PDA-assisted low glycemic index dietary intervention for type II diabetes: a pilot study

ORIGINAL ARTICLE

PDA-assisted low glycemic index dietaryintervention for type II diabetes: a pilot study

Y Ma1, BC Olendzki1, D Chiriboga1, M Rosal1, E Sinagra1, S Crawford1, AR Hafner1, SL Pagoto1,

RP Magner1 and IS Ockene2

1Division of Preventive and Behavioral Medicine, Department of Medicine, University of Massachusetts Medical School, Worcester, MA,USA and 2Division of Cardiovascular Medicine, University of Massachusetts Medical School, Worcester, MA, USA

Background: Epidemiological and dietary intervention studies suggest that a low-glycemic index (GI) diet is beneficial for bloodglucose control; however, long-term clinical utility of the low GI diet has not been fully investigated.Objectives: To evaluate the feasibility and efficacy of a nutritionist-delivered low-GI dietary intervention, with the support of apersonal digital assistant (PDA), for adult patients with poorly controlled type II diabetes.Method: The low-GI intervention consisted of six counseling sessions and the use of a PDA-based food database with GI scoresfor 6 months. Study outcomes included feasibility measures, glycosylated hemoglobin levels (HbA1c), GI and glycemic load (GL)score of self-reported dietary intake, body weight, depression and quality of life (QOL). Measures were obtained at baseline, 3and 6 months.Results: Of 31 adult patients approached, 15 met study eligibility criteria and were enrolled in the study. Thirteen patients(87%) completed all study assessments. Findings included decreases in average HbA1c (�0.5% P¼ 0.02), body weight, hipcircumference, blood pressure, dietary GI and daily caloric intake. Diabetes impact scores also decreased. All but one participantcompleted all components of the intervention. There were mixed reports regarding the usefulness of the PDAs; however,participants offered helpful suggestions for further development.Conclusions: Results of this pilot study support the feasibility of implementing a nutritionist-delivered, PDA-assisted low-GIdietary intervention for patients with poorly controlled type II diabetes. Encouraging initial efficacy data require further testing inthe context of a randomized clinical trial.

European Journal of Clinical Nutrition (2006) 60, 1235–1243. doi:10.1038/sj.ejcn.1602443; published online 17 May 2006

Keywords: diabetes mellitus; glycemic index; diet; PDA

Introduction

The glycemic index (GI) is based on a ranking of carbohy-

drate-containing foods according to the rate at which they

raise blood glucose levels after eating (Jenkins et al., 1981). A

recent meta-analysis of randomized clinical trials (RCTs)

(Brand-Miller et al., 2003a) suggests that choosing low-GI

foods has a modest but clinically useful effect on medium-

term glycemic control in patients with type II diabetes.

However, in most of the published RCTs, patients were fed

experimental diets, leaving controversy over the applicabi-

lity of GI-based diets in the clinical setting for the manage-

ment of diabetes. The American Diabetes Association (ADA)

states that evidence is lacking to support a change in dietary

recommendations which are based on carbohydrate count-

ing (Franz, 2003). One of the barriers to implementation of a

low-GI diet among diabetic patients is the relative complex-

ity involved in ascertaining the GI score of each food or

meal; whereas carbohydrate counting is much more familiar,

and available on every label or through averages from the

exchange list (Franz, 2003). Carbohydrate counting and use

of a low-GI diet are not mutually exclusive; thus, integrating

a low-GI diet with carbohydrate counting may be optimal for

glycemic control.Received 5 October 2005; revised 27 January 2006; accepted 16 February

2006; published online 17 May 2006

Correspondence: Dr Y Ma, Division of Preventive and Behavioral Medicine,

Department of Medicine, University of Massachusetts Medical School, 55 Lake

Avenue North, Worcester, MA 01655, USA.

E-mail: [email protected]

Guarantor: Y Ma.

Contributors: Study concept and design: YM, BCO, DC and ISO; statistical

analysis: ARH, YM and SC; drafting of the manuscripts: YM and BCO; critical

revision of the mannuuscript for important intellectual contents: DC, MR, ES,

SC, ARH, SLP, RPM and ISO.

European Journal of Clinical Nutrition (2006) 60, 1235–1243& 2006 Nature Publishing Group All rights reserved 0954-3007/06 $30.00

www.nature.com/ejcn

Advances in the technology of personal digital assistants

(PDAs) including smaller size, user friendliness, familiarity,

easily modifiable software, and lower cost have contributed

to their increasing use in psychological and medical studies

since the early 1990s. PDAs have successfully assisted self-

monitoring in studies of diabetes (Clarke et al., 1995),

respiratory diseases (Tiplady et al., 1997), panic disorder

(Taylor et al., 1990), smoking cessation (Shiffman et al.,

1996), effects of night shift work on psychological symptoms

during the menstrual cycle (Totterdell et al., 1995) and

menstrual cycling (Johannes et al., 2000). PDAs might also

reduce the burden of GI self-monitoring in the management

of diabetes.

The objectives of this study are to evaluate the feasibility

of implementation and explore the potential efficacy of a

nutritionist-delivered low-GI dietary intervention in a

sample of patients diagnosed with type II diabetes, which

was poorly controlled. Outcomes of interest included patient

completion of intervention sessions, relevant physiological

variables (glycosylated hemoglobin (HbA1c), blood pressure

and weight), dietary intake and quality of life (QOL)

measures.

Subjects and methods

Subject recruitment and enrollment

The study protocol was approved by the University of

Massachusetts (UMass) Medical School Institutional Review

Board for use of human subjects in medical research. All

study participants provided informed consent before being

enrolled into the study.

UMass Memorial Health Care patients with a diagnosis of

type II diabetes were recruited between July and October

2004 through referrals from a network of participating

health-care providers. Eligibility criteria included: having a

diagnosis of type II diabetes and having an HbA1c level of

7.0% or greater. We did not include fasting glucose levels as

an inclusion criterion, as we were interested in the long-term

control of blood glucose which is reflected in HbA1c.

Exclusion criteria included: planning to become pregnant

or to relocate during the study period, being pregnant,

inability to provide informed consent, being on glucocorti-

coid therapy within the previous 3 months, having experi-

enced an acute coronary event within the previous 6

months, having a diagnosis of a medical condition

that precluded adherence to study dietary recommenda-

tions, that is, Crohn’s disease, ulcerative colitis, end-stage

renal disease, diagnosis of a severe psychiatric illness,

significant visual impairment or currently following a low-

carbohydrate, high-fat diet such as the Atkins Diet (Atkins,

1998). Because the Atkins diet is low in all sources of

carbohydrates, modification of the type of carbohydrate

(GI) will have limited effect on glycemic load (GL) and

therefore HbA1c.

Study design and intervention format

This was a pilot study with a single treatment arm. The goal

of the intervention was to improve HbA1c levels via a low-GI

diet that was customized to the participant’s taste prefer-

ences and eating habits. The 6-month nutritionist-delivered

low-GI dietary intervention consisted of two group sessions,

one individual session and three telephone counseling

sessions. All participants received a Palm Zire 21 PDA to

use for the duration of the study.

The initial 2.5 h group session was held in a demonstration

kitchen by a trained nutritionist and involved information

on low-GI diet. Factors that influence the GI value of a food

(i.e., acid, particle size, fiber content and cooking time), how

to integrate a low-GI diet into one’s life, and low-GI

substitutions for common high-GI foods were discussed.

The group then sampled low-GI foods, were provided with

low-GI recipes and given hands-on training in the use of the

PDA. At the subsequent 1-h individual visit with a nutri-

tionist 2 weeks later, low-GI diet information was reviewed

and patient-centered counseling was provided to facilitate

the following behavior change strategies: goal-setting, self-

monitoring and problem-solving to address each patient’s

adherence barriers (to diet or use of the PDA). At the

individual visit, participants’ PDA data were uploaded to our

study database; these data were used by the nutritionist to

guide the nutritional counseling. The group reconvened for

a 2-h grocery store tour at week 4. Subsequent 30-min

telephone counseling sessions with the nutritionist were

completed at 2-, 3- and 5- month time points. These sessions

addressed the progress towards goals and challenges to

adherence.

Low-GI diet and PDA GI database

Patients were trained to locate and record individual foods in

their diets on the PDA. The Glycemic Index Meal Planner

software was provided by the Glycemic Diet Software, LLC

(http://www.glycemicdietsw.com). Study investigators pro-

vided the software company with nutrient and GI values for

food equivalent to those from the International Table of GI

(Foster-Powell et al., 2002), as well as those on 400 additional

foods that are commonly consumed by Americans (Ma et al.,

2005, 2006). The Meal Planner can be accessed easily from

the PDA home screen. Once in the program, the patient can

choose breakfast, lunch, dinner, or snack as the meal type,

and then select the food group from a list. Foods in the

selected group will be displayed, along with the GI score or

calories per serving. The ranges for high (469), medium (56–

69), and low (o56) GI values are displayed to assist the

patient in choosing foods (Foster-Powell et al., 2002; Brand-

Miller et al., 2003b). The software calculates the overall GI for

individual meals and for the entire day. It also provides

carbohydrate, calorie, protein, fat, fiber and sodium intake

analysis by meal and by day. The program has the capacity to

store up to 30 days of data, and data can be uploaded to

external databases (via cradle). Patients can view a log which

PDA-assisted dietary GI score and diabetesY Ma et al

1236

European Journal of Clinical Nutrition

displays the GI, GL, carbohydrate content and calories for

each of the last 30 dates on which they have entered meal

data. The software also allows patients to view data on meals

and individual foods eaten on each of the dates. At the

telephone counseling sessions, the nutritionist reviewed the

log with participants, and identified days with high GI.

Specific high-GI foods eaten on that day could be identified.

Assessment of study outcomes

The study data collection period was 6 months. Feasibility of

study implementation was assessed based on retention rate

and attendance records. Treatment acceptability was assessed

via questionnaire at the end of the study.

Data were collected at baseline, and then 3 and 6 months

following study enrollment and included: HbA1c, blood

pressure, body weight, height, waist and hip circumference

and medication use; behavioral measures such as dietary

intake and physical activity level; and psychosocial measures

including QOL and depression.

Fasting blood samples were used to assess HbA1c. Weight

and height were measured using a balance beam scale, with

participants wearing light clothing and no shoes. Blood

pressure measurements were taken using a Dinamap XL

automated blood pressure monitor (Sturrock et al., 1997),

which allows standardized measurements.

Dietary intake was assessed by the Seven-Day Diet Recall

(7DDR) (Hebert et al., 1997b). The 7DDR also included a brief

validated physical activity assessment, described elsewhere

(Matthews et al., 1997, 2000). The GI score was determined

from the intake data collected through the 7DDR using

published tables (Foster-Powell et al., 2002; Brand-Miller

et al., 2003b). GI and GL (GI of a food times the amount of

carbohydrate in grams eaten divided by 100) were also

calculated, as in previous reports (Ma et al., 2005, 2006).

Use of oral hypoglycemics, insulin, lipid-lowering and

antihypertensive medications was assessed using a ques-

tionnaire at each study visit (baseline, 3 and 6 months).

Depression was assessed using the Center for Epidemiolo-

gical Studies-Depression Scale (CES-D) (Radloff, 1977; Devins

and Orme, 1985) and QOL was assessed using the Diabetes

Impact Survey (Schwartz et al., 2005). Age, gender, educa-

tion, occupation, ethnicity, income data and diabetic

complications were collected by questionnaire at baseline.

Statistical analyses

Demographic data were summarized using mean (s.d.) or

frequency (%) as appropriate. For physiological, dietary,

physical activity and psychosocial variables, mean and

standard deviation (s.d.) were computed at baseline. We

conducted longitudinal analyses to evaluate within-partici-

pant change over time in physiological, dietary, physical

activity and psychosocial variables, using linear mixed

modeling (Littell et al., 1996). Consistent with our previous

longitudinal analyses of HbA1c (Rosal et al., 2005), we

employed compound symmetry for the within-subject

covariance structure, estimating subject-specific intercepts,

that is, subject was considered to be a random effect. Each

variable was modeled as a function of categorized time

(baseline, 3 and 6 months). Within-subject changes between

each of the three time points were estimated, to assess both

short-term and long-term changes, and stability between 3

and 6 months.

Results

Of the 31 patients approached for the study, 15 met

eligibility criteria and were enrolled. Major reasons for

ineligibility were HbA1co7 (eight subjects). We could not

assess eligibility for two subjects because of inability to

contact the patients. Two subjects indicated they were too

busy to participate, and one subject was excluded for each of

the following reasons: planned to have weight loss surgery,

moving out of the area within 6 months, had type I diabetes,

and already on low-GI diet. Of the subjects recruited, 15 were

enrolled in the study with one participant dropping out after

the first group session owing to job relocation. In addition,

one patient was excluded from the final analysis owing to a

diagnosis of advanced breast cancer during the study and

subsequent inability to adhere to the low-GI diet. Thirteen

patients completed the study and their data were used for

the analysis. Of these, all but one completed all components

of the study intervention (one subject missed two telephone

counseling sessions).

Participants’ characteristics are presented in Table 1. They

were predominately white, married and well-educated. The

average age was 58 years with a range of 39–72 years. Nine

men and four women participated. The majority worked

full-time and all but one had an annual household income

of $60 000 or more. Eighty-five percent of the participants

reported diabetic complications, the most common being

hypertension, neuropathy and heart disease. However, the

majority of participants characterized their health as ‘very

good’ or ‘good’. In addition, most (10 out of 13) of the

participants had first-degree relatives with diabetes. All

participants were being treated with diet and medication.

When asked at the beginning of the study if they thought

the PDA would be helpful in tracking their diet, all

participants answered ‘yes’.

The average baseline HbA1c was 8%. Most (nine out of 13)

participants’ HbA1c levels decreased over the course of the

intervention (Table 2). For the group as a whole, the decrease

in HbA1c was 0.5%, which is clinically significant. There was

a statistically significant decrease in HbA1c levels at both the

3-month (P¼0.02) and 6-month (P¼0.02) time points for

the overall group compared to baseline. HbA1c remained

stable between 3- and 6-month visits (P¼0.98).

Oral hypoglycemic medications at baseline visit included

metformin (used by five subjects or 38%), pioglitazone

(31%), glyburide (25%), rosiglitazone (15%), and glimepiride


1237


(8%). Five subjects (38%) were on insulin. Over the course of

the study, several participants reported medically supervised

changes in their diabetes treatment (Table 2). Two decreased

their insulin dosage, one discontinued an oral hypoglycemic

agent and one decreased oral hypoglycemic dosage. Two

participants switched oral hypoglycemic agents, one started

taking an additional oral hypoglycemic agent and one began

using insulin. When grouping the study results in a category

of ‘effective GI intervention’, meaning either improved in

HbA1c, associated with no change in medication or no

change in HbA1c associated with a decrease in medication

dosage (a total of seven subjects), the average decrease in

HbA1c in this subgroup was �5.1% at 3 months and �3.5%

at 6 months as compared to baseline. Four subjects in this

group (60%) decreased their medication dosage, or the

number of medications. On the other hand, in the group

where medication dosage was increased, and/or HbA1c

increased during the study period (n¼6), categorized as

‘not-effective GI intervention’, HbA1c increased by 4.1% at 3

months and 1.2% at 6 months, as compared to baseline.

Three of these patients (50%) required increased dosage of

medication or addition of a different medication.

Table 3 presents baseline as well as changes in physiolo-

gical, dietary, physical activity and psychosocial variables.

Baseline average BMI was 32.81 kg/m2 (s.d.¼5.87). Daily

caloric intake was 1994 kcal (s.d.¼669). GI was 81.96

(s.d.¼7.20). Forty-four percent of calories were from total

fat and 35% were from carbohydrate. Weight decreased by an

average of 1.8 kg over the course of the study but this was

marginally statistically significant owing to the small sample

size (P¼0.07). Systolic and diastolic blood pressure also

decreased over the 6-month period (�7.6 and �6.1 mm Hg

on average, respectively), the latter attaining statistical

significance. Hip circumference decreased significantly over

the 6 months for the group overall (�1.9 cm, P¼0.03);

however, waist circumference did not change significantly

during the study period. Most (nine out of 12) participants’

dietary GI decreased from baseline at 3 months but this

change did not reach statistical significance. However, when

we examined the change of GI at the individual level, we

noted that one subject’s GI score increased by 24 units at the

3-month time point, which affected the overall significance

of the change in GI score for the entire group; when we

excluded from the analysis the GI value from this subject,

the average GI decreased by 4 units for the group as a whole,

at 3 months (P¼0.05). Daily calorie intake also decreased,

with a mean reduction of 185 kcal at 6 months, but this

reduction was not statistically significant. Depression scores

were stable over the course of the study but a decrease in the

diabetes impact score was observed and was marginally

significant at 6 months (P¼0.09), indicating a better QOL.

Participants completed a questionnaire at the end of the

intervention to assess acceptability of the intervention. Most

participants (11 out of 13) reported that they liked the

intervention. Three reported that they had difficulty adher-

ing to the lower-GI diet, three were neutral and seven found

it easy, with all participants but one reporting that they liked

the low-GI diet. Opinions about the PDA database were

mixed. Eight (62%) participants reported that it was difficult

to use owing to inability to find all the foods they consumed,

Table 1 Baseline characteristics of participants in PDA pilot study

Count (%) or Mean (s.d.)

Demographic informationAge (years) 57.9 (10.4)Gender

Male 9 (69.2%)Female 4 (30.8%)

EducationHS diploma 2 (15.4%)Some college 2 (15.4%)Bachelor’s degree 4 (30.8%)Graduate or professional degree 5 (38.5%)

EthnicityWhite 12 (92.3%)Other 1 (7.7%)

Marital statusSingle 0Married or living with partner 10 (76.9%)Separated/divorced/widowed 3 (23.1%)

Work statusFull-time 8 (61.5%)Part-time 2 (15.4%)Disabled (unable to work) 1 (7.7%)Retired 1 (7.7%)

Health informationSmoking status

Current smoker 0 (0%)Former smoker; no longer smokes 6 (46.2%)Never smoked 7 (53.9%)

Family historyFirst-degree relatives with diabetes 10 (76.9%)

Diabetes treatmentDiet only 0Diet and medication 13 (100%)Exercise 5 (38.5%)

Diabetes medications usedOral hypoglycemics 10 (76.9%)Insulin 5 (38.5%)

Frequency of primary care visits0–2 times/year 2 (15.4%)2–4 times/year 10 (76.9%)More than 4 times/year 1 (7.7%)

Technology experienceUsed PDA before

Yes 6 (46.2%)No 7 (53.9%)

Access to personal computer at homeYes 10 (76.9%)No 3 (23.1%)


1238


tedium of continuously recording diet, or screen size too

small. Five (38%) participants reported that it was fairly easy

to use the PDA.

Discussion

This study provides encouraging data regarding the feasi-

bility of a nutritionist-delivered low-GI intervention, assisted

by a PDA food database, for patients with type II diabetes.

The results support the feasibility of implementing a fully

powered RCT to explore the impact of a PDA-assisted low-GI

intervention on glycemic control in type II diabetes. There

were mixed reports regarding the usefulness of the PDAs;

however, participants offered helpful suggestions for further

development.

Glycemic response is influenced by several factors includ-

ing the nature of the monosaccharides absorbed, the amount

of carbohydrate consumed, the rate of absorption, and

colonic fermentation (Wolever, 2003). In addition, several

other factors affect the glycemic response to carbohydrates,

including acid content (Ostman et al., 2005), the proportion

of certain nutritional components (fat, protein) in the mixed

meal, the physical state of a food item (e.g., whole vs

ground), the content of non-nutrients in foods like legumes

(e.g., phytates), and, perhaps most important, the overall

fiber content of a meal (Nuttall, 1993; Wursch and Pi-Sunyer,

1997). The mechanisms of the proposed beneficial effects of

dietary fiber and a low-GI diet on glucose control remain

unclear, but they may include delayed gastric emptying,

altered transit time in the small intestine, insulation of

carbohydrate from digestive enzymes, and digestive enzyme

inhibition. Reducing overall carbohydrate intake can in-

crease postprandial serum free-fatty acids, whereas low-GI

diets have been shown to both reduce serum free-fatty acids

and improve glycemic control (Wolever, 2003).

Results from our study are consistent with a recent meta-

analysis of RCTs of low-GI diets among type I and type II

diabetic patients suggesting that choosing low-GI foods in

place of high-GI foods has a modest but clinically useful

effect on medium-term glycemic control (Brand-Miller et al.,

2003a). Among type I diabetic patients (Weyman-Daum

Table 2 Subject-specific HbA1c changes and medication changes

Subject ID Baseline medications BaselineHbA1c

Change(3-mon. tobaseline)

Change(6-mon. tobaseline)

Change(6-mon. to

3-mos.)

Medication changes

4 Metformin 500 mg 2� /day 10.1 �1.5 0 1.5 Switched to glyburide at 3 months;began insulin glargine at 6 months

7 Not on medication 8.2 — �0.3 — No change8 Metformin 100 mg 2� /day,

pioglitazone 35 mg 1� /day7.8 0 0 0 No change

10 Metformin 8 mg 1� /day, glyburide32 mg 12/day

8.3 �0.1 0 0.1 Discontinued glyburide, switchedfrom metformin to rosiglitazone4 mg at 6 months

12 Pioglitazone 1� /day, metformin3� /day, insulin lispro 15–75 unitsat mealtime, insulin glargine 50units at night

7.5 �0.1 �1.3 �1.2 Decreased insulin lispro by 80% at 6months

13 Insulin lispro 15 units 1� /day,insulin glargine 50 units 1� /day

7.0 0.2 �0.5 �0.7 No change

14 Rosiglitazone 2 mg 2� /day 8.2 �0.5 �1.7 �1.2 Added metformin at 6 months15 Metformin 850 mg 2� /day 7.1 0 0.1 0.1 No change18 Rosiglitazone 4 mg 1� /day 7.3 �0.1 0.7 0.8 Non-compliant—dropped

medications and increased dietaryand GI intake

24 Glimepiride 3 mg 1� /day,pioglitazone 15 mg 1� /day

7.4 �0.9 �0.9 0 Halved glimepiride dosage at 6months

25 Glyburide 5 mg 2� /day 8.0 �1.1 �1.2 �0.1 Added metformin at 3 months29 Insulin glargine 70 units 1� /day 9.7 �1.6 �1.3 0.3 Lowered insulin to 60 units at 6

months31 Glipizide 2 mg 2� /day,

pioglitazone 1 mg 1� /day, insulinlispro 16 units 1� /day and 20 units1� /day

7.4 �0.2 0 0.2 No change

Mean (s.d.) 8.00 (0.95)Mean (s.e.) �0.49 (0.20) �0.49 (0.19) �0.004 (0.20)P-value* 0.02 0.02 0.98

*P-values were from a mixed model for comparison of differences of least-squares means. Model was fit using HbA1c as dependent variable, time as independent

variable and subject as random effect. The overall P-value for the model was 0.03.


1239


et al., 1987; Giacco et al., 2000; Gilbertson et al., 2003), a low-

GI diet resulted in significant decreases in postprandial

plasma glucose levels, improved glycemic control and

reduction in hypoglycemic events. Results were consistent

among type II diabetic patients (Jenkins et al., 1988, 2002;

Gulliford et al., 1989; Wolever et al., 1992; Jarvi et al., 1999;

Tsihlias et al., 2000; Heilbronn et al., 2002; Kabir et al., 2002;

Brynes et al., 2003; Gilbertson et al., 2003), where low-GI diet

improved plasma glucose control and had the added benefit

of significantly decreasing cholesterol.

Supporting the results from our study, a limited number of

clinical trials have demonstrated that low-GI dietary educa-

tional interventions delivered by a clinical nutritionist can

lead to improved metabolic control in patients with diabetes

(Frost et al., 1994; Gilbertson et al., 2001, 2003; Rizkalla et al.,

2004). These reports are consistent with our findings. For

example, in a study performed in Britain, 60 patients with

type II diabetes were randomized to either a standard advice

group or a low-GI advice group over a 12-week period (Frost

et al., 1994). Results demonstrated that patients in the low-

GI diet group were successful in decreasing the GI of the food

they consumed. Fasting blood glucose, fructosamine, total

cholesterol and triglyceride levels showed a significantly

greater decrease in the low-GI group as compared to those of

the patients in the standard advice group.

As in our study, PDAs were used to monitor diet and blood

glucose readings for 6 months in a pilot study with 20

diabetic patients (Tsang et al., 2001). The dietary records

were transmitted to a computerized diabetes monitoring

system, and patients received immediate dietary feedback.

Although no other intervention was involved, a significant

improvement in glycemic control was reported. The study

also found that 95% of the patients found PDAs easy to

operate and 63% found them useful.

Critics of low-GI diets for diabetes often suggest that the

concept of the GI is too complicated and restrictive to be of

practical importance (Franz et al., 2003). Another potential

limitation that is shared with carbohydrate counting is that,

if not carefully chosen, a low-GI diet can be high in saturated

fat and cholesterol, both of which can increase serum lipids

and the associated cardiovascular disease risk (Neff, 2003).

Nutrition counseling can help ensure that the patient will

choose a healthy and balanced low-GI diet, with results from

initial trials suggesting that a structured low-GI diet can be a

safe and usable tool for the management of diabetes,

regardless of the patients’ education levels (Frost et al.,

1994; Gilbertson et al., 2001, 2003).

PDAs have been used for gathering data in several studies

(Taylor et al., 1990; Clarke et al., 1995; Shiffman et al., 1996;

Tiplady et al., 1997; Johannes et al., 2000). PDAs are ideally

suited as an intervention tool for several reasons: they are

compact and portable, and for the most part self-explanatory

in their operation. PDA prices have decreased significantly,

leading to a wider use in the population. The use of the PDA

in our study also provided a channel for interaction between

the user and the provider. Whereas further development of

the software is ongoing, the option of immediate feedback

on GI programmed into the PDA proved to be quite useful to

Table 3 Within-subject changes in physiological, dietary, physical activity, and psychosocial variables in PDA study between baseline and 3-monthpoint, between baseline and 6-month point and between 3- and 6-month points

Baseline Change (3 months to baseline) Change (6 months to baseline) Change (6 months to 3 months)

Mean (s.d.) Mean (s.e.) P-value* Mean (s.e.) P-value* Mean (s.e.) P-value*

Physiological variablesWeight (kg) 95.79 (19.60) �0.68 (0.92) 0.47 �1.77 (0.95) 0.07 �1.08 (0.95) 0.26Hip circumference (cm) 114.69 (11.14) �1.58 (0.78) 0.05 �1.85 (0.81) 0.03 �0.26 (0.81) 0.75Waist circumference (cm) 109.02 (15.12) 2.21 (0.99) 0.04 1.82 (1.03) 0.91 �0.39 (1.03) 0.71Systolic blood pressure (mmHg) 133.00 (13.35) 5.04 (4.23) 0.25 �7.64 (4.23) 0.08 �12.68 (4.34) 0.008Diastolic blood pressure (mmHg) 81.008 (7.06) �2.35 (2.03) 0.26 �6.07 (2.03) 0.007 �3.72 (2.08) 0.09

Dietary variablesGlycemic index (GI) 81.96 (7.20) �1.15 (2.42) 0.64 �1.86 (2.49) 0.47 �0.71 (2.56) 0.78Glycemic load (GL) 155.87 (74.86) 10.66 (32.30) 0.74 �27.30 (33.24) 0.42 �37.96 (34.11) 0.28Daily caloric intake (Kcal) 1993.54 (668.49) 135.41 (324.74) 0.68 �184.85 (334.36) 0.59 �320.25 (343.10) 0.36Fat percentage 43.95 (5.71) �0.64 (2.14) 0.77 1.45 (2.20) 0.52 2.09 (2.26) 0.37Saturated fat percentage 14.37 (2.56) �0.44 (0.74) 0.56 0.52 (0.76) 0.50 0.96 (0.78) 0.23Total daily carbohydrate (g) 179.25 (83.89) 16.82 (38.34) 0.67 �23.98 (39.45) 0.55 �40.80 (40.45) 0.32Carbohydrate percentage 34.98 (5.77) 0.69 (0.30) 0.77 �0.29 (2.37) 0.90 �0.98 (2.43) 0.69Daily dietary fiber (g) 13.31 (6.04) 1.14 (2.70) 0.68 0.11 (2.79) 0.97 �1.03 (2.86) 0.72

Leisure physical activity (min/week) 176.82 (168.30) �16.01 (53.23) 0.77 �21.61 (53.24) 0.69 �5.60 (57.84) 0.92

Psychosocial variablesDepression score (CES-D) 6.38 (4.84) 0.44 (1.80) 0.81 0.50 (1.85) 0.79 0.07 (1.87) 0.97Diabetes impact score 51.77 (14.01) �2.58 (3.39) 0.45 �6.16 (3.50) 0.09 �3.57 (3.52) 0.32

*P-values were from a mixed model from which we estimated mean within-subject differences over time. Models were fit using each physiological, dietary, physical

activity and psychosocial variable as the dependent variable, time as the independent variable and subject as a random effect.


1240


participants. Use of the PDA GI software addresses some of

the perceived barriers (Franz, 2003) to the implementation of

a low-GI nutrition intervention, as users are provided with

readily accessible information on low-GI foods. As suggested

by Henker and Colleagues (Henker et al., 2002), PDAs have

great potential to add to the effectiveness of protocol-driven

treatment interventions.

We observed a decrease in HbA1c and a change in overall

GI score during the study. At the same time, we noted that

total carbohydrate intake in the group as a whole also

decreased from an average of 180 g/day at baseline to 140 g/

day at 6 months. Although the change did not achieve

statistical significance (P¼0.32), it could have contributed to

the decrease in HbA1c levels. In addition, GL decreased 27

points between baseline and 6 months, so the change in

HbA1c may also be related to the decrease in GL.

There are some weaknesses in our study, including lack of

control group, small sample size, limited generalizability

owing to population characteristics, need for refinement of

the PDA food database and software, and potential measure-

ment error of the GI score. First, owing to the pilot nature

and design of our study, we were unable to examine which

components (nutritionist or PDA) of the intervention were

most important in achieving the results. Given that there

was only one arm, our results could be explained by the

combination of the intervention, placebo effects, and

regression to the mean. Second, firm conclusions about

effectiveness cannot be derived from our data given the

small sample size and demographic characteristics of the

study population. The fact that changes were generally in a

consistently favorable direction, and that some changes were

large enough to be statistically significant, however, is

encouraging and indicates that a larger study is warranted.

Participants were primarily white, highly educated, above

average income levels and, although about half had no PDA

experience, most had very positive attitudes towards the

possibility that the PDAs would help them. Thus, general-

izability is limited. This pilot study provided important

information on efficacy, rather than effectiveness. We are

investigating this further in a more diverse population,

which is currently underway. Third, the PDA database and

software require further development. The PDA develop-

ment would include expansion of the database to increase

patient satisfaction with their ability to obtain information

on common foods, and an automated and personalized

feedback system, in as close to real time as possible. Finally,

structured assessment instruments, such as the 7DDR, used

in our pilot study, have been associated with a variety of

measurement errors (Hebert et al., 1997a, 2001, 2002) with

particular concern regarding the estimation of the GI.

Collapsing foods into groups, rather than using specific

foods and food quantities, could increase measurement error

of GI with the 7DDR. Therefore, several 24-h diet recalls,

which utilize the multiple pass recall methodology, may

generate greater accuracy in future studies; however, this is

more labor intensive.

Strengths of this pilot study include the fact that it was

conducted in the clinical, ‘real-world’ setting, and that it

attempted to address the perceived barriers to implementa-

tion of the GI score concept, using a technological device

that is gaining acceptability in the general population.

The Dietary Guidelines for Americans (Marlett et al., 2002;

US Department of Health and Human Services, 2005) stress

an increase in consumption of whole grains, dietary fiber,

fruit and vegetables, all of which are intrinsic to the low-GI

diet, but are not necessarily part of the carbohydrate

counting method. Optimal dietary management of diabetes

may involve a combination of the two: decreasing the GI

score of the diet, as well as the amount of carbohydrates

consumed per day.

Diet is a fundamental component in the management of

type II diabetes, but adherence to dietary recommendations

is a challenging issue. Refinements in instruments that foster

adherence are urgently needed. Technological devices such

as the PDAs to promote adherence to dietary interventions

via facilitation of food choices and real-time feedback appear

to improve diabetes control.

In conclusion, results of this pilot study support the

feasibility of implementing a nutritionist-delivered, PDA-

assisted, low-GI dietary intervention for patients with poorly

controlled type II diabetes. Randomized controlled trials

testing the efficacy of PDA-assisted low-GI diet on glycemic

control are justified. On the other hand, although partici-

pants’ perceptions regarding the practicality of the PDAs

were positive overall, a more comprehensive GI database

with ‘real-time’ personalized feedback is needed.

Acknowledgements

The project described was supported by an internal grant

from the University of Massachusetts Medical School to Dr

Yunsheng Ma. We thank Dr Judith K Ockene for her

consistent encouragement and support of this work; Mr

Hongzhen Zhang who helped with the innovative idea of

proactive control and management of human activities

through computer software; Mr Keith Crozier and Dr Frank

J Domino for the development of the PDA software; Ms Mary

Jane Campbell for her assistance with study recruitment and

data collection; Ms Victoria Andersen for her assistance with

the intervention; Ms Jennifer Griffith for her assistance in

creating GI database for PDA; and Ms Annie Culver, BS

Pharm for her assistance in understanding pharmacological

issues related to participants.

References

Atkins RC (1998). Dr. Atkins’ New Diet Revolution. Avon Books:New York.

Brand-Miller J, Hayne S, Petocz P, Colagiuri S (2003a). Low-glycemicindex diets in the management of diabetes: a meta-analysis ofrandomized controlled trials. Diabetes Care 26, 2261–2267.


1241


Brand-Miller J, Wolever TM, Foster-Powell K, Colagiuri S (2003b). TheNew Glucose Revolution: The Authoritative Guide to the GlycemicIndex—The Dietary Solution for Lifelong Health. Marlowe & Com-pany: New York, NY.

Brynes AE, Lee JL, Brighton RE, Leeds AR, Dornhorst A, Frost GS(2003). A low glycemic diet significantly improves the 24-h bloodglucose profile in people with type 2 diabetes, as assessed usingthe continuous glucose MiniMed monitor. Diabetes Care 26,548–549.

Clarke WL, Cox DJ, Gonder-Frederick LA, Julian D, Schlundt D,Polonsky W (1995). Reduced awareness of hypoglycemia in adultswith IDDM. A prospective study of hypoglycemic frequency andassociated symptoms. Diabetes Care 18, 517–522.

Devins G, Orme C (1985). Center for epidemiologic studiesdepression scale. Sweetland R (ed). Test Critiques. Test Corp ofAmerica, a subsidiary of Westport Publishers, Inc.: Kansas City.pp 144–160.

Foster-Powell K, Holt SH, Brand-Miller JC (2002). International tableof glycemic index and glycemic load values: 2002. Am J Clin Nutr76, 5–56.

Franz MJ (2003). Meta-analysis of low-glycemic index diets in themanagement of diabetes: response to brand—Miller et al. andMann. Diabetes Care 26, 3364–3365.

Franz MJ, Bantle JP, Beebe CA, Brunzell JD, Chiasson JL, Garg A et al.(2003). Evidence-based nutrition principles and recommendationsfor the treatment and prevention of diabetes and relatedcomplications. Diabetes Care 26, S51–S61.

Frost G, Wilding J, Beecham J (1994). Dietary advice based on theglycaemic index improves dietary profile and metabolic control intype 2 diabetic patients. Diabetes Med 11, 397–401.

Giacco R, Parillo M, Rivellese AA, Lasorella G, Giacco A, D’Episcopo Let al. (2000). Long-term dietary treatment with increased amountsof fiber-rich low-glycemic index natural foods improves bloodglucose control and reduces the number of hypoglycemic eventsin type 1 diabetic patients. Diabetes Care 23, 1461–1466.

Gilbertson HR, Brand-Miller JC, Thorburn AW, Evans S, Chondros P,Werther GA (2001). The effect of flexible low glycemic indexdietary advice versus measured carbohydrate exchange diets onglycemic control in children with type 1 diabetes. Diabetes Care24, 1137–1143.

Gilbertson HR, Thorburn AW, Brand-Miller JC, Chondros P, WertherGA (2003). Effect of low-glycemic-index dietary advice on dietaryquality and food choice in children with type 1 diabetes. Am J ClinNutr 77, 83–90.

Gulliford MC, Bicknell EJ, Scarpello JH (1989). Differential effect ofprotein and fat ingestion on blood glucose responses to high- andlow-glycemic-index carbohydrates in noninsulin-dependentdiabetic subjects. Am J Clin Nutr 50, 773–777.

Hebert J, Ebbeling C, Matthews C, Ma Y, Clemow L, Hurley T et al.(2002). Systematic errors in middle-aged women’s estimates ofenergy intake: comparing three self-report measures to totalenergy expenditure from doubly labeled water. Ann Epidemiol 12,577–586.

Hebert JR, Ma Y, Clemow L, Ockene IS, Saperia G, Stanek E et al.(1997a). Gender difference in social desirability and socialapproval bias in dietary self-report. Am J Epidemiol 146,1046–1055.

Hebert JR, Ockene IS, Hurley TG, Luippold R, Well AD, Harmatz MG(1997b). Development and testing of a seven-day dietary recall.J Clin Epidemiol 50, 925–937.

Hebert JR, Peterson KE, Hurley TG, Stoddard AM, Cohen N, Field AEet al. (2001). The effect of social desirability trait on self-reporteddietary measures among multi-ethnic female health centeremployees. Ann Epidemiol 11, 417–427.

Heilbronn LK, Noakes M, Clifton PM (2002). The effect of high- andlow-glycemic index energy restricted diets on plasma lipid andglucose profiles in type 2 diabetic subjects with varying glycemiccontrol. J Am Coll Nutr 21, 120–127.

Henker B, Whalen CK, Jamner LD, Delfino RJ (2002). Anxiety, affect,and activity in teenagers: monitoring daily life with electronicdiaries. J Am Acad Child Adolesc Psychiatry 41, 660–670.

Jarvi AE, Karlstrom BE, Granfeldt YE, Bjorck IE, Asp NG, Vessby BO(1999). Improved glycemic control and lipid profile and normal-ized fibrinolytic activity on a low-glycemic index diet in type 2diabetic patients. Diabetes Care 22, 10–18.

Jenkins AL, Jenkins DJ, Zdravkovic U, Wursch P, Vuksan V (2002).Depression of the glycemic index by high levels of beta-glucanfiber in two functional foods tested in type 2 diabetes. Eur J ClinNutr 56, 622–628.

Jenkins DJ, Wolever TM, Buckley G, Lam KY, Giudici S, Kalmusky Jet al. (1988). Low-glycemic-index starchy foods in the diabeticdiet. Am J Clin Nutr 48, 248–254.

Jenkins DJ, Wolever TM, Taylor RH, Barker H, Fielden H, Baldwin JMet al. (1981). Glycemic index of foods: a physiological basis forcarbohydrate exchange. Am J Clin Nutr 34, 362–366.

Johannes CB, Crawford SL, Woods J, Goldstein RB, Tran D, MehrotraS et al. (2000). An electronic menstrual cycle calendar:comparison of data quality with a paper version. Menopause 7,200–208.

Kabir M, Oppert JM, Vidal H, Bruzzo F, Fiquet C, Wursch P et al.(2002). Four-week low-glycemic index breakfast with a modestamount of soluble fibers in type 2 diabetic men. Metabolism 51,819–826.

Littell R, Milliken G, Stroup W, Wolfinger R (1996). SAS System forMixed Models. SAS Institute Inc.: Cary, NC.

Ma Y, Li Y, Chiriboga D, Olendzki B, Hebert J, Li W et al. (2006).Association between carbohydrate intake and serum lipids. J AmColl Nutr 25(2), 155–163.

Ma Y, Olendzki B, Chiriboga D, Hebert J, Li Y, Li W et al. (2005).Association between dietary carbohydrates and body weight. Am JEpidemiol 161, 359–367.

Marlett JA, McBurney MI, Slavin JL (2002). Position of the AmericanDietetic Association: health implications of dietary fiber. J Am DietAssoc 102, 993–1000.

Matthews CE, Freedson P, Hebert J, Stanek E, Ockene I, Merriam P(2000). Comparison of physical activity assessment methods inthe seasonal variation of blood cholesterol levels study. Med SciSports Exerc 32, 976–984.

Matthews CE, Hebert JR, Ockene IS, Saperia G, Merriam PA (1997).The inter-relation between leisure time physical activity andselected dietary variables. Med Sci Sports Exerc 29, 1199–1207.

Neff LM (2003). Evidence-based dietary recommendations forpatients with type 2 diabetes mellitus. Nutr Clin Care 6, 51–61.

Nuttall FQ (1993). Dietary fiber in the management of diabetes.Diabetes 42, 503–508.

Ostman E, Granfeldt Y, Persson L, Bjorck I (2005). Vinegarsupplementation lowers glucose and insulin responses andincreases satiety after a bread meal in healthy subjects. Eur J ClinNutr 59, 983–988.

Rosal MC, Olendzki B, Reed GW, Gumieniak O, Scavron J,Ockene I (2005). Diabetes self-management among low-incomeSpanish speaking patients: a pilot study. Ann Behav Med 29(3),225–235.

Radloff L (1977). The CES-D scale: a self-report depression scalefor research in the general population. Appl Psychol Meas 1,385–401.

Rizkalla SW, Taghrid L, Laromiguiere M, Huet D, Boillot J, Rigoir Aet al. (2004). Improved plasma glucose control, whole-bodyglucose utilization, and lipid profile on a low-glycemic index dietin type 2 diabetic men: a randomized controlled trial. DiabetesCare 27, 1866–1872.

Schwartz CE, Welch G, Santiago-Kelley P (2005). Cultural challengesin developing a diabetes CAT tool for Hispanics and non-Hispanicspatients. Ann Behav Med 29, S38.

Shiffman S, Gnys M, Richards TJ, Paty JA, Hickcox M, Kassel JD(1996). Temptations to smoke after quitting: a comparison oflapsers and maintainers. Health Psychol 15, 455–461.


1242


Sturrock ND, Pound N, Peck GM, Soar CM, Jeffcoate WJ (1997). Anassessment of blood pressure measurement in a diabetic clinicusing random-zero, semi-automated, and 24-hour monitoring.Diabetes Med 14, 370–375.

Taylor CB, Fried L, Kenardy J (1990). The use of a real-time computerdiary for data acquisition and processing. Behav Res Ther 28, 93–97.

Tiplady B, Crompton G, GE BF Matusiewicz S, Campbell L,Brackenridge D (1997). The use of electronic diaries in respiratorystudies. Drug Inf J 31, 759–764.

Totterdell P, Spelten E, Pokorski J (1995). The effects of nightwork onpsychological changes during the menstrual cycle. J Adv Nurs 21,996–1005.

Tsang MW, Mok M, Kam G, Jung M, Tang A, Chan U et al. (2001).Improvement in diabetes control with a monitoring system basedon a hand-held, touch-screen electronic diary. J Telemed Telecare 7,47–50.

Tsihlias EB, Gibbs AL, McBurney MI, Wolever TM (2000). Compar-ison of high- and low-glycemic-index breakfast cereals with

monounsaturated fat in the long-term dietary management oftype 2 diabetes. Am J Clin Nutr 72, 439–449.

US Department of Health and Human Services (HHS) and theDepartment of Agriculture (USDA) (2005). Dietary Guidelines forAmericans.

Weyman-Daum M, Fort P, Recker B, Lanes R, Lifshitz F (1987).Glycemic response in children with insulin-dependent diabetesmellitus after high- or low-glycemic-index breakfast. Am J ClinNutr 46, 798–803.

Wolever TM (2003). Carbohydrate and the regulation of bloodglucose and metabolism. Nutr Rev 61, S40–S48.

Wolever TM, Jenkins DJ, Vuksan V, Jenkins AL, Wong GS, Josse RG(1992). Beneficial effect of low-glycemic index diet in overweightNIDDM subjects. Diabetes Care 15, 562–564.

Wursch P, Pi-Sunyer FX (1997). The role of viscous soluble fiberin the metabolic control of diabetes. A review with specialemphasis on cereals rich in beta-glucan. Diabetes Care 20,1774–1780.


1243


PDA-assisted low glycemic index dietary intervention for type II diabetes: a pilot study

Documents

Transcript of PDA-assisted low glycemic index dietary intervention for type II diabetes: a pilot study