Do Automated Calls with Nurse Follow-up
Improve Self-Care and Glycemic Control among
Vulnerable Patients with Diabetes?
John D. Piette, PhD, Morris Weinberger, PhD, Stephen J. McPhee, MD, Connie A. Mah, BA,
Fredric B. Kraemer, MD, Lawrence M. Crapo, MD, PhD
PURPOSE: We sought to evaluate the effect of automated telephone assessment and self-care education calls with nurse
follow-up on the management of diabetes.
SUBJECTS AND METHODS: We enrolled 280 English- or
Spanish-speaking adults with diabetes who were using hypoglycemic medications and who were treated in a county health care
system. Patients were randomly assigned to usual care or to
receive an intervention that consisted of usual care plus biweekly automated assessment and self-care education calls with
telephone follow-up by a nurse educator. Outcomes measured
at 12 months included survey-reported self-care, perceived glycemic control, and symptoms, as well as glycosylated hemoglobin (Hb A1c) and serum glucose levels.
RESULTS: We collected follow-up data for 89% of enrollees
(248 patients). Compared with usual care patients, intervention
patients reported more frequent glucose monitoring, foot in-
spection, and weight monitoring, and fewer problems with
medication adherence (all P #0.03). Follow-up Hb A1c levels
were 0.3% lower in the intervention group (P 5 0.1), and
about twice as many intervention patients had Hb A1c levels
within the normal range (P 5 0.04). Serum glucose levels were
41 mg/dL lower among intervention patients than usual care
patients (P 5 0.002). Intervention patients also reported better glycemic control (P 5 0.005) and fewer diabetic symptoms
(P ,0.0001), including fewer symptoms of hyperglycemia and
hypoglycemia.
CONCLUSIONS: Automated calls with telephone nurse follow-up may be an effective strategy for improving self-care behavior and glycemic control, and for decreasing symptoms
among vulnerable patients with diabetes. Am J Med. 2000;108:
20 –27. q2000 by Excerpta Medica, Inc.
R
munication (7). Spanish-speaking patients who are limited to English-only services have poorer outcomes than
those who receive language-appropriate care (8).
Because telephones are almost universally available
(9,10), clinicians can use them to conduct health status
assessments and provide self-care education for patients
who have difficulty obtaining outpatient care. As part of a
chronic disease management strategy, telephone services
can improve outcomes (11,12) and decrease treatment
costs (13). Telephone-supported diabetes care improves
glycemic control (14,15). However, such programs can
be labor-intensive and costly, and frequently are inaccessible to non-English speakers.
Automated calling systems represent a pragmatic and
inexpensive way to improve telephone care. These systems use specialized computer technology to deliver messages and collect information from patients using their
touch-tone keypad or voice-response technology. Automated telephone calls are acceptable to vulnerable patients, including non-English-speakers (16), and can be
an effective way to monitor patients (17–20) and to promote behavior change (21–24).
We conducted a randomized, controlled trial of automated telephone assessment and self-care education calls
with nurse follow-up among diabetic patients treated in a
public health care system to determine whether this intervention could improve self-care, glycemic control, and
symptoms.
egular assessments of blood glucose levels and effective self-care may improve glycemic control,
thereby reducing the risk of complications from
diabetes (1– 4). Unfortunately, many patients fall short of
targeted glucose levels because of problems obtaining
treatment or inadequate self-care (5). In particular, lowincome patients with diabetes use fewer outpatient services and have more hospitalizations than those with
higher incomes (6), and patients who cannot speak English have difficulty achieving improvements in glycemic
control that accompany effective provider-patient com-
From the Center for Health Care Evaluation/HSR&D Field Program
(JDP, CAM) VA Palo Alto Health Care System, Palo Alto, California;
Department of Health Research and Policy (JDP), Stanford University,
Stanford, California; Roudenbush VA Medical Center (MW), Indiana
University School of Medicine, and Regenstrief Institute for Health
Care, Indianapolis, Indiana; Department of Medicine (SJM), University
of California, San Francisco, California; VA Palo Alto Health Care System (FBK), Palo Alto, California, and Stanford University School of
Medicine, Stanford, California; Department of Endocrinology (LMC),
Santa Clara Valley Medical Center, San Jose, California.
Supported by the Clinical Research Grants Program of the American
Diabetes Association and by the Health Services Research and Development Service and Mental Health Strategic Health Group, Department
of Veterans Affairs.
Requests for reprints should be addressed to John D. Piette, PhD,
Center for Health Care Evaluation, VA Palo Alto Health Care System,
Menlo Park Division (152), 795 Willow Road, Menlo Park, California
94025.
Manuscript submitted February 8, 1999 and accepted in revised form
July 22, 1999.
20
q2000 by Excerpta Medica, Inc.
All rights reserved.
0002-9343/00/$–see front matter
PII S0002-9343(99)00298-3
Impact of Automated Telephone Care with Nurse Follow-up/Piette et al
METHODS
Patient Enrollment
Participants were enrolled from two general medicine
clinics of a county health care system. Research assistants
reviewed medical records of patients with scheduled appointments to identify adults with a diagnosis of diabetes
mellitus or an active prescription for a hypoglycemic
agent. We excluded patients who were .75 years of age,
had a diagnosed psychotic disorder, disabling sensory impairment, or life expectancy of ,12 months, or whose
primary language was neither English nor Spanish. For
patients with some fluency in both languages, we identified their primary language using an established scale
(25). Potentially eligible patients were interviewed to exclude patients who controlled their glucose levels without
hypoglycemic medication, were newly diagnosed (,6
months), planned to discontinue receiving services from
the clinic within the 12-month follow-up period, or did
not have a touch-tone telephone. Informed consent procedures were conducted according to a protocol approved by the Institutional Review Boards at the medical
center and Stanford University. Randomization was
based on a table of randomly permuted numbers (26).
Neither providers, research staff, nor prospective participants had knowledge of group assignment until the patient had consented to participate.
Description of the Intervention
The core of the intervention was a series of automated
telephone assessments designed to identify patients with
health and self-care problems (27). These assessments
were used to focus the efforts of a diabetes nurse educator
on patients experiencing the greatest problems. The automated telephone calls also were used to deliver targeted
and tailored self-care education messages.
Automated telephone calls. The automated calling
component of the intervention was developed and implemented using a Teleminder Model IV automated telephone messaging computer (Decision Systems, Los Altos, California). The calls consisted of hierarchically
structured messages composed of statements and queries
recorded in a human voice. All calls were outbound (ie, to
the patients) and were placed at the times patients indicated were most convenient for them. Using up to six
attempted calls, we sought to determine patients’ health
status biweekly with a 5- to 8-minute assessment. Patients
interacted with the system using their touch-tone keypad;
responses were stored and determined the subsequent
content of the message. Each query included a check for
invalid responses and a prompt instructing patients to
correct out-of-range values. Intervention patients received no training other than the general description provided as part of the informed consent process and a onepage summary provided in their enrollment packets.
During each assessment, patients reported information about self-monitored blood glucose readings, selfcare, perceived glycemic control, and symptoms of poor
glycemic control, foot problems, chest pain, and breathing problems. Assessments periodically included additional questions addressing issues unlikely to change on a
biweekly basis (eg, whether the patient had a retinal examination in the prior year). At the end of each assessment, patients were given the option to listen to a randomly cycling diabetes “health tip.” Health tips were 30
to 60 seconds in length and were based on literature published by the Centers for Disease Control and Prevention
and the American Diabetes Association. Patients also
were given the option to participate in a 3- to 5-minute
interactive self-care education module focusing on diet
and weight control.
After participating in the intervention for several
months, patients were offered additional automated selfcare education calls that focused on glucose self-monitoring, foot care, and medication adherence. Within these
calls, patients reported specific barriers to self-care (eg,
no glucose monitoring strips) and received tailored education and advice. Within the medication adherence segment of the calls, patients were asked about their adherence to insulin, oral hypoglycemic medications, antihypertensive medications, and antilipidemic medications.
For each type of medication, patients without adherence
problems received positive feedback and reinforcement.
Patients reporting less than optimal adherence were
asked about specific barriers and were given advice about
overcoming each barrier.
Telephone nurse follow-up. Each week, the automated
assessment system generated reports organized according
to the urgency of reported problems, and the nurse
used these reports to prioritize patient contacts. During
follow-up calls, the nurse addressed problems reported
during the assessments and provided more general selfcare education. The nurse was located outside of the clinics and had neither face-to-face contact with patients nor
ready access to their medical records. She had access only
to medical record data that were abstracted at enrollment,
automated assessment reports, and her notes from prior
telephone contacts.
The nurse also made periodic calls to follow up on
issues discussed in a prior week or check on patients who
rarely responded to the automated calls. A small number
of contacts was initiated by patients using a toll-free telephone number. Depending on the nature and acuity of
patients’ problems, the nurse contacted their primary
care physician via fax, voicemail, or pager.
Spanish translation. We used a standard translation
procedure to produce Spanish-language versions of the
automated telephone messages (28). A bilingual, bicultural, first-generation Mexican-American translated the
January 2000
THE AMERICAN JOURNAL OF MEDICINEt
Volume 108 21
Impact of Automated Telephone Care with Nurse Follow-up/Piette et al
messages into Spanish, and a second bilingual speaker
independently translated them back into English. Differences between the original English version and the backtranslation were resolved through discussions among the
two translators and the principal investigator. Before using the messages in the trial, they were pilot tested with
nonenrolled Spanish-speaking patients. The intervention
nurse was competent, but not fluent, in conversational
Spanish.
Usual Care
Patients assigned to the usual care control group had no
systematic monitoring between clinic visits or reminders
of upcoming clinic appointments. Providers used their
discretion to schedule follow-up visits. Additional visits
were scheduled at the patient’s initiative. Although a telephone triage nurse, a diabetes education clinic, and an
interpreter service were available at the study site, demand for these services often exceeded capacity.
Data Collection and Measures
Survey data. At baseline and 12 months, trained interviewers surveyed patients over the telephone in their native language. During the survey, it was not possible to
ensure that interviewers were blinded to a patient’s group
assignment.
For glucose self-monitoring, foot inspection, and
weight monitoring, we measured self-care using a 5-point
Likert scale (0 5 “never” to 5 5 “daily”). Patients were
considered to have a problem with medication adherence
if they reported that they “sometimes forget to take their
medication,” “sometimes stop taking their medication
when they feel better,” or “sometimes stop taking their
medication when they feel worse.” Patients reported their
perceived glycemic control using a 5-point scale ranging
from 1 5 “poor” to 5 5 “excellent.” Although an imperfect measure of glucose control, greater perceived control
has a consistent association with lower glycosylated hemoglobin (Hb A1c) and serum glucose levels, and fewer
diabetes-related symptoms (29). In addition, patients’
perceptions of glycemic control often reflect both their
average control as well as their experience with periodic
episodes of hyperglycemia and hypoglycemia. Perceived
glycemic control may capture the difference between actual control of glucose levels and patient-specific goals
that vary owing to patient age, comorbid conditions, and
physician practice styles. Finally, perceived glycemic control is correlated with other important outcomes, such as
health-related quality of life, in ways that Hb A1c and
serum glucose levels are not (30,31).
During their interviews, patients reported whether
they experienced each of 22 diabetes-related symptoms in
the prior week, including symptoms of hyperglycemia
(eg, “frequent urination at night”), hypoglycemia (eg,
“shakiness or weakness”), vascular problems (eg, “pain in
22
January 2000
THE AMERICAN JOURNAL OF MEDICINEt Volume 108
the calf muscles when walking”), or other problems (eg,
“painful urination”).
Medical record and laboratory data. At enrollment, we
measured patients’ height and weight wearing light clothing and calculated their body mass index (kg/m2). Sociodemographic data were obtained at baseline from medical records and screening interviews. We abstracted information about active prescriptions, diabetes-related
complications, and comorbid chronic diseases from
medical records. Glycosylated hemoglobin and serum
glucose levels were measured blindly at baseline and 12
months in a single laboratory. We used formulae provided by the manufacturer of the glycosylated hemoglobin assay to convert values to Hb A1c units, with a normal
range of 4.7% to 6.4%. We identified a priori both the
mean Hb A1c level and the proportion of patients within
the normal range as outcomes.
Measures of health service use. Because it was unlikely
that we would observe a decrease in resource use associated with improved glycemic control during the 12month observation period, health service utilization was
not a primary outcome. However, we monitored patients’ inpatient and outpatient utilization to examine
whether the intervention increased the appropriate use of
preventive services such as retinal exams. Information
about inpatient admissions was collected from the medical center’s administrative databases and corroborated
by patients’ self-reports. We also were able to use these
databases to identify podiatry clinic, ophthalmology
clinic, and emergency department visits. We used patients’ responses to survey questions to measure outpatient service use within the 6 months before their baseline
and 12-month interviews.
Statistical Analysis
We used Student’s t test, the chi-square test, and the Wilcoxon rank sum test to compare the baseline characteristics of intervention and usual care patients. Outcome
analyses were conducted on an intent-to-treat basis, and
all P values were two-tailed. Despite randomization, the
intervention and usual care groups were not equivalent at
baseline (eg, in terms of insulin use). To adjust for these
differences, as well as for baseline values of endpoint measures, we used multivariate regression models, including
ordinary least-squares regression for continuous outcomes and Likert scale scores, logistic regression for binary outcomes, and Poisson regression for symptom
counts. The adjusted effects, with 95% confidence intervals, are presented.
RESULTS
Of the 588 patients identified as potentially eligible, 46
patients were excluded at the request of their physician,
Impact of Automated Telephone Care with Nurse Follow-up/Piette et al
Table 1. Sociodemographic and Clinical Characteristics of Intervention and Usual Care Patients
at Enrollment
Sociodemographic characteristics
Age (years)
Female gender
Race
White
Hispanic
Other
Married
Living alone
Income ,$10,000/year
Spanish speaking
Body mass index (kg/m2)
Number of diabetic complications
Number of comorbidities
Self-care
Glucose monitoring*
Foot inspection*
Weight monitoring*
Any medication problem
Glycemic control
Hemoglobin A1c level (percent)
Normal hemoglobin A1c level†
Serum glucose (mg/dL)
Insulin
Self-reported control‡
Diabetic symptoms, median (interquartile range)
All symptoms
Hyperglycemic symptoms
Hypoglycemic symptoms
Vascular symptoms
Other symptoms
Intervention
(n 5 124)
Usual Care
(n 5 124)
56 6 10
76 (61)
53 6 10
70 (56)
36 (29)
59 (47)
29 (23)
36 (29)
22 (18)
73 (59)
30 (24)
34 6 9
1 (0–1)
1 (1–3)
36 (29)
64 (52)
24 (19)
40 (32)
18 (15)
71 (57)
34 (27)
33 6 8
1 (0–1)
1 (0–2)
0.6
0.5
0.7
0.6
0.6
0.6
0.03
3.7 6 1.7
3.8 6 1.5
1.8 6 1.6
69 (56)
3.5 6 1.8
4.2 6 1.3
1.3 6 1.4
69 (56)
0.3
0.03
0.007
1.00
8.8 6 1.8
10 (8)
233 6 84
54 (44)
2.7 6 1.1
8.6 6 1.8
9 (7)
221 6 106
38 (31)
2.8 6 1.1
0.3
0.8
0.4
0.04
0.5
5 (2–9)
2 (1–4)
1 (0–3)
1 (0–1)
1 (0–1)
5 (3–9)
2 (1–4)
2 (0–3)
1 (0–2)
1 (0–1)
P Value
0.07
0.4
0.7
0.7
0.6
0.4
0.5
0.9
Data shown as number (percent), mean 6 SD, or median (interquartile range).
* 0 5 never, 1 5 ,monthly, 2 5 monthly, 3 5 weekly, 4 5 almost daily, 5 5 daily.
† Less than 6.4%.
‡ 1 5 poor, 2 5 fair, 3 5 good, 4 5 very good, 5 5 excellent.
148 declined participation, and 114 were not enrolled for
some other reason (eg, they left the clinic before they
could be approached). The remaining 280 patients were
enrolled. Compared with patients who were potentially
eligible but not enrolled, enrollees were somewhat more
likely to be female (51% versus 59%, P 5 0.04) and
somewhat younger (mean age [6 SD] 57 6 10 years versus 55 6 10 years, P ,0.01).
We collected outcome data at 12 months for 89% of
enrollees (248 patients). Equal numbers of intervention
and usual care patients failed to complete the study because they moved out of the area (7 patients in each
group) or because they refused follow-up data collection
(4 patients in each group). Fewer patients in the intervention group than in the usual care group died during the
study (1 versus 4), and fewer were lost to follow-up (1
versus 4). In most respects, baseline characteristics of in-
tervention and usual care patients were similar (Table 1).
However, patients in the intervention group had more
comorbidities, less frequent foot inspections, and more
frequent weight monitoring. They were also more likely
to be using insulin at baseline (44% versus 31%, P 5
0.04).
Dose of the Intervention
Intervention patients completed an average of 1.4 automated assessment and self-care education calls each
month (Table 2). Half of the patients completed 78% or
more of their attempted assessments (interquartile range
44% to 91%). During completed assessments, patients
reported their self-monitored blood glucose values an average of 80% of the time (median 86%, interquartile
range 45% to 94%).
On average, patients had 6 minutes of nurse telephone
contact per month. Most nurse contacts (72%) included
January 2000
THE AMERICAN JOURNAL OF MEDICINEt
Volume 108 23
Impact of Automated Telephone Care with Nurse Follow-up/Piette et al
Table 2. Automated Telephone and Nurse Contact over the Year among Intervention Patients
Automated telephone contacts
Number of contacts of all types
Number of assessment calls
Home glucose readings
Health tip selections
Dietary module selections
Number of self-care education calls
Telephone nurse contacts
Contacts of all types
Total number of episodes
Total time (minutes)
Automated assessment follow-up
Total number of episodes
Total time (minutes)
Other nurse-initiated contacts
Total number of episodes
Total time (minutes)
Patient-initiated contacts
Total number of episodes
Total time (minutes)
some discussion directly related to patients’ glycemic
control. Hypoglycemic medications (eg, adherence problems and side effects) were discussed in 45% of contacts,
and glucose self-monitoring was discussed in 57% of contacts. In addition, nondiabetes medications were discussed in 32% of contacts, nondiabetic symptoms were
discussed in 49% of contacts, and psychological problems, such as symptoms of depression and anxiety, were
discussed in 18% of contacts.
Intervention Effects
In unadjusted analyses, intervention patients reported
significantly better glucose self-monitoring, foot inspection, and weight monitoring at follow-up than did usual
care patients (P #0.01, Table 3). Intervention patients
also were substantially less likely to report problems with
medication adherence (P 5 0.003). Adjusting for baseline values and insulin use had no appreciable effect on
the differences in self-care between the two groups. After
adjustment for baseline differences, the intervention decreased the proportion of patients with medication adherence problems by 21% (from 69% to 48%, P 5
0.003).
Intervention patients had minimally lower Hb A1c levels (0.1%) at follow-up than usual care patients. Adjustment for baseline levels and insulin use increased the
magnitude of the estimated intervention effect to 0.3%
(P 5 0.1, Figure). In adjusted analyses, the intervention
increased the proportion of patients with normal Hb A1c
levels by 9% (from 8% normal in the usual care group to
17% normal in the intervention group, P 5 0.04). The
24
January 2000
THE AMERICAN JOURNAL OF MEDICINEt
Volume 108
Total
Contacts
Per Patient Contacts
(Mean 6 SD)
2,123
1,631
1,098
761
574
492
17 6 12
13 6 8
968
666
566
465
695
8,616
664
70 6 13
319
4,233
363
34 6 46
301
3,592
262
29 6 31
75
791
161
6 6 13
intervention decreased serum glucose levels by 41 mg/dL
(P 5 0.002), and improved patients’ self-reported glycemic control (P 5 0.005). Among the patients using
insulin at baseline, 50 (93%) of 54 intervention patients
were still using insulin at follow-up compared with all 38
of the usual care patients. In addition, 2 of the 70 intervention patients and 6 of the 86 usual care patients who
were using oral hypoglycemic agents at baseline were using insulin at follow-up.
In adjusted analyses (Table 3), symptoms of all types
were less frequent at follow-up among intervention patients than among usual care patients. The greatest differences were in reported hyperglycemic and hypoglycemic
symptoms.
During follow-up, 24% of intervention patients and
23% of usual care patients were hospitalized (P 5 0.9);
48% of intervention patients and 40% of usual care patients were seen in the emergency department (P 5 0.2).
Similar proportions of intervention (22%) and usual care
(26%) patients were seen in podiatry clinics (P 5 0.5),
and slightly more intervention (49%) than usual care
(41%) patients were seen in ophthalmology clinics (P 5
0.2). Intervention patients reported an average of three
diabetes-related outpatient visits and six visits of all types
during the final 6 months of follow-up, virtually identical
to those among patients in the usual care group. Adjustment for prior visit rates and insulin use had no appreciable effect on the between-group comparison.
Subgroup analyses. Despite the small number of Spanish-speaking patients in the study, we observed signifi-
Impact of Automated Telephone Care with Nurse Follow-up/Piette et al
Table 3. Diabetes Self-Care, Glycemic Control, and Symptoms at Follow-Up among Intervention (n 5 124) and Control
(n 5 124) Patients*
Unadjusted Outcomes
Self-Care
Glucose monitoring†
Foot inspection†
Weight monitoring†
Any medication problem
Glycemic Control
Hemoglobin A1c level
(percent)
Normal hemoglobin A1c
level‡
Serum glucose (mg/dL)
Self-reported control§
Diabetic Symptoms¶
All symptoms
Hyperglycemic symptoms
Hypoglycemic symptoms
Vascular symptoms
Other symptoms
Outcomes Adjusted for Baseline Values and Insulin Use
P
Usual
Value Intervention Care
P
Value
Difference
95%
Confidence
Interval
0.03
0.02
0.001
0.003
0.4
0.3
0.6
221
0.04 to 0.7
0.1 to 0.6
0.1 to 1.0
234 to 27
20.7 to 0.1
Intervention
Usual
Care
4.2 6 1.4
4.7 6 0.7
2.1 6 1.7
55 (44)
3.7 6 1.7
4.3 6 1.3
1.6 6 1.6
78 (63)
0.01
0.006
0.008
0.003
8.2 6 1.9
8.3 6 1.9
0.8
25 (20)
14 (11)
0.06
181 6 68
3.1 6 1.1
220 6 110
2.8 6 1.1
4 (1–7)
1 (0–3)
1 (0–2)
1 (0–1)
0 (0–1)
6 (3–10)
2 (1–4)
2 (1–3)
1 (0–2)
1 (0–1)
4.1
4.7
2.1
48
3.7
4.4
1.6
69
8.1
8.4
0.1
20.3
17
8
0.04
9
7 to 30
0.009
0.05
180
3.1
221
2.7
0.002
0.005
241
0.4
267 to 215
0.1 to 0.6
0.001
0.003
0.002
0.07
0.11
4.0
1.6
1.1
0.9
1.0
21.4
20.7
20.5
20.2
20.3
21.8 to 21.0
21.9 to 20.4
20.7 to 20.2
20.4 to 0.1
20.6 to 20.0
5.4 ,0.0001
2.3
0.0005
1.6
0.001
1.1
0.2
1.3
0.07
* Unadjusted outcomes are reported as mean 6 SD and adjusted outcomes as means, or unadjusted outcomes are number (percent) and adjusted
outcomes are percent.
† 0 5 never, 1 5 ,monthly, 2 5 monthly, 3 5 weekly, 4 5 almost daily, 5 5 daily.
‡ Less than 6.4%.
§ 1 5 poor, 2 5 fair, 3 5 good, 4 5 very good, 5 5 excellent.
¶ Unadjusted outcomes are median count (interquartile range); adjusted outcomes are counts.
cant effects of the intervention on their glycemic control.
The average endpoint Hb A1c level among Spanishspeakers in the intervention group was 1.1% lower (95%
confidence interval [CI] 0.2% to 1.9%) than among those
in the usual care group, and six times as many intervention patients had normal Hb A1c levels (18% versus 3%,
P 5 0.05). In addition, the mean serum glucose level at
the end of follow-up was 71 mg/dL lower (95% CI 13 to
129 mg/dL) among patients in the intervention group,
and intervention patients had 1.6 fewer symptoms (95%
CI 0.0 to 3.2) than controls.
Among patients who reported at baseline that no physician had examined their feet in the prior 6 months (n 5
77), those in the intervention group were somewhat more
Figure. Mean glycosylated hemoglobin (Hb A1c) levels among intervention and usual care patients. Adjusted follow-up means were
calculated adjusting for baseline glycosylated hemoglobin levels and insulin use (any versus none).
January 2000
THE AMERICAN JOURNAL OF MEDICINEt
Volume 108 25
Impact of Automated Telephone Care with Nurse Follow-up/Piette et al
likely to be seen in podiatry clinics during the study
(19%) than those in the usual care group (7%, P 5 0.1).
Among patients who reported during their baseline survey that they had not had their eyes checked in the prior 6
months (n 5 145), 57% of intervention patients were
seen in ophthalmology clinics during the study compared
with 37% of usual care patients (P 5 0.02).
DISCUSSION
The results of this randomized, controlled trial suggest
that automated telephone assessment and self-care education calls with nurse follow-up improved patients’ selfcare and glycemic control, and decreased their symptom
burden. These improvements were achieved with an average of less than 6 minutes per month of nurse-patient
contact. Through automated telephone assessments, the
nurse was able to use time more judiciously, focusing on
the patients who most needed assistance. The automated
calling system also allowed patients to access languageappropriate self-care information and behavioral reinforcement.
Some of the greatest effects of the intervention were in
the areas of self-reported glucose self-monitoring, foot
inspection, and weight monitoring. Self-monitoring of
glucose levels is an important component of diabetes selfcare (32–34), although there have been only a few randomized trials of its effects on glycemic control among
patients with type 2 diabetes (35). However, regular glucose self-monitoring is important for some patients, and
many others may benefit during periods when they are at
risk for poor control (eg, soon after a change in medication or during times of intercurrent illness). The benefits
are most likely to be realized if self-monitoring is linked
either to patient education or to medical intervention (eg,
medication adjustment). The results of this study suggest
that automated assessments with follow-up nurse telephone calls may be a useful strategy for linking self-monitoring to a clinical response.
Improved foot self-care results in fewer serious foot
lesions and amputations (36,37). As with glucose selfmonitoring, improvement in patients’ foot self-care has
the greatest effect on outcomes when it is linked to a provider-based intervention (38). The intervention evaluated in this study may provide a way to ensure that such
linkages are possible.
We are not aware of any studies that have examined the
independent effect of weight self-monitoring on health
outcomes. However, behavioral programs that include
weight self-monitoring can be moderately effective in decreasing obesity (39), and improvements for some patients have been large enough to have a long-term impact
on glycemic control (40). With the advent of new drug
therapies for obesity, weight self-monitoring may take on
increased importance (41).
26
January 2000
THE AMERICAN JOURNAL OF MEDICINEt
Volume 108
One of the unique contributions of this study is its
implementation within a public health care system in
which many patients have low incomes, little formal education, and psychosocial problems that complicate their
diabetes care. Interventions such as the one we evaluated
may improve public providers’ ability to serve more patients using fewer staff. It also is encouraging that this
system appeared to be effective in improving outcomes
among Spanish-speaking patients, whose outcomes often
are especially poor. Circumventing language barriers may
be one way that interventions such as this one may improve patient outcomes.
Several limitations of this study must be considered. It
was conducted at a single site, and telephone nursing services were provided by a single nurse. We are now performing a similar trial in a Department of Veterans Affairs health care system. Finally, many of the outcomes in
this study were self-reported, and patients in the intervention group may have reported more favorable outcomes than actually occurred.
This study does not suggest that automated patient
surveillance and education can replace clinical vigilance
or the provider-patient relationship that is central to diabetes care. Automated systems are a way to augment
service delivery in primary care. At the heart of the encounter between a patient and a clinician is the ability to
uncover problems that may have gone unrecognized by
patients and unanticipated by even the most sophisticated automated assessment algorithm (42). The provider-patient relationship itself is therapeutic, and lack of
such a relationship can lead to poor adherence and dissatisfaction with care (43). With these caveats, we conclude that diabetes care supported by automated telephone assessments and patient education may be an effective means of improving vulnerable patients’ self-care
and glycemic control, and decreasing their symptoms.
ACKNOWLEDGMENTS
We thank Eunice Coeter and the clinicians and patients of the
Moorpark Clinics, Santa Clara Valley Medical Center; Carol
Gangitano, MSN, for her contribution as intervention nurse;
and Edgar O. Alvarez, Dara J. Amboy, and Vivian Schiedler for
assistance with translation, data collection, and data processing.
REFERENCES
1. The Diabetes Control and Complications Trial Research Group.
The effect of intensive treatment of diabetes on the development
and progression of long-term complications in insulin-dependent
diabetes mellitus. NEJM. 1993;329:977–986.
2. The Diabetes Control and Complications Trial Research Group.
Lifetime benefits and costs of intensive therapy as practiced in the
diabetes control and complications trial. JAMA. 1997;278:25.
3. UK Prospective Diabetes Study (UKPDS) Group. Intensive bloodglucose control with sulphonylureas or insulin compared with con-
Impact of Automated Telephone Care with Nurse Follow-up/Piette et al
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
ventional treatment and risk of complications in patients with type
2 diabetes (UKPDS 33). Lancet. 1998;352:837– 853.
UK Prospective Diabetes Study (UKPDS) Group. Effect of intensive blood-glucose control with metformin on complications in
overweight patients with type 2 diabetes (UKPDS 34). Lancet. 1998;
352:854 – 865.
Kurtz SM. Adherence to diabetes regimens: empirical status and
clinical implications. Diabetes Educ. 1990;16:50 –59.
Bindman AB, Grumbach K, Osmond D, et al. Preventable hospitalizations and access to health care. JAMA. 1995;274:305–311.
Greenfield S, Kaplan SH, Ware JE, et al. Patients’ participation in
medical care: effects on blood sugar control and quality of life in
diabetes. J Gen Intern Med. 1988;3:448 – 457.
Perez-Stable EJ, Napoles-Springer A, Miramontes JM. The effects
of ethnicity and language on medical outcomes of patients with
hypertension or diabetes. Med Care. 1997;35:1212–1219.
Anderson JE, Nelson DE, Wilson RW. Telephone coverage and
measurement of health risk indicators: data from the National
Health Interview Survey. Am J Public Health. 1998;88:1392–1395.
Phoneless in America. (Statistical Brief). Washington, DC: US Bureau of the Census; 1994:94 –16.
Weinberger M, Tierney WM, Booher P, Katz BP. Can the provision
of information to patients with osteoarthritis improve functional
status? A randomized, controlled trial. Arthritis Rheum. 1989;32:
1577–1583.
Debusk RF, Houston Miller N, Superko HR, et al. A case-management system for coronary risk factor modification after acute myocardial infarction. Ann Intern Med. 1994;120:721–729.
Wasson J, Gaudette C, Whaley F, et al. Telephone care as a substitute for routine clinic follow-up. JAMA. 1992;267:1788 –1793.
Weinberger M, Kirkman MS, Samsa GP, et al. A nurse-coordinated
intervention for primary care patients with non-insulin-dependent
diabetes mellitus: impact on glycemic control and health-related
quality of life. J Gen Intern Med. 1995;10:59 – 66.
Aubert RE, Herman WH, Waters J, et al. Nurse case management to
improve glycemic control in diabetic patients in a health maintenance organization. A randomized, controlled trial. Ann Intern
Med. 1998;129:605– 612.
Tanke ED, Leirer VO. Automated telephone reminders in tuberculosis care. Med Care. 1994;32:380 –389.
Christ G, Siegel K. Monitoring quality-of-life needs of cancer patients. Cancer. 1990;65:760 –765.
Searles JS, Perrine MW, Mundt JC, et al. Self-report of drinking
using touch-tone telephone: extending the limits of reliable daily
contact. J Stud Alcohol. 1995;56:375–382.
Mahoney D, Tennstedt S, Friedman R, Heeren T. An automated
telephone system for monitoring the functional status of community-residing elders. Gerontologist. 1999;39:229 –234.
Piette JD, Mah CA. The feasibility of automated voice messaging as
an adjunct to diabetes outpatient care. Diabetes Care. 1997;20:15–
21.
Leirer VO, Morrow DG, Pariante G, et al. Increasing influenza vaccination adherence through voice mail. J Am Geriatr Soc. 1989;37:
1147–1150.
Friedman RH. Automated telephone conversations to assess health
behavior and deliver behavioral interventions. J Med Syst. 1998;22:
95–102.
23. Dini EF, Linkins RW, Chaney M. Effectiveness of computer-generated telephone messages in increasing clinic visits. Arch Pediatr
Adolesc Med. 1995;149:902–905.
24. Leirer VO, Morrow DG, Tanke ED, et al. Elders’ nonadherence: its
assessment and medication reminding by voice mail. Gerontologist.
1991;31:514 –520.
25. Marin G, Sabogal F, Marin BV, et al. Development of a short acculturation scale for Hispanics. Hispanic J Behav Sci. 1987;9:183–205.
26. Fleiss JL. The Design and Analysis of Clinical Experiments. New
York: Wiley & Sons; 1986:390 – 416.
27. Piette JD. Moving diabetes management from clinic to community:
development of a prototype based on automated voice messaging.
Diabetes Educ. 1997;23:672– 679.
28. Brislin RW, Lonner WJ, Thorndike EM. Cross-Cultural Research
Methods. New York: Wiley & Sons; 1973.
29. Piette JD, McPhee SJ, Weinberger M, et al. Use of automated telephone disease management calls in an ethnically diverse sample of
low-income patients with diabetes. Diabetes Care. 1999;22:1302–
1309.
30. Weinberger M, Kirkman MS, Samsa GP, et al. The relationship
between glycemic control and health-related quality of life in patients with non-insulin-dependent diabetes mellitus. Med Care.
1994;32:1173–1181.
31. The Diabetes Control and Complications Trial Research Group.
Influence of intensive diabetes treatment on quality-of-life outcomes in the Diabetes Control and Complications Trial. Diabetes
Care. 1996;19:195–203.
32. Butler RN, Rubenstein AH, Gracia AG, Zweig SC. Type 2 diabetes:
patient education and home glucose monitoring. Geriatrics. 1998;
53:60 – 67.
33. Consensus Panel Statement. Self-monitoring of blood glucose. Diabetes Care. 1996;19(suppl 1):S62–S66.
34. Walker EA, Cypress ML. Self-monitoring: the patient-practitioner
alliance. Nurse Pract Forum. 1991;2:175–177.
35. Faas A, Schellevis FG, Van Eijk JTM. The efficacy of self-monitoring
of blood glucose in NIDDM subjects: a criteria-based literature
review. Diabetes Care. 1997;20:1482–1486.
36. Malone JM, Snyder M, Anderson G, et al. Prevention of amputation
by diabetic education. Am J Surg. 1989;158:520 –524.
37. Pieber TR, Holler A, Siebenhofer A, et al. Evaluation of a structured
teaching and treatment program for type 2 diabetes in general practice in a rural area of Austria. Diabet Med. 1995;12:349 –354.
38. Mayfield JA, Reiber GE, Sanders LJ, et al. Preventive foot care in
people with diabetes. Diabetes Care. 1998;21:2161–2177.
39. Maggio CA, Pi-Sunyer FX. The prevention and treatment of
obesity: application to type 2 diabetes. Diabetes Care. 1997;20:
1744 –1766.
40. Wing RR, Koeske R, Epstein LH, et al. Long-term effects of modest
weight loss in type 2 diabetic patients. Arch Intern Med. 1987;147:
1749 –1753.
41. Hollander PA, Elbein SC, Hirsch IB, et al. Role of Orlistat in the
treatment of obese patients with type 2 diabetes. A 1-year randomized double-blind study. Diabetes Care. 1998;21:1288 –1294.
42. Naylor CD. Grey zones of clinical practice: some limits to evidencebased medicine. Lancet. 1995;345:840 – 842.
43. Leopold N, Cooper J, Clancy C. Sustained partnership in primary
care. J Fam Pract. 1996;42:129 –137.
January 2000
THE AMERICAN JOURNAL OF MEDICINEt
Volume 108 27