diabetes research and clinical practice 102 (2013) 158–166

Contents available at ScienceDirect

Diabetes Research
and Clinical Practice
jou rnal hom ep ag e: w ww.e l s e v i er . c om/ loca te / d i ab r es

A short message service (SMS) intervention to

prevent diabetes in Chinese professional drivers
with pre-diabetes: A pilot single-blinded
randomized controlled trial

Carlos K.H. Wong a, Colman S.C. Fung a,*, S.C. Siu b, Yvonne Y.C. Lo a,
K.W. Wong b, Daniel Y.T. Fong c, Cindy L.K. Lam a
a
Department of Family Medicine and Primary Care, The University of Hong Kong, Hong Kong
b
Diabetes Centre, Department of Medicine and Rehabilitation, Tung Wah Eastern Hospital, Hong Kong
c
School of Nursing, The University of Hong Kong, Hong Kong

article info abstract

Article history: Aim: To determine the efficacy of delivering short-message service (SMS) to provide diabe-
Received 7 February 2013 tes-related information in reducing the risk of developing diabetes in Chinese professional
Received in revised form drivers with pre-diabetes.
20 June 2013 Methods: A pilot single-blinded randomized controlled trial was conducted in Hong Kong
Accepted 1 October 2013 between 05/2009 and 04/2012. Professional drivers with impaired glucose tolerance (IGT)
Available online 8 October 2013 were randomly allocated to either a SMS group receiving messages comprising knowledge
and lifestyle modification on diabetes or to a control group with usual care. Primary
Keywords: outcomes were the incidence rate of diabetes mellitus over 12 and 24 months period.
Drivers Results: Fifty-four, out of 104 professional drivers recruited, were randomly allocated to
Chinese intervention group. Fewer subjects developed diabetes at 12 months in intervention group
Pre-diabetes (5.56%) compared to control group (16.00%). Relative risk (RR) of diabetes onset was 0.35
Diabetes (95%CI: 0.10–1.24) and the number needed to treat (NNT) for preventing one diabetes was
Short-message service 9.57. At 24 months, RR increased to 0.62 (95%CI: 0.24–1.61) with a NNT of 10.58. Logistic
Cellular phone regression showed a significant odds ratio of 0.04 (P = 0.021) for intervention group com-
pared to control group at 12-month follow-up for completers and a non-significant odds
ratio of 0.34 (P = 0.303) at 24-month follow-up.
Conclusions: The SMS program proved to have potential to reduce the risk of developing
diabetes at 12 months but additional measures should be integrated to prevent or delay
disease progression.
# 2013 Elsevier Ireland Ltd. All rights reserved.

* Corresponding author at: Department of Family Medicine & Primary Care, Li Ka Shing Faculty of Medicine, The University of Hong Kong,
3/F, Ap Lei Chau Clinic, 161 Ap Lei Chau Main Street, Ap Lei Chau, Hong Kong. Tel.: +852 2518 5756; fax: +852 2814 7475.
E-mail address: cfsc@hku.hk (Colman S.C. Fung).
Abbreviations: DM, diabetes mellitus; T2DM, type 2 DM; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; 2HPPG, two-
hour post-load plasma glucose; SMS, short-messaging service; FPG, fasting plasma glucose; BMI, body mass index; SBP, systolic blood
pressure; DBP, diastolic blood pressure; TC, total cholesterol; TG, triglycerides; HDL-C, high density lipoprotein cholesterol; LDL-C, low
density lipoprotein cholesterol; RR, relative risk; NNT, the number needed to treat; ANOVA, analysis of variance.
0168-8227/$ – see front matter # 2013 Elsevier Ireland Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.diabres.2013.10.002
 diabetes research and clinical practice 102 (2013) 158–166
1. Introduction
Diabetes mellitus (DM) is a leading cause of morbidityand
complications including cardiovascular diseases, blindness,
leg amputation and renal failure, placing enormous burden on
both hospital and community health care [1,2]. Over the past
two decades, the rising epidemic of obesity has contributed to
the increased rates of diabetes and pre-diabetes [3]. People
with pre-diabetes, defined as having impaired fasting glucose
(IFG) or impaired glucose tolerance (IGT) [4], are at very high
risk of developing type 2 DM (T2DM) [5]. In Hong Kong, DM is
the ninth leading cause of death for both men and women in
2011 [6]. Almost 10% of the Hong Kong adult population has
DM, with another 15% having IGT and 1.2% having IFG [7]. A
high two-hour post-load plasma glucose (2HPPG) level was the
major predictor for the progression of IGT to T2DM in previous
prospective studies [8–10].
Professional drivers are at higher risk of ischemic heart
diseases, including various coronary heart diseases and
myocardial infarction [11–14] and musculoskeletal disorders
[15] than other occupational groups. This was attributed to the
drivers’ unfavorable lifestyles of high fat and low fiber diet, low
physical activity level, smoking and inadequate sleep [16]. A
recent diabetes screening program identified one case of pre-
diabetes in ten (10.0%) professional drivers and one undiag-
nosed T2DM in almost twelve (8.1%) drivers [17]. The program
reported that long working hours and shifting duties had
detrimental effect on their health-related quality of life [18]
compared with general population. Their work nature made it
difficult for them to access medical services, and developing
alternative ways to manage this high risk group of people is an
imminent challenge.
Mobile phones allow drivers to send and receive text
messages through short-message service (SMS), which pro-
vides a perfect medium for delivering health information and
support. Web-based SMS messaging service allows large
batches of text messages to be sent at once to different
mobile numbers, thus minimizing cost. SMS program has been
deployed in many countries for providing information support
on smoking cessation [19] and managing chronic diseases
such as T2DM [20,21,22,23,24]. For instance, in SMS-based
diabetes management program [20–22], patients send their
fasting glucose levels daily by mobile phone or through the
internet to the program staff. A nurse practitioner replies to
patients weekly with recommendations by SMS. Results
demonstrated an improvement in glycaemic control and
modification in lifestyle to reduce the incidence of T2DM in
subjects with IGT [25,26,27] and subsequently reduce risk
indicators of cardiovascular diseases [28]. Since SMS was little
exploited in clinical research or practice in Chinese popula-
tion, we put forward a simple yet innovative intervention to
test the usefulness of a SMS intervention program with a
randomized controlled trial design for professional drivers
with pre-diabetes.
The aim of this pilot randomized controlled trial was to
determine the efficacy of using SMS to provide IGT and DM
knowledge and to reduce the risk of developing T2DM at 12
and 24 months among Chinese professional drivers with pre-
diabetes.
159
2. Patients and methods
2.1. Subjects
A community health promotion project to screen for
undiagnosed DM and pre-diabetes among Chinese profes-
sional drivers in Hong Kong was launched in May 2007 [17].
This project screened 3376 drivers and provided continuous
follow-up for those who were newly diagnosed with DM but
not to those 10% (n = 337) who were found to have pre-
diabetes (i.e. IGT and IFG). The latter were informed of their
screening results and advised to consult their own family
doctors.
All Chinese professional drivers screened and found to
have pre-diabetes were invited to take part in our study.
Subjects were included if they (i) had been identified within
the last 3 months with pre-diabetes which is defined as a FPG
level of 5.6–6.9 mmol/L or a 2HPPG of 7.8–11.0 mmol/L after a
75 grams glucose load [29]; and (ii) were accessible by mobile
phone that could receive Chinese text messages. Subjects
were excluded from the study if they (i) had a history of DM; (ii)
were currently on medications known to alter glucose
tolerance; (iii) did not have a mobile phone; (iv) were unable
to read Chinese characters; and (v) refused to take part in
current study. 93 (27.6%) of all eligible subjects identified in the
community screening project agreed to take part and an
additional 11 subjects were recruited through media adver-
tisement, resulting in a total of 104 subjects. There was no
gender or racial bias in the selection of subjects. Subjects were
enrolled from May 2009 to May 2010, and followed up at 12 and
24 months. Fig. 1 shows the CONSORT flowchart of the
progression of subject recruitment, randomization, allocation
and participation.
2.2. Randomization and blinding
Subjects were assigned to the SMS intervention or control
groups by simple randomization on a 1:1 basis with random
numbers generated by a computer. Randomization was
undertaken by a research staff (C.K.H.W.) who did not involve
in subject recruitment or contacts with the subjects. Clinical
staff (S.C.S. and K.W.W.), who recruited the drivers and
undertook clinical and laboratory measurements, were
blinded to the group allocation but subjects and outcome
assessors were not blinded.
2.3. SMS intervention groups
This SMS intervention is a multifaceted intervention based on
two theories relating to human behaviors, namely, the theory
of planned behavior [30] and the social cognitive theory [31]
behavior. In the theory of planned behavior, human behavior
is influenced by the attitude toward the behavior, the
subjective norms which is the perceived social pressure to
perform or not the behavior and the degree of perceived
behavioral control which refers to the perceived ease or
difficulty of performing the behavior. In the social cognitive
theory, the concept of self-efficacy refers to the people’s
beliefs about their own capabilities to perform the behavior.
160 diabetes research and clinical practice 102 (2013) 158–166

Recruited by driver project Recruited by media

(n=93) advertisement (n=11)

Randomized (n=104)

SMS intervention group (n=54) Control group (n=50)

Baseline (n=54) Baseline (n=50)

• Completed: 53 • Completed: 48
• Uncompleted: 1 • Uncompleted: 2
oDBP Missing: 1 oLDL Missing: 2

6th months follow-up (n=54) 6th months follow-up (n=50)

• Completed: 31 • Completed: 27
• Default: 14 • Default: 11
• Withdrawal: 9 • Withdrawal: 12

12th months follow-up (n=45) 12th months follow-up (n=38)

• Completed: 33 • Completed: 25
• Uncompleted: 5 • Uncompleted: 4
oWaistline Missing: 1 oTC, HDL-C, TG, LDL-C Missing: 2
oLDL-C: 1 oWaistline, SBP, DBP, LDL-C: 1
oTC, HDL-C, TG, LDL: 2 oWaistline TC, HDL-C, TG, LDL-C: 1
oHeight, Weight, BMI, Waistline, SBP, • Discontinued due to DM diagnosis: 8
DBP, TC, HDL-C, TG, LDL-C: 1 • Default: 0
• Discontinued due to DM diagnosis: 3 • Withdrawal: 1
• Default: 3
• Withdrawal: 1

24th months follow-up (n=41) 24th months follow-up (n=29)

• Completed: 23 • Completed: 18
• Uncompleted: 1 • Uncompleted: 0
oSBP, DBP Missing: 1 • Discontinued due to DM diagnosis: 1
• Discontinued due to DM diagnosis: 3 • Default: 10
• Default: 14

Fig. 1 – Flowchart on the subject recruitment, randomized allocation and participation.

We attempted to construct self-management goals as part of
the framework for the content of text messages.
The goal of the SMS intervention was to prevent patients
with pre-diabetes developing diabetes. At 2007, the computer-
based text message database based on the above two
psychological theories that underpin behavior change was
developed by a multidisciplinary team including doctors,
nurses, and dieticians. Thus, the text messages were grouped
under four broad themes: (i) information about diabetes and
pre-diabetes, (ii) information about lifestyle modification, (iii)
social norms of how others would appreciate the lifestyle
modification and (iv) self-efficacy enhancing statements of
how to control and stay on behavior control. Examples of
message included ‘‘Diabetic complications include eye pro-
blems and feet problems’’ under theme i, ‘‘Should choose lean
meat with skin and fat trimmed off’’ under theme ii, ‘‘Smoking
 diabetes research and clinical practice 102 (2013) 158–166 161

is old fashioned, quit smoking is the trend’’ under theme iii, Relative risk (RR) and its associated 95% confidence interval
‘‘Walking 30 min a day, you can do it’’ under theme iv. One text (CI) were calculated for the primary outcome of T2DM onset, a
message was randomly sent to people in the intervention RR less than one indicates a lower risk of T2DM onset in the
group at specified times as follows. In Phase 1 (first 3 months), intervention group. The number needed to treat (NNT) was
text message were sent 3 times a week (total 36 text messages). reported as the reciprocal of absolute reduction in risk of
In Phase 2 (the subsequent 3 months), text messages were sent T2DM onset, in which the number who needed to be treated to
once per week (total 12 text messages). In Phase 3 (the
subsequent 6 months) and Phase 4 (the subsequent 12
Table 1 – Subject characteristics and biometric data of
months), text messages were sent once per month (total 18
SMS intervention and control groups at baseline.
text messages). The sequence of the text messages sent was
generated randomly by computer. Characteristics* Intervention Control
Participants in both groups were given information book- (n = 54) (n = 50)
lets by the research nurse on pre-diabetes, diabetes, and Socio-demographic
health behavior information when they had their baseline Age 54.1  6.1 55.2  6.5
Sex
laboratory (oral glucose tolerance test and full lipid profile)
Male 49 (90.7%) 48 (96.0%)
results. Participants in the control group were treated with
Female 5 (9.3%) 2 (4.0%)
usual care by their own doctors. Marital status
Body mass index (BMI) and waist circumference were Non-married 6 (11.1%) 6 (12.0%)
measured at the end of phase 2 (6 months follow-up), 3 (12 Married 48 (88.9%) 44 (88.0%)
months follow-up), and 4 (24 months follow-up) by the nurse Occupational profile
specialist blinded to the group allocation of subjects. Systolic Type of vehicles
blood pressure (SBP), diastolic blood pressure (DBP), and Taxi 32 (59.3%) 26 (53.1%)
laboratory blood test for FPG, 2HPPG, total cholesterol (TC), Bus/minibus 13 (24.1%) 12 (24.5%)
triglycerides (TG), high-density lipoprotein cholesterol (HDL- Lorry 6 (11.1%) 3 (6.1%)
Private car 3 (5.6%) 7 (14.3%)
C) and low-density lipoprotein cholesterol (LDL-C) were
Other/missing value 0 (0.0%) 1 (2.0%)
measured at the end of phase 3 and 4.
Work experience (year) 18.3  11.4 21.3  11.4
Ethics approval for this study was obtained from the Work hours weekly 58.7  17.7 55.5  12.9
Institutional Review Board (IRB) of the University of Hong Kong Shift base 12 (22.2%) 14 (28.0%)
and Hong Kong West and East Cluster of the Hospital
Lifestyle (%)
Authority (#UW 07-368 and #HKEC-2008-055). Active smoker 9 (17.0%) 4 (8.0%)
Active drinker 32 (59.3%) 32 (64.0%)
2.4. Outcome measures Regular exercise 15 (27.8%) 20 (40.0%)
Frequency of eating out weekly
The primary outcome measures were the incidence rate of DM <6 time(s) 28 (51.9%) 23 (46.0%)
6–10 times 17 (31.5%) 15 (30.0%)
in pre-diabetic drivers over 12 and 24 month period. The
>10 times 9 (16.7%) 12 (24.0%)
secondary outcomes were the changes in biometric data
including BMI, weight, waist circumference, 2HPPG, FPG, SBP, Clinical
Family history of diabetes 25 (46.3%) 16 (32.0%)
DBP, TC, TG, HDL-C, and LDL-C. The SBP, DBP, waist
mellitus
circumference, weight and BMI were measured and recorded
Family history of CHD 7 (13.0%) 10 (20.0%)
at baseline and the end of Phases 2–4. FPG, 2HPPG, TC, TG, History of high blood 8 (14.8%) 11 (22.0%)
HDL-C, LDL-C were measured at baseline and the end of Phase pressure
3 and 4 (12 and 24 months after baseline). Blood samples were
Biometric data
taken for glucose and lipid profile after fasting for 12 h; and Weight (kg) 69.49  10.52 72.32  10.01
plasma glucose, TG, TC and HDL-C were measured using the BMI (kg/m2) 25.55  2.94 26.25  2.95
Abbott Architect c16000 chemistry analyzer. LDL-C was Waist (cm) 89.86  7.42 92.04  8.05
derived from the Friedewald formula. SBP (mmHg) 136.54  15.88 133.90  16.45
Socio-demographics (age, gender and marital status), DBP (mmHg) 80.32  10.67 80.86  11.04
FPG (mmol/L) 5.86  0.42 5.90  0.49
medical history (family history of DM, family history of
2HPPG (mmol/L) 7.28  1.87 7.53  2.00
coronary heart disease, past history of high blood pressure),
TC (mmol/L) 5.35  0.72 5.49  0.93
lifestyle (smoking, drinking, exercise, frequency of eating out) HDL (mmol/L) 1.28  0.40 1.32  0.39
and occupational characteristics (type of vehicle, working TG (mmol/L) 1.71  0.87 1.77  1.09
experience, working hours, and shift base) of professional LDL (mmol/L) 3.34  0.70 3.47  0.85
drivers were measured at baseline. Note: CHD: Coronary heart disease; BMI: body mass index; SBP:
systolic blood pressure; DBP: diastolic blood pressure; FPG: Fasting
2.5. Statistical analysis plasma glucose; 2HPPG: two-hour post-load plasma glucose; TC:
total cholesterol; HDL: high density lipoprotein; TG: triglyceride;
LDL: low density lipoprotein.
Chi-square test and independent t-test were used to test the *
No significant difference between treatment group by indepen-
possible imbalance in baseline socio-demographic, occupa-
dent T-test or Chi-square test.
tional, lifestyle and clinical characteristics collected between
intervention and control groups.
162 diabetes research and clinical practice 102 (2013) 158–166

Table 2 – RR and NNT of T2DM onset during 12-month period and 24-month period in SMS intervention group compared
to control group according to ITT and complete case analysis.
T2DM onset Intervention Control RR 95% CI NNT
12-month period
Intention-to-treat 3/54 (5.56%) 8/50 (16.00%) 0.35 (0.10, 1.24) 9.57
Complete case 3/41 (7.32%) 8/37 (21.62%) 0.34 (0.10, 1.18) 6.99

24-month period
Intention-to-treat 6/54 (11.11%) 9/50 (18.00%) 0.62 (0.24, 1.61) 10.58
Complete case 6/30 (20.00%) 9/27 (33.33%) 0.60 (0.25, 1.46) 7.50
Note: RR: relative risk; NNT: number needed to treat.

prevent one case pf new onset T2DM compared with the
control group. Associations of T2DM onset at 12-month period
and 24-month period with treatment groups were tested by
logistic regression models with the adjustment of biometric
data at baseline since the baseline plasma glucose concentra-
tion [8] and smoking habit [32] were significant prognostic
factors of new T2DM among Chinese patients with pre-DM.
Given the continuous nature of secondary outcomes,
repeated measures analysis of variance (ANOVA) were
conducted to determine any significant difference in biometric
data between intervention and control groups, over time and
their interactions.
Our analyses were undertaken on an intention-to-treat
approach in primary analysis using the SPSS Version 20.0 for
Windows (IBM SPSS lnc., Chicago, IL, USA) with statistical
significance taken at p-value < 0.05. Missing values at subse-
quent follow-ups or at subjects who were lost to follow-ups
(i.e. defaulted or withdrawal) were imputed with last observed
value carried forward. Sensitivity analysis was performed on
complete case (per-protocol) to assess the robustness and
uncertainty of the analysis with data imputation.
3. Results
One hundred and four subjects were recruited and randomly
assigned to either the SMS intervention or control group
(Fig. 1). There was no false inclusion or ineligible subjects
reported after randomization and allocation. In intervention
(control) group, 31(27) completed at six months follow-up but
14(11) defaulted and 9(12) subjects had withdrawn that follow-
up. 45(38) and 41(29) in the intervention (control) group were
followed up at 12 and 24 months, respectively.
Socio-demographic, occupational, lifestyle characteristics
at baseline between intervention and control groups were well
balanced with no significant statistical differences (Table 1).
The mean age was 54.1 years in intervention group and 55.2
years in the control group. Intervention group was less likely to
have obese subjects than control group (67.9% vs 86.0%).
Of 45(38) subjects who completed the 12-month follow-up
in intervention (control) group, 3(8) were diagnosed with T2DM
and 5(4) had incomplete biometric data. At the 24-month
follow-up, T2DM was diagnosed in 6 (out of 54) in the
intervention group and 9 (out of 50) in the control group.
Table 2 shows the RR and NNT for new onset T2DM at 12-
month and 24-month follow-up of the SMS intervention group
compared with the control group. The RR for T2DM onset was
0.35 (95% CI: 0.10–1.24) at 12-month and 0.62 (95% CI: 0.24–1.61)
at 24-month assessments, while the NNT for preventing one
case of T2DM at 12-month was 9.6 and at 24-month was 10.6.
Table 3 shows the effects of SMS intervention on the T2DM
onset in pre-diabetes drivers during the 12-month and 24-
month periods tested by logistic regressions. After adjusting
for baseline characteristics in intention-to-treat analysis,
T2DM onset in the intervention group was marginally lower
at 12 months (P = 0.059) and showed no significant difference

Table 3 – Effect of the SMS intervention on T2DM onset of pre-diabetes drivers during 12-month and 24-month period by
logistic regression.
T2DM onset Intention-to-treat Complete case

Odds ratio 95% CI P-value Odds ratio 95% CI P-value

12-month period
FPG (mmol/L) 72.14* (3.52, 1478.27) 0.005 130.98* (3.63, 4731.86) 0.008
2HPPG (mmol/L) 1.55 (0.93, 2.58) 0.095 2.01 (1.00, 4.06) 0.051
Smoking 15.35* (1.08, 218.67) 0.044 63.40* (1.37, 2926.54) 0.034
Intervention 0.15 (0.02, 1.07) 0.059 0.04* (0.00, 0.61) 0.021

24-month period
FPG (mmol/L) 7.29* (1.42, 37.30) 0.017 27.59* (2.49, 305.38) 0.007
2HPPG (mmol/L) 1.78* (1.14, 2.78) 0.011 2.54* (1.19, 5.43) 0.016
Smoking 9.19* (1.29, 65.29) 0.027 18.15 (0.92, 358.00) 0.057
Intervention 0.51 (0.13, 2.03) 0.336 0.34 (0.04, 2.68) 0.303
T2DM: Type 2 diabetes mellitus; FPG: fasting glucose; 2HPPG: two-hour post-load plasma glucose.
*
Statistically significant ( p < 0.05) difference by logistic regression.
Table 4 – Effect of the SMS intervention on the change in the level of biometric data by repeated measure ANOVA.
Repeated measure ANOVA

Time Intention-to-treat Complete case

Baseline 6 months 12 months 24 months Group Time Group  time Group Time Group  time
* * *
Weight (kg) 0.094 <0.001 0.020 0.316 <0.001 0.149
Control 72.32  10.01 72.58  10.29 72.30  10.49 71.91  10.88

diabetes research and clinical practice 102 (2013) 158–166

Intervention 69.49  10.52 69.01  10.40 68.40  10.19 68.47  10.35
BMI (kg/m2) 0.112 0.005* 0.019* 0.994 0.003* 0.482
Control 26.25  2.95 26.24  2.99 26.28  3.14 26.18  3.27
Intervention 25.55  2.94 25.31  3.02 25.18  3.10 25.11  3.04
Waist (cm) 0.124 0.453 0.990 0.455 0.833 0.140
Control 92.04  8.05 91.78  8.29 91.70  8.33 91.72  8.50
Intervention 89.86  7.42 89.45  7.22 89.38  7.08 89.34  7.47
SBP (mmHg) 0.602 0.584 0.440 0.869 0.389 0.538
Control 133.90  16.45 135.18  17.65 132.48  19.13 133.74  18.65
Intervention 136.54  15.88 135.06  17.16 135.46  19.65 134.96  16.31
DBP (mmHg) 0.774 0.033* 0.158 0.593 0.079 0.227
Control 80.86  11.04 80.34  11.22 80.12  13.02 79.74  11.94
Intervention 80.32  10.67 81.87  17.94 77.76  12.73 77.85  11.64
FPG (mmol/L) 0.468 0.695 0.517 0.920 0.979 0.590
Control 5.90  0.49 NA 5.98  0.60 5.94  0.59
Intervention 5.86  0.42 NA 5.85  0.60 5.89  0.62
2HPPG (mmol/L) 0.171 0.069 0.245 0.931 0.170 0.526
Control 7.53  2.00 NA 8.06  3.00 8.15  2.83
Intervention 7.28  1.87 NA 7.11  2.25 7.63  2.53
TC (mmol/L) 0.318 0.223 0.607 0.175 0.769 0.526
Control 5.49  0.93 NA 5.45  0.99 5.42  0.87
Intervention 5.35  0.72 NA 5.24  0.77 5.28  0.90
HDL (mmol/L) 0.946 0.011* 0.605 0.368 0.025* 0.817
Control 1.32  0.39 NA 1.22  0.27 1.21  0.26
Intervention 1.28  0.40 NA 1.24  0.27 1.22  0.25
TG (mmol/L) 0.357 0.886 0.392 0.538 0.992 0.826
Control 1.77  1.09 NA 1.95  1.90 1.93  1.91
Intervention 1.71  0.87 NA 1.61  1.15 1.65  1.20
LDL (mmol/L) 0.277 0.965 0.687 0.134 0.910 0.659
Control 3.47  0.85 NA 3.50  0.92 3.49  0.80
Intervention 3.34  0.70 NA 3.32  0.72 3.33  0.77
Note: BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; FPG: fasting glucose; 2HPPG: two-hour post-load plasma glucose; TC: total cholesterol; HDL: high density
lipoprotein; TG: triglyceride; LDL: low density lipoprotein; NA: not applicable.
*
Statistically significant ( p < 0.05) difference.

163
164 diabetes research and clinical practice 102 (2013) 158–166
Fig. 2 – Mean BMI changes over time for intervention and
control group.
at 24 months (P = 0.336). After adjusting for baseline char-
acteristics in complete case analysis, subjects in the SMS
intervention group had significantly lower odds of having
T2DM onset during the 12-month period (P = 0.021) but
insignificantly different odds of at 24-months. On the other
hand, higher FPG had a higher odds ratio for T2DM onset at
both 12 and 24 months.
Table 4 shows the effect of the SMS intervention on the
change in the level of biometric data by repeated measure
ANOVA. Significant interaction between treatment group and
time was found in BMI (P = 0.019). Mean BMI decreased over
time in intervention group but remain stable in control group
(Fig. 2). There were significant mean differences on DBP and
HDL-C over time (P = 0.033; P = 0.011) in the intention-to-treat
analysis. In complete case analysis, significant mean differ-
ences on BMI and HDL-C (P = 0.003; P = 0.025) were also found
over time. However, the mean change in waist circumference,
SBP, FPG, 2HPPG, TC, TG and LDL-C were not significantly
different between groups, over time or in interaction effect
between groups and time.
4. Discussions
This randomized controlled trial delivered one-way SMS
information support via mobile phone for two years in
Chinese professional drivers. To our knowledge, this is the
first pilot trial to assess the effectiveness of SMS intervention
program in preventing the progression of pre-diabetes to
T2DM and controlling glucose level, blood pressure and lipid
profile in a Chinese population. One of the strengths of the
current trial was a RCT design resulting in baseline char-
acteristics not significantly different between the intervention
and control groups because it was less affected by selection or
sampling bias.
Overall 10.6% and 14.4% of pre-diabetic Chinese profes-
sional drivers progressed to T2DM at 12 and 24 months,
respectively. Subjects in the intervention group had a lower
incidence rate of T2DM than subjects in the control group,
providing some evidence that SMS-based program was
effective in the risk reduction of T2DM from pre-diabetes
groups in complete case analysis.
Logistic regressions identified fasting plasma glucose as the
main significant independent predictor of the progression to
T2DM in the first 12 months, while both fasting plasma glucose
and 2-h post-meal glucose were key predictors for the
progression to T2DM in the 24 months. These findings were
consistent with those found in other studies. In a cohort study
of Chinese subjects with risk factors for T2DM [8], 50% of those
with IGT progressed to develop T2DM in 4.34 years with an
estimated annual conversion rate of 11.5%, and the major
predictor of the progression of IGT to T2DM was the 2HPPG
level collected at the first visit. In a Bedford survey under an
observation period of over four years in the UK [10] and Japan
[9], an increase in the initial level of 2HPPG was associated with
a higher likelihood of progression to T2DM with a total of 15–
16.7% of IGT worsened to T2DM.
It was surprising that the SMS intervention program
significantly decreased BMI, but previous clinical trials
delivering short message service via mobile phone or internet
to patients with Type 1 DM [20] and T2DM [21,22] were not
efficacious in reducing BMI. In contrast to the results of
previous studies [21,22], our trial did not find any significant
improvement in 2HPPG in the intervention group compared
with the control group. Although the changes in 2HPPG value
itself were not statistically different, the mean 2HPPG decrease
in SMS group from 7.3 mmol/L to 7.1 mmol/L coupled with the
rise of mean 2HPPG from 7.5 mmol/L to 8.1 mmol/L in the
control group at 12-month may explain clinically why there
were less newly diagnosed DM patients in the SMS group at 12-
month. The other consistent findings were that the interven-
tion group did not have lower fasting plasma glucose, total
cholesterol, and triglyceride than the control group at the end
of the trial. Another point worth noting is that, despite both
SMS and control groups having a significant decreasing trend
in HDL-C over time, the SMS group showed a comparatively
smaller drop of HDL-C than the control group.
Our study has provided preliminary evidence that a SMS
intervention program may be effective in reducing the
incidence rate of T2DM in Chinese professional drivers with
pre-diabetes. The SMS intervention program has the superi-
ority of low monetary cost with a budget of
HKD$39.60 = USD$5.08 per subject (each SMS costs about
HKD$0.6 = USD$0.08), but a more precise cost versus effec-
tiveness in comparison to usual care or other interventions is
still unknown. Other costs related to the additional demand
for clinical and laboratory examinations from the perspective
of health service providers need to be considered. More in-
depth cost-benefit analysis or cost-effectiveness analysis
should be utilized to estimate the public willingness-to-pay
for reduction in conversion to T2DM (or gains in effectiveness),
and to model on alternative intervention strategies. Although
the SMS intervention program is slightly more expensive than
standard usual care, the one-way communication is expected
to reduce the intangible costs incurred during the waiting time
spent at the primary health care system. Further studies
designing a two-way interactive mode of SMS intervention
program by advocating a communication channel between
 diabetes research and clinical practice 102 (2013) 158–166
pre-diabetic or T2DM patients and clinicians should be
warranted in Chinese populations.
4.1. Limitations
There were several drawbacks in this pilot study. Due to
limitations of workup resource in outpatient setting, this
study at the six months follow-up assessment did not collect
the short-term outcome for the readings of fasting glucose and
full lipid profile. Overall follow-up rates dropped substantially
in the last follow-up at 24 months, particularly in the control
group. The attrition rates from both intervention and control
groups were relatively high, which inflated the missing data
and may in turn lead to an underestimation of the incidence of
DM. Meanwhile, individual changes in lifestyle behaviors were
difficult to be estimated due to the unsatisfactory response
rate to lifestyle questions at follow-up assessments. A
complimentary incentive may be offered to patients upon
completion in future clinical trials, or following up subjects by
their own primary care doctors might improve the study
completion rates. Secondly, subjects who were diagnosed with
DM had missing values of biometric data and lifestyle
questions at latter follow-up assessments because they were
terminated from delivering SMS intervention and may be
referred to primary and secondary care service providers
following diabetes management and establishing treatment
goals routinely. It may not be feasible to collect the biometric
data among these newly diagnosed diabetic patients as they
may select other clinics to follow-up their DM subsequently.
Finally, our study adopted the most commonly used carrying
forward of last observation to deal with the missing data.
Other simpler methods like worst case, regression or multiple
imputation could be alternatives in dealing with missing data
[33] were not tested due to limitation of the scope of our study.
In conclusion, subjects in the intervention group had a
lower incidence rate (5.6% at 12 months and 11.1% at 24
months) of T2DM than subjects in the control group (16.0% at
12 months and 18.0% at 24 months). This pilot study provided
some evidence to show that SMS intervention might help to
prevent or delay pre-diabetic Chinese professional drivers
from developing T2DM at 12 months but the long-term effects
need further exploration. Compliance with follow-up was
found to be a problem among drivers, which would need to be
addressed in both clinical trials and practice.
Author contributions
C.K.H.W. wrote/edited the manuscript and researched data.
C.S.C.F. wrote/edited the manuscript. S.C.S. and K.W.W.
contributed to acquisition of data and reviewed/edited the
manuscript. Y.Y.C.L. contributed to study design. D.Y.T.F.
reviewed/edited the manuscript, contributed to statistical
analysis and interpretation of results. C.L.K.L. reviewed/edited
the manuscript.
Clinical trial registration number
NCT01556880, ClinicalTrials.gov.
165
Conflict of interest statement
The authors declare that they have no conflict of interest.
Acknowledgments
The study is funded by the Board of the Tung Wah Group of
Hospitals. The authors would like to thank Miss H.Y. Chung,
Mr. Kelvin Wong and Mr. Eric Wan for assistance in subject
enrolment, data collection and data analysis.
references
[1] Wild S, Roglic G, Green A, Sicree R, King H. Global
prevalence of diabetes. Diabetes Care 2004;27(5):1047–53.
[2] Whiting DR, Guariguata L, Weil C, Shaw J. IDF Diabetes
Atlas: global estimates of the prevalence of diabetes for
2011 and 2030. Diab Res Clin Pract 2011;94(3):311–21.
[3] Little P, Margetts B. The importance of diet and physical
activity in the treatment of conditions managed in general
practice. Br J Gen Pract 1996;46(404):187–92.
[4] American Diabetes Association. The prevention or delay of
type 2 diabetes. Diabetes Care 2002;25(4):742–9.
[5] de Vegt F, Dekker JM, Jager A, Hienkens E, Kostense PJ,
Stehouwer CDA, et al. Relation of impaired fasting and
postload glucose with incident type 2 diabetes in a dutch
population. JAMA 2001;285(16):2109–13.
[6] Hospital Authority Hospital Authority Statistical Report
2009–2010; 2011. Available from: http://www.ha.org.hk/
upload/publication_15/321.pdf.
[7] Janus ED, Wat NMS, Lam KSL, Cockram CS, Siu STS, Liu LJ,
et al. The prevalence of diabetes, association with
cardiovascular risk factors and implications of diagnostic
criteria (ADA 1997 and WHO 1998) in a 1996 community-
based population study in Hong Kong Chinese. Diabet Med
2000;17(10):741–5.
[8] Ko GT, Li JK, Cheung AY, Yeung VT, Chow CC, Tsang LW,
et al. Two-hour post-glucose loading plasma glucose is the
main determinant for the progression from impaired
glucose tolerance to diabetes in Hong Kong Chinese.
Diabetes Care 1999;22(12):2096–7.
[9] Kadowaki T, Miyake Y, Hagura R, Akanuma Y, Kajinuma H,
Kuzuya N, et al. Risk factors for worsening to diabetes in
subjects with impaired glucose tolerance. Diabetologia
1984;26(1):44–9.
[10] Keen H, Jarrett R, McCartney P. The ten-year follow-up of
the Bedford Survey (1962–1972): Glucose tolerance and
diabetes. Diabetologia 1982;22(2):73–8.
[11] Tüchsen F, Endahl LA. Increasing inequality in ischaemic
heart disease morbidity among employed men in Denmark
1981–1993: the need for a new preventive policy. Int J
Epidemiol 1999;28(4):640–4.
[12] Rosengren A, Anderson K, Wilhelmsen L. Risk of coronary
heart disease in middle-aged male bus and tram drivers
compared to men in other occupations: a prospective
study. Int J Epidemiol 1991;20(1):82–7.
[13] Bigert C, Gustavsson P, Hallqvist J, Hogstedt C, Lewne M,
Plato N, et al. Myocardial infarction among professional
drivers. Epidemiology 2003;14(3):333–9.
[14] Quah CH, Ng JM, Puar TH. Does occupational driving
increase the risk of cardiovascular disease in people with
diabetes? Diab Res Clin Pract 2013;99(1):e9–11.
166 diabetes research and clinical practice 102 (2013) 158–166
[15] Szeto G, Lam P. Work-related musculoskeletal disorders in
urban bus drivers of Hong Kong. Journal of Occupational
Rehabilitation 2007;17(2):181–98.
[16] Korelitz JJ, Fernandez AA, Uyeda VJ, Spivey GH, Browdy BL,
Schmidt RT. Health habits and risk factors among truck
drivers visiting a health booth during a trucker trade show.
Am J Health Promot 1993;8(2):117–23.
[17] Siu SC, Wong KW, Lee KF, Lo YYC, Wong CKH, Chan AKL,
et al. Prevalence of undiagnosed diabetes mellitus and
cardiovascular risk factors in Hong Kong professional
drivers. Diab Res Clin Pract 2012;96(1):60–7.
[18] Wong CKH, Fung CSC, Siu SC, Wong KW, Lee KF, Lo YYC,
et al. The impact of work nature, lifestyle and obesity on
health-related quality of life in Chinese professional
drivers. J Occup Environ Med 2012;54(8):989–94.
[19] Rodgers A, Corbett T, Bramley D, Riddell T, Wills M, Lin R-B,
et al. Do u smoke after txt? Results of a randomised trial of
smoking cessation using mobile phone text messaging.
Tob Control 2005;14(4):255–61.
[20] Rami B, Popow C, Horn W, Waldhoer T, Schober E.
Telemedical support to improve glycemic control in
adolescents with type 1 diabetes mellitus. Eur J Pediatr
2006;165(10):701–5.
[21] Kim HS. A randomized controlled trial of a nurse short-
message service by cellular phone for people with diabetes.
Int J Nurs Stud 2007;44(5):687–92.
[22] Yoon KH, Kim HS. A short message service by cellular
phone in type 2 diabetic patients for 12 months. Diab Res
Clin Pract 2008;79(2):256–61.
[23] Hanauer DA, Wentzell K, Laffel N, Laffel LM. Computerized
Automated Reminder Diabetes System (CARDS): E-mail and
SMS cell phone text messaging reminders to support
diabetes management. Diabetes Technol Ther
2009;11(2):99–106.
[24] Hussein WI, Hasan K, Jaradat AA. Effectiveness of mobile
phone short message service on diabetes mellitus
management; the SMS-DM study. Diab Res Clin Pract
2011;94(1):e24–6.
[25] Gillies CL, Abrams KR, Lambert PC, Cooper NJ, Sutton AJ,
Hsu RT, et al. Pharmacological and lifestyle interventions to
prevent or delay type 2 diabetes in people with impaired
glucose tolerance: systematic review and meta-analysis.
BMJ 2007;334(7588):299.
[26] Pan XR, Li GW, Hu YH, Wang JX, Yang WY, An ZX,
et al. Effects of diet and exercise in preventing NIDDM
in people with impaired glucose tolerance: the Da Qing
IGT and diabetes study. Diabetes Care 1997;20(4):
537–44.
[27] Lindström J, Louheranta A, Mannelin M, Rastas M,
Salminen V, Eriksson J, et al. The Finnish Diabetes
Prevention Study (DPS): lifestyle intervention and 3-year
results on diet and physical activity. Diabetes Care
2003;26(12):3230–6.
[28] The Diabetes Prevention Program Research Group. Impact
of intensive lifestyle and metformin therapy on
cardiovascular disease risk factors in the diabetes
prevention program. Diabetes Care 2005;28(4):888–94.
[29] American Diabetes Association. Diagnosis and
classification of diabetes mellitus. Diabetes Care
2012;35(Suppl. 1):S64–71.
[30] Ajzen I. The theory of planned behavior. Organ Behav Hum
Decis Process 1991;50(2):179–211.
[31] Bandura A. Social foundations of thought and action: a
social cognitive theory. Eaglewood Cliffs, NJ: Prentice Hall;
1986.
[32] Ko GTC, Chan JCN, Cockram CS. Change of glycaemic status
in Chinese subjects with impaired fasting glycaemia.
Diabet Med 2001;18(9):745–8.
[33] Wood AM, White IR, Thompson SG. Are missing outcome
data adequately handled? A review of published
randomized controlled trials in major medical journals.
Clin Trials 2004;1(4):368–76.