variables that were monitored is presented following the remaining description of the required materials.

The data-receiving computer is a dedicated IBM-compatible (Dell) personal computer (PC) running proprietary software developed by AvidCare. Aside from storing each patient’s electronic health file, the computer automatically receives and displays patient transmissions as well as audible/visual alarms if any data exceed preset alarm parameters. The receiving computer is positioned on the cardiac surveillance (telemetry) unit of the UIC Medical Center, directly adjacent to the telemetry technician’s station. At the UIC, a telemetry technician is on duty 24 hours per day, seven days per week as part of the existing inpatient telemetry service. All telemetry technicians (and telemetry nursing staff) were given special training with the data-receiving computer, preparing them to effectively operate the computer and follow triage algorithms for any alarms received.

Alarm parameters were established according to each patient’s baseline condition, as deemed appropriate by the medical director and an advanced practice nurse (APN) who collaboratively coordinated the study. Maximum and minimum deviations from baseline were programmed into the monitoring station software such that when a patient’s physiologic data exceeded these limits, an alarm would sound, prompting the attention of the telemetry technicians. The telemetry technicians subsequently followed a triage algorithm that guided the resolution of all alarms. The telemetry technician also paged the APN with all alarm data and actions taken. The utilization of existing telemetry personnel was intended to augment the efficiency of the method.

The UIC has an extensive network server system. Therefore, aside from receiving alphanumeric pages from the telemetry staff regarding aberrant data, the study coordinator and/or the study medical director could also access a patient’s file via a server connection. This allowed for data to be accessed from any remote location, with a computer/laptop modem by dialing into the UIC network.

Nearly 50 million Americans have hypertension,₁₈,₁₉ and after coronary heart disease, it is the most common antecedent of HF.1 Clearly, control of hypertension is critically important to the management of HF, yet national data suggest that overall BP control (< 140/90 mmHg) was achieved by only 27% of diagnosed hypertensive adults from 1991-1994.19 Alarmingly, this percentage is less than the 29% achieved from 1988-1991.18,19 Moreover, BP was chosen as a clinical outcome for this study because hypertension is particularly pervasive among African Americans,18,19 who comprise a racial majority of the UIC patient population. See Table 1 for patient demographics and other characteristics.

The non-invasive blood pressure (NIBP) module within the HM is consistent with the American National Standard for Electronic or Automated Sphygmomanometers, ANSI/AAMI SP10-1992. The module is fully automatic, not requiring manual inflation, and is rated for automatic self-monitoring. The NIBP module is incorporated within the HM control unit and communicates to the HM’s microprocessor. Blood pressure measurements determined with this device are considered equivalent, within the limits prescribed by the standard, to those obtained by intra-arterial measurement. The NIBP device also determines heart rate, although this was not a clinical outcome of this study. Heart rate is an essential physiologic variable to be monitored, and can be especially useful for the management of medications with chronotropic properties.

Oxygen saturation (O2Sat), measured by pulse oximetry, was not a specific or direct clinical outcome in this study. However, guidelines from the Agency for Health Care Policy and Research (AHCPR) recommend that O2Sats below 90 percent, not due to pulmonary disease, are indicative of a need for hospitalization.₂₀ Therefore, oximetry has the potential to provide useful information in the absence of face-to-face contact, which is inherent to telemanagement. Furthermore, since pulse oximetry is evaluated from arterial pulsations, HR data is also made available.

Hyper- and hypovolemia, best measured by a scale, are particularly challenging problems in the management of chronic HF. Daily weights are recommended by most HF experts20-22 for early detection of fluid accumulation or depletion. While fluid weight gain/loss was not a clinical outcome, it indirectly affects multiple other HF outcomes; therefore, objective daily weights are essential data for chronic HF management. Patients in this study were instructed to weigh themselves daily in the morning after voiding, before eating/drinking, and in the same amount of clothing each time. The integral scale stand (to which the home monitor was attached) was capable of weights up to 700 pounds.

All patients admitted to the UIC Medical Center between September 1995 and April 1997 with a diagnosis of HF and meeting inclusion/exclusion criteria were asked to participate. All qualifying participants were required to sign an IRB-approved informed consent.

Inclusion criteria included at least one of the following: 1) documented diagnosis of HF as determined by radiographic evidence of pulmonary congestion; 2) documented New York Heart Association (NYHA) functional classification III or IV; 3) conventional clinical HF symptoms of dyspnea and edema that responded to diuresis; 4) echocardiographic evidence of heart failure (either systolic or diastolic dysfunction).

Exclusion criteria included: 1) unstable angina; 2) renal failure; 3) severe dementia or other debilitating psychiatric disorder; 4) end-stage heart failure requiring regular inotropic infusions; 5) anticipated survival of less than six months; 6) planned discharge to a long-term care facility; ₇) current illicit drug use; ₈) participation in another HF research protocol within the last six months; ₉) scheduled HF-specific home health nursing; and/or ₁₀) lack of an operating home telephone line.

One hundred thirty hospitalized patients were identified during the above time frame with a diagnosis of HF. Seventy patients were excluded from participation for the following reasons: twenty patients had participated in other HF-related research within the last six months; eight patients required dialysis and another four patients were being considered for dialysis; ten patients were scheduled to have HF-specific home nursing services upon discharge; six patients had recent myocardial infarctions resulting in unstable angina; six patients were discharged to long-term care facilities; five patients died before discharge; nine patients did not have operating telephone lines; and two patients were suspected of current drug abuse other than alcohol. No eligible patients refused participation.

Sixty patients (n = 60) met the inclusion/exclusion criteria described above and comprised the study sample. Mean age of participants was 61 ± 13 years and mean NYHA class was 3.2 ± 0.5. The sample was 62% African American, 31% Caucasian, 6% Hispanic, and 1% American Indian. Sixty-seven percent of the sample was female.

This sample is representative of the ethnic and gender composition of the general population of patients who receive health care services at the UIC Medical Center (52% African American). During the study time period, UIC census information further indicates that 54% of the patients hospitalized for HF were female. See Table 1 for other patient demographics and characteristics.

Figure 1

Table 1: Patient demographics and characteristics

NYHA = New York Heart Association; LVEF = left ventricular ejection fraction; HTN = hypertension; CAD = coronary artery disease; IDCM = idiopathic dilated cardiomyopathy; CVA = cerebral vascular attack; HX = history of; DM = diabetes mellitus; COPD = chronic obstructive pulmonary diseases; HS = high school. (values are presented as mean ± SD or rounded percent of total sample)

After discharge from a heart failure hospitalization, patients used the HM for two to three months (mean 76.3 ± 20.4 days) and transmitted data 1-2 times daily (mean 2.0 ± 0.78 times/day). Patients were asked to transmit more frequently to assess responses when medications (especially ACE-inhibitors and beta-blockers) were being actively titrated. Patients/families received HM operating instructions from a trained bio-technician who installed the device in each patient’s home. Installation/instruction time was approximately 30 minutes.

During the monitoring period, telephone interventions were prompted by the alarms received, and telemanagement interventions were based on the existing, evidence-based clinical recommendations/guidelines for HF,20, ₂₁, ₂₂, ₂₃, ₂₄, ₂₅, ₂₆, ₂₇ and hypertension.18,19 Initial calls were made by the telemetry technician who, following triage algorithms, separated emergency from non-emergency alarms. The telemetry technician then communicated all alarm information to the APN (via phone or pager), who subsequently contacted the patient regarding the abberrant data, if necessary. The majority of medication interventions and patient education calls were made directly by the APN following clinical guidelines, but the study medical director was available by pager at all times for consultation and/or questions from the APN. Monitoring was discontinued when a patient demonstrated a continuous two-week period with no significant alarms. Following discontinuation from the physiologic monitoring phase of the TM method, regardless of the monitoring period duration, each patient received monthly assessment calls until one year post study enrollment. No patients resumed use of the HM as a result of the monthly assessment calls, however, several patients (16) were instructed to make outpatient clinic appointments with their respective physicians.

The primary outcomes were analyzed at three, six and twelve months following study enrollment. These data were compared to each participant’s previous history of HF admissions, length of stay, and hospitalization charges from three, six and twelve months prior to enrollment.

The secondary outcome of quality of life was measured using the Minnesota Living with Heart Failure Questionnaire (MLHF), a 21-item questionnaire developed by Rector, Kubo, & Cohn.₂₈, ₂₉, ₃₀, ₃₁ The MLHF provides an overall total score as well as physical and emotional sub-scale scores relating how persons with HF perceive how their condition has affected the quality of their lives in the past month. Only the overall score data only is presented here. The response scale of each item ranges from 0 (none) to 5 (very much). The total score is the sum of the responses to the 21 items, with the sub-scale scores equal to the sum of each of the respective sub-scale items. Total scores can range from 0 to 105, with higher scores indicative of a less favorable quality of life. Baseline quality of life scores were assessed at the beginning of the physiologic monitoring period and compared to scores obtained at three, six and twelve-month time periods thereafter; the post-discharge assessments were each compared to the respective patient’s baseline score.

Data for all variables were tested for normal distribution using the Kolmogorov-Smirnov Goodness-of-Fit test. Paired t-tests for within-group pre-post comparisons were used for normally distributed, continuous variables (with Bonferoni correction whenever appropriate); the chi-square test for any discrete variables; and the Wilcoxon rank-sum test for non-normally distributed categorical and continuous variables. Descriptive statistics and frequencies were used for demographics and patient characteristics. An analysis of co-variance (ANCOVA) was used to test for confounding variables that might have affected the statistical significance of the results. The SPSS 8.0+ for Windows statistical software was used for these analyses. An alpha level of < 0.05, determined a priori, indicated the level of statistical significance.

Positive outcomes at three, six and twelve months after enrollment in the study, demonstrated the clinical and cost effectiveness of the TM method (Table 2). At three months post study enrollment, data for the 58/60 surviving patients indicate an 81% change in HF admissions (p < 0.001), 90% change in LOS for these admissions (p < 0.001), and a 91% decrease in associated hospital charges (p < 0.001) compared to their respective data for the three months preceding study enrollment. The above percentages represent three-month pre/post decreases in HF admissions from 62 to 12, inpatient days from 326 to 35, and hospital charges from $699,880 to $65,613.

At six months post study enrollment, a 77% change in HF admissions (p<0.001), 86% change in LOS for these admissions (p < 0.001), and an 88% decrease in associated hospital charges ( p < 0.001) was demonstrated by the 56/58 surviving patients. These represent decreases in HF admissions from 70 to 16, inpatient days from 395 to 56, and hospital charges from $890,159 to $110,797.

At 12 months post study enrollment, data for the 48 surviving patients indicate a 70% change in HF admissions (p < 0.001), 73 % decrease in LOS (p < 0.001), and a 76 % (p < 0.001) change in hospital charges compared to the 12-month period prior to study enrollment. These percentages represent decreases in HF hospital admissions from 74 to 22, LOS from 388 to 105 and hospital charges from $880,695 to $214,603. See Table 2 below for primary outcome summary.

Figure 2

Table 2: Primary outcome summary

Primary outcomes: heart failure (HF) readmissions, length of stay of HF readmissions, cost of HF readmissions and quality of life at 3, 6 and 12 months. The data at each interval is cumulative. Therefore, as each time point for data comparison increases in duration, so do the number of hospital admissions within each interval. *Percent change was determined by difference between pre intervention and post intervention, at the interval indicated by the “Endpoint” column, relative to values pre intervention.

Significant improvement in quality of life was demonstrated by a 26% change (p < 0.002) from baseline (figure 1). At six and twelve months post enrollment, quality of life continued to remain significantly improved 26% ( p < 0.002) and 23% (p < 0.003) from baseline, respectively.

Figure 4

Quality of life was measured with the Minnesota Living with Heart Failure Questionnaire (permission granted by the Department of Research, University of Minnesota). Higher scores indicate less favorable quality of life as perceived by the patient; 105 is the highest possible score. Quality of Life was assessed at baseline (beginning of home monitoring phase), at discontinuation of the physiologic monitoring (2-3 months post), and every month thereafter to 12 months from study enrollment. Significant change from baseline to discontinuation from monitoring phase depicted in above figure.

Mean arterial pressure (MAP) was assessed for patients who required blood pressure reduction at baseline. Because of the need for afterload reduction in most HF patients, criteria for BP reduction were baseline systolic (SBP) and diastolic (DBP) blood pressures greater than 120 and/or 80 mmHg, respectively. Baseline mean arterial pressures (MAP) were calculated from the SBPs and DBPs transmitted over the first 5 days of the monitoring period. Baseline MAPs were then compared to end MAPs (calculated from the last 5 days of a monitoring period). Patients requiring blood pressure reduction at study onset demonstrated a significant reduction in MAP; baseline MAP among these patients was 99.8 mmHg and end MAP was 93.5 mmHg (p < 0.001), providing further evidence of clinical effectiveness. See figure below for change in mean arterial pressure (MAP) from baseline (beginning of home monitoring) to discharge from monitoring system. Data is only for the patients who required blood pressure lowering.

Figure 5

With the exception of the APN who focused on patient education, provider time was most related to responding to alarming data. In response to aberrant data for any parameter, telemetry technicians telephoned patients 1,183 times; the APN was consulted 1,601 times, and telephoned patients 1,390 times to assess any changes in status and reinforce dietary and medication compliance; cardiology fellows were consulted 527 times and telephoned patients on 152 occasions; attending cardiologists were consulted on 298 alarms and telephoned patients regarding 42 alarms. Medications were adjusted in response to 186 alarms, a clinic visit was scheduled following 31 alarms, and patients were directed to an emergency room in twelve instances, which were followed by nine hospitalizations for HF decompensation. Efficiency was demonstrated by the relatively minimal amount of time and cost involved, compared to the outcomes achieved. Provider time (minutes/patient/day) and total method costs (cost/patient/day) are represented in figures 3 and 4 below.

Figure 6

{image:6}

Safety was evaluated by adverse events in relation to an unanswered alarm. No adverse events were reported in any patient’s electronic record or are known to have occurred as a result of an unanswered alarm or technical malfunction. The latter was assessed by verifying patient/family phone calls and auditing traditional medical records.

Patient satisfaction was evaluated at the end of each monitoring period with a satisfaction questionnaire, scoring items on a 1-5 Likert scale, with 5 rating the highest level of satisfaction. Satisfaction questions asked patients to rate their perceived satisfaction with: the overall quality of the TM method, the convenience of the service, the professional manner of the TM staff in general, and particularly the telemetry and installation technicians, and also their satisfaction with having care provided over the phone by the APN. Two overall questions were also included regarding any perceived interference in daily activities do to the operation of the device or calls received from health care providers. These were scored with 1 representing very interfering and 5 indicative of no interference. A mean satisfaction score of > 4.75 was recorded for 95% of the sample.