Using telemonitoring to construct knowledge about homebound patient populations: vital-signs alert rates

Abstract

Three home health agencies conducted daily telemonitoring of patients in western Montana. The agencies all used monitoring equipment of the same type, which provided up to six vital-signs measurements (heart rate, oxygen saturation, systolic blood pressure, diastolic blood pressure, glucose and bodyweight). There were 337 patients in all, two-thirds of whom were female. These patients were monitored for a total of 16,999 person-days. The rate of occurrence of any vital-sign measurement falling outside acceptable ranges was 33.8 per patient per 60-day period. The highest alert rate for a specific vital sign was for decreased SpO₂ (6.4 per patient per 60-day period). The central station nurse took follow-up action at a rate of 26.0 per patient per 60-day period; case manager nurses took follow-up action at a rate of 8.2 per patient per 60-day period. There were some differences between male and female patients in the alert rates, and between the agencies. The alert rates measured in the present study are expected to be useful to home care administrators in estimating the staffing requirements for telemonitoring.

Introduction

Telemonitoring is often used to monitor a patient's vital signs at home. The goal is to improve the health status or recovery of patients. The patient follows instructions from the home monitoring equipment for recording the vital signs, and the monitor transmits the measurements to a central station at the health-care agency. The vital signs that are commonly measured are heart rate, oxygen saturation, systolic blood pressure, diastolic blood pressure, glucose and bodyweight. When monitoring begins, the patient's doctor decides whether to use standardized ranges or patient-specific upper and lower limits. The latter allow closer monitoring of a particular vital sign where tighter control is needed, or they can avoid unnecessary action when tighter control is not warranted.

If a vital-sign measurement falls outside the range of acceptable values, then a ‘red alert’ is generated at the central monitoring station. A nurse at the central station then decides what action is required. Follow-up actions will commonly be taken by the case manager nurse, the patient/caregiver, the physician, the emergency department or the hospital.

We have examined the red alerts and the follow-up actions in patients being tele-monitored. This information is expected to be useful to planners, especially to those in health-care agencies in rural settings.

Methods

We used data from underserved and tribal patients in five contiguous counties in western Montana. The study was approved by the appropriate ethics committee. Montana has ageing and impoverished subpopulations, a large area and low population density. It therefore presents challenges to home health agencies.

Three home health agencies provided data for the study (Table 1). Agency 1 was an independent agency affiliated with two local hospitals, Agency 2 was a hospital-based home health agency and Agency 3 was an independent agency with a county affiliation. The agencies were not-for-profit. Agencies 1 and 3 provided services to populations where there were major medical centres within about 10 km. Agency 2 provided services to a population whose nearest medical centre was 90 km away.

Table 1

Study characteristics

	Agency 1	Agency 2	Agency 3	Total
Number of monitors available	30	15	20	65
Average age of the patients monitored (years)	78	78	77	78
Number of patients monitored	146	99	92	337
% female patients monitored	66	66	71	67
Days monitored (males)	3537	1179	1063	5779
Days monitored (females)	4540	1966	4714	11,220
Total days monitored	8077	3145	5777	16,999
% of total days monitoring females	56	63	82	66

Equipment

The monitoring equipment consisted of a PC with blood pressure cuffs, SpO₂ finger probes, bodyweight scales and (when required) glucose monitors. The agencies all used monitoring equipment of the same type (Honeywell HomMed LLC). Other studies have reported the accuracy and reliability of measurements made with these devices.^{1
–3}

A nurse set up the equipment in the patient's home and instructed the patient how to use it. The monitoring equipment was connected to the home telephone line so that the data could be transmitted automatically to the central station. If the patient failed to transmit expected data, the central station nurse contacted the patient, either by telephone or, if necessary, by making a home visit.

Central station

Vital-signs data were collected each day at the central station. The data were collected at a time agreed with the patient, ideally early enough in the day that any alerts could be dealt with in a timely manner. There was one nurse dedicated to work at the central station. This nurse monitored the station periodically and provided information to the patient's case manager nurse and physician as required, and recorded follow-up actions.

Alerts

Proprietary software at the central station provided an alert when a vital sign fell outside the acceptable range. The participating agencies agreed on acceptable ranges for the vital signs, which were:

Heart rate 55–100 /min;

Systolic blood pressure 90–180 mmHg;

Diastolic blood pressure 50–100 mmHg;

SpO₂ 89–100%;

Glucose 80–200 mg/dL (4.4–11.1 mmol/L).

In about 10% of cases, the patient's physician decided that different ranges should be used, depending on the needs of that patient. The bodyweight limits were always individualized to each patient.

Patients

The study population comprised all patients in the three study agencies who met the inclusion criteria and were telemonitored daily in their home during the period October 2007 to September 2008, inclusive. The inclusion criteria were: (1) the patient qualified for skilled nursing services; (2) the patient had home health services with Medicare or Medicaid as the primary payer; (3) the patient lived in a home with electricity and an ordinary telephone; (4) the patient had not previously used telemonitoring; and (5) the patient and/or caregiver had the necessary cognitive ability to perform telemonitoring and manage the patient's disease state in the home.

Episode of care

An episode of care is the unit of payment for home care services paid by Medicare. Services are paid by a single episode of care every 60 days. We therefore calculated alert rates for a 60-day episode of care.

Alert rates

Alerts could occur for heart rate, oxygen saturation, systolic blood pressure, diastolic blood pressure, glucose or bodyweight. The alert rate, R, for a particular vital sign during the study period was defined as:

where n = total number of alerts recorded during the study period for the particular vital sign and p = total number of person-days of monitoring during the study period for all the study patients.

That is, R was the number of alerts for the vital sign in question that would be expected from a person under care who was monitored continuously for a 60-day period.

The overall alert rate (i.e. for any of the signs) was the sum of the six vital sign-specific alert rates.

Follow-up rates

Follow-up actions after an alert could be taken by the central station nurse, case manager nurse, patient or caregiver, physician, emergency department or hospital. The responder-specific follow-up rate for the study period, F, was defined as:

where a = total number of follow-up actions by the particular type of responder during the study period and p = total number of person-days of monitoring, as above.

That is, F was the number of follow-up actions by the relevant type of responder that would be expected in relation to a person under care who was monitored continuously for a 60-day period.

The overall follow-up rate (i.e. for any type of responder) was the sum of the six responder-specific follow-up rates.

Data

Patient reports could be generated using the proprietary central station software. We used data from two such reports:

Patient compliance report. This report provided the number of days monitored;

Tabular trends report. This report provided the number of alerts for each vital sign and the number of follow-ups for each type of responder.

The number of alerts for a patient could be greater than the number of days monitored because the home monitors could provide more than one set of vital-signs readings per day and because the patient could have an alert for more than one type of vital sign on any day.

We obtained data for males and females separately within each agency. Since there was a higher proportion of women (65%) than men receiving home health services, we investigated whether there were gender differences in the rates.

Results

There were 337 patients in all, two-thirds of whom were female (Table 1). Of the patients studied, 44% received home health services for 25 days or less, 26% for 25–50 days and 30% for longer than 50 days.

Alert rates

The rate of occurrence across both genders and all three agencies of any vital-sign measurement falling outside the limits was 33.8 per patient per 60-day period (see Table 2). Thus over a 60-day episode of care, patients in the study population could expect to generate about 34 alerts (so-called abnormal readings) for one or more types of vital sign. This represents one alert per patient every two days. Among the different vital signs, the highest alert rate was for decreased SpO₂ (6.4 per person per 60 days).

Table 2

Red alert rates (per person per 60 days)

Alert			Agency 1	Agency 2	Agency 3	Total
Heart rate	Increased	Male	5.46	7.79	4.35	5.73
		Female	5.75	6.99	4.39	5.40
		Total	5.62	7.29	4.38	5.51
	Decreased	Male	0.76	2.04	1.02	1.07
		Female	1.08	2.66	2.02	1.75
		Total	0.94	2.42	1.84	1.52
SpO₂	Increased	Male	0.08	0.00	0.00	0.05
		Female	0.05	0.00	0.00	0.02
		Total	0.07	0.00	0.00	0.03
	Decreased	Male	6.40	10.79	3.67	6.79
		Female	7.41	4.36	5.60	6.12
		Total	6.97	6.77	5.24	6.35
Systolic BP	Increased	Male	0.17	4.43	4.06	1.75
		Female	2.21	9.80	4.16	4.36
		Total	1.31	7.78	4.14	3.47
	Decreased	Male	3.43	3.46	3.33	3.42
		Female	0.78	0.61	2.18	1.34
		Total	1.94	1.68	2.39	2.04
Diastolic BP	Increased	Male	0.34	1.98	2.65	1.10
		Female	0.56	3.45	2.25	1.78
		Total	0.46	2.90	2.33	1.55
	Decreased	Male	4.02	14.15	1.13	5.55
		Female	1.88	8.39	3.05	3.51
		Total	2.82	10.55	2.70	4.21
Glucose	Increased	Male	0.39	0.00	0.00	0.24
		Female	1.14	0.00	1.32	1.02
		Total	0.81	0.00	1.08	0.75
	Decreased	Male	0.17	0.00	0.00	0.10
		Female	0.17	0.00	0.00	0.07
		Total	0.17	0.00	0.00	0.08
Weight	Increased	Male	3.48	5.45	4.01	3.98
		Female	3.05	2.47	4.34	3.49
		Total	3.24	3.59	4.28	3.66
	Decreased	Male	5.38	2.54	4.23	4.59
		Female	3.52	4.73	5.68	4.64
		Total	4.33	3.91	5.41	4.62
Overall (i.e. any of the vital signs)		Male	30.08	52.63	28.45	34.37
		Female	27.60	43.46	34.99	33.50
		Total	28.68	46.89	33.79	33.79

Follow-up rates

The rate of occurrence across both genders and all three agencies of follow-up by any responder was 42.0 per patient per 60-day period (see Table 3). This represents one follow-up per patient every two out of three days. There were more follow-ups on average than red alerts, because one alert often generated multiple follow-ups; also the central station nurse sometimes identified a trend not producing an alert, or the provider requested a follow-up. Each alert or adverse trend identified triggered a follow-up action by the central station nurse, who might then contact the case manager nurse, patient or caregiver, physician, hospital emergency room, and/or other provider. Among the different responders, the highest follow-up rate was for central station nurse (26.0 per person per 60 days).

Table 3

Follow-up rates (per person per 60 days)

		Agency 1	Agency 2	Agency 3	Total
Central station nurse	Male	21.7	34.2	25.1	24.9
	Female	22.4	30.1	29.3	26.6
	Total	22.1	31.6	28.5	26.0
Case manager nurse	Male	6.0	18.8	5.1	8.4
	Female	6.0	18.1	6.0	8.1
	Total	6.0	18.4	5.9	8.2
Patient or caregiver	Male	4.3	3.4	6.9	4.6
	Female	3.7	3.2	8.4	5.6
	Total	3.9	3.3	8.1	5.2
Physician	Male	2.4	1.1	2.5	2.2
	Female	2.6	0.7	3.6	2.7
	Total	2.5	0.9	3.4	2.5
Emergency department	Male	0.07	0.00	0.11	0.06
	Female	0.01	0.03	0.08	0.04
	Total	0.04	0.02	0.08	0.05
Hospital	Male	0.03	0.00	0.00	0.02
	Female	0.00	0.00	0.05	0.02
	Total	0.00	0.00	0.04	0.02
Overall (i.e., any of the 6 responders)	Male	34.5	57.5	39.7	40.2
	Female	34.7	52.1	47.4	43.1
	Total	34.5	54.2	46.0	42.0

Gender differences

There were few gender differences in the calculated rates. For the alerts, we only observed differences in blood pressure. First, men had more decreased systolic and diastolic blood pressure alerts in comparison with women (3.4 versus 1.3 for systolic and 5.6 versus 3.5 for diastolic, each a difference of over 2.0 per person per 60 days). Clinically, low blood pressure alerts indicate possible low perfusion or systemic lack of circulation, or a possible adverse response to a prescribed medication. Second, women had more increased systolic blood pressure alerts than men (4.4 versus 1.8, a difference of over 2.6 per person per 60 days). Clinically, high blood pressure alerts indicate possible atherosclerosis, claudication, increased vascular resistance and/or a response to stress or fatigue. There were only minor gender differences in the follow-up rates. The rate of follow-up by the central station nurse for female patients was slightly higher than for males (26.6 versus 24.9); the absolute values of all the other rate differences were lower.

Agency differences

There were some differences between agencies in alert rates or follow-up action rates. Rates for increased diastolic blood pressure (0.5 versus 2.9 versus 2.3), decreased diastolic blood pressure (2.8 versus 10.6 versus 2.7) and increased systolic blood pressure (1.3 versus 7.8 versus 4.1), had large coefficients of variation (greater than or equal to 0.5). Agency 2 had much higher rates for decreased diastolic and increased systolic blood pressures, and Agency 1 had lower rates for increased diastolic and increased systolic blood pressures. Each of the three agencies served different populations, and these differences in blood pressure probably reflected the differences in their patients. For Agencies 1 and 2, most of the follow-up actions beyond the central station nurse were with the case manager nurse (6.0 and 18.4), whereas in Agency 3, most beyond the central station nurse were with the patient/caregiver (8.1). Agency 2 had a lower rate for physicians (0.9 compared with 2.5 and 3.4), possibly because of their distance from a major medical centre.

Gender and agency differences

Between the genders and agencies, males in Agency 2 had both the highest alert rate (52.6 per person per 60 days) and the highest follow-up rate (57.5 per person per 60 days). Agency 2 rates were zero for both increased and decreased glucose, for both males and females, because this agency did not telemonitor patients for glucose.

Discussion

This is the first time, to our knowledge, that alert data from home telemonitoring have been reported in detail. Access to care has typically been measured by using numbers of patients who have telemonitoring equipment placed in their homes. Increases in these numbers have been used to indicate improved access to health care. Our study demonstrates that access can also be measured by rates of alerts, and rates of follow-ups to alerts. Such rates have the potential to document the levels at which problems can be detected early and addressed (e.g. through changes in medication), which may improve or at least prevent worsening of patients' conditions. In our study, the central station nurse responded to every red alert. In this sense, the patients accessed care each time they produced a red alert. This is therefore a more sensitive measure of access to care than the mere presence of equipment in the home. We also noted that there was more access to care than the red alert rates indicated. For example, the central station nurse was trained to observe vital-sign trends and sometimes intervened with the patient even before a red alert occurred.

We observed different patterns of alerts within agencies and genders, which in turn reflected differences in patient populations or differences in practice patterns. The observed rates were highest for compromised oxygen carrying capacity indicated by SpO₂, followed closely by elevated heart rate. They were also relatively high for both under- and overweight and, less commonly, for decreased diastolic blood pressure. The study showed that most of the follow-up actions were by the central station nurse, as expected. The central station nurse is the health-care provider assigned to interpret the alert and to decide what should be done. On first receiving an alert, the central station nurse would normally call the patient and talk to him or her, provide instruction as needed, ask the patient to take repeat readings on the telemonitor, and then contact the case manager nurse only if necessary. The central station nurse's follow-up actions often avoided the need for action by other providers.

The present study was limited to homebound patients reimbursed under the Medicare Prospective Payment System (PPS) and Medicaid fee-for-service. Patients who were not homebound, but could benefit from home telemonitoring for long-term disease management, were not included in the study. Our study populations only included patients who lived in homes that had landlines. As mobile phones become more common, there are more homes without landlines. Telemonitoring was impossible if patients had no landline for the necessary modem.

The present study was subject to certain sources of bias. One source was physician and nurse reluctance to use home telemonitoring. In particular, there was some physician resistance to the greatly increased quantity of patient data provided by telemonitoring. These concerns became less frequent as the study progressed, with increased staff training and programme flexibility. Another source of bias was patient refusal to use the technology. Patient education was used successfully to mitigate these fears. The technology of home monitoring itself also produces limitations. Telemonitoring equipment does not usually allow continuous monitoring of vital signs. Consequently, alert rates could be biased by factors such as the time of day that data are collected.

Alert rates are expected to be useful to home care administrators in estimating the staffing requirements for telemonitoring. Estimating appropriate staffing levels when planning new telemonitoring services may be difficult in agencies with no prior experience. The alert data from the present study provide information to estimate the staffing requirements necessary to follow up on alerts.

Footnotes

Acknowledgements

Parts of this paper were presented at the meeting of the American Public Health Association in Philadelphia on 10 November 2009, and at the conference on Technology, Knowledge and Society in Berlin on 15–17 January 2010. The work was funded by a grant (H2ATH07759) from the Office for the Advancement of Telehealth, Health Resources and Services Administration, Department of Health and Human Services.

References

Paré

, Jaana

, Sicotte

. Systematic review of home telemonitoring for chronic diseases: the evidence base. J Am Med Inform Assoc 2007;14:269–77

Abdoh

, Krousel-Wood

, Re

. Accuracy of telemedicine in detecting uncontrolled hypertension and its impact on patient management. Telemed J E Health 2003;9:315–23

Artinian

, Washington

, Klymko

, Marbury

, Miller

, Powell

. What you need to know about home blood pressure telemonitoring: but may not know to ask. Home Healthc Nurse 2004;22:680–6