p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408

Contents lists available at ScienceDirect

Primary Care Diabetes

journal homepage: http://www.elsevier.com/locate/pcd

Review

The health effects of medical nutrition therapy by

dietitians in patients with diabetes: A systematic
review and meta-analysis
Nutrition therapy and diabetes

Jalaledin Mirzay Razaz a,∗ , Jamal Rahmani a , Hamed Kord Varkaneh b ,

Jacqueline Thompson c , Cain Clark d , Hebatullah M. Abdulazeem e
a Department of Community Nutrition, Faculty of Nutrition and Food Technology, National Nutrition and Food
Technology Research Institute, Shahid Beheshti University of Medical Sciences, Tehran, Iran
b Department of Clinical Nutrition and Dietetics, Faculty of Nutrition and Food Technology, National Nutrition and

Food Technology Research Institute, Student Research Committee, Shahid Beheshti University of Medical Sciences,
Tehran, Iran
c Nuffield Department of Orthopaedics, Rheumatology & Musculoskeletal Sciences, University of Oxford, Oxford,

OX3 7LD, United Kingdom

d School of Life Sciences, Coventry University, Coventry, CV1 5FB, United Kingdom
e Arab diploma in Family Medicine, AICPD, Egypt

a r t i c l e i n f o a b s t r a c t

Article history: Aims: Intensive lifestyle, dietary interventions and patient education have been recom-
Received 3 April 2019 mended as key milestones in to facilitate the management of Diabetes and contain the
Received in revised form growing incidence. We performed a systematic review and meta-analysis to assess the
29 April 2019 health benefits of medical nutrition therapy among patients with diabetes.
Accepted 12 May 2019 Design: A systematic search was performed in MEDLINE/PubMed, SCOPUS, and Cochrane
Available online 9 June 2019 library from onset up to February 2019 to identify trials investigating the health effect of
Medical nutrition (MNT) in patients with diabetes. Random-effects models were used to
Keywords: calculate the effect sizes as weighted mean difference (WMD) and 95% confidence intervals
Nutrition therapy (CI).
Diabetes Results: Eleven studies containing 1227 participants were included in the meta-analysis.
Lifestyle Pooled results showed a significant reduction in Fasting blood sugar (FBS) (WMD=
Dietitians −8.85 mg/dl, 95% CI: −14.41, −3.28), HbA1c (WMD: −0.43%, 95% CI: −0.69, −0.17), weight
Diet (WMD: −1.54 kg, 95% CI: −2.44, −0.64), Body mass index (BMI) (WMD: −0.34 Kg/m2, 95% CI:
−0.52, −0.17), waist circumference (WMD: −2.16 cm, 95% CI: −4.09, −0.23), cholesterol (WMD:
−4.06 mg/dl, 95% CI: −7.31, −0.81), Systolic blood pressure (SBP) (WMD: −7.90 mmHg, 95% CI:
−13.03, −2.77). Results of meta-regression analysis based on age of participants and duration
of intervention were not significant.

∗
Corresponding author at: Department of Community Nutrition, Faculty of Nutrition Sciences and Food Technology, National Nutrition
and Food Technology Research Institute, Shahid Beheshti University of Medical Sciences, West Arghavan St., Farahzadi Blvd., P.O. Box
19395-4741 Tehran, Iran.
E-mail address: jmrazaz2018@gmail.com (J.M. Razaz).
https://doi.org/10.1016/j.pcd.2019.05.001
1751-9918/© 2019 Primary Care Diabetes Europe. Published by Elsevier Ltd. All rights reserved.
400 p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408

Conclusions: Patients with diabetes who received medical nutrition therapy showed signifi-
cant improvements in outcome measures of FBS, HbA1c, weight, BMI, waist circumference,
cholesterol, and SBP.
© 2019 Primary Care Diabetes Europe. Published by Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
2.1. Search strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
2.2. Eligibility criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
2.3. Data extraction and quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
2.4. Quality assessment of studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
2.5. Quantitative data synthesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .401
3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
3.1. Study characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
3.2. Interventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
3.3. Results of meta-analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
3.4. Subgroup analysis and meta-regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
3.5. Publication bias and sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
4.1. Comparison with previous findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406
4.2. Limitations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .407
5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
Funding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
Conflict of interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407

1. Introduction that include physicians, nurses, pharmacists, and (or) dieti-

tians, have shown better patient outcomes such as reduction
The global prevalence of diabetes mellitus, defined as fasting in weight, BMI, and fasting blood sugars compared to usual
blood glucose equal to or higher than 7 mmol/L, among adults care [8,11–14]. However, many of these studies were poorly
over 18 years of age has risen from 4.7% in 1980 to 8.5% in designed. Methodological flaws such as measurement bias,
2014 (90–95% of which is Type 1I). Diabetes has been directly small sample size, lack of generalizability, short interven-
linked to 1.6 million deaths globally. High blood glucose alone tion and follow-up periods were some of the issues identified
was the cause of another 2.2 million deaths in 2015, and is an [11,15–17]. There is evidence to suggest that engaging other
established risk factor for coronary heart disease, ischaemic skilled health professionals such as dietitians in primary care
stroke, and other vascular diseases [1–3]. Higher health care improves patient outcomes [12]. Dietitians, both quality care
use, economic burden and associated societal costs have been providers and educators have an important role to play within
reported among people with diabetes when compared to their multidisciplinary diabetes care teams. The Diabetes Preven-
normal counterparts [4]. In the US, approximately 20% of the tion Program (DPP) provides clear indications for integrating
nation’s health resource is spent treating people with diabetes dietitians into lifestyle management to help patients change
[5]. In the UK, diabetes management accounts for around 10% their eating and exercise behaviours and prevent diabetic
of the National Health Service budget and is projected to rise complications [18].
to around 17% by 2035 [6]. Nutritional therapy is one of the key components of pre-
Medical nutrition therapy have been recommended as vention and management strategies for Type 1 and II diabetes
milestones in the management of type 2 diabetes. There is mellitus. A number of studies have also demonstrated sus-
evidence supporting the effectiveness of patient education tained improvements in HbA1c at 12 months [13,14]. These
and adherence to self-management strategies on health out- effects were sustained for longer periods when a registered
comes [7,8]. National recommendations suggest promoting dietitian provided follow-up visits ranging from quarterly
the role of diabetes educators alongside treating physicians to monthly sessions [13,14]. Ongoing medical counselling
to improve the coordination of care and facilitate better in nutritional management by a trained dietitian has been
outcomes in diabetes management [9,10]. Currently, studies shown to lead to better long-term metabolic control [19]. In
evaluating the role of multidisciplinary diabetes care teams a clinical trial evaluating a 24-month intervention, weight
 p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408 401

(−0.7 vs. + 2.1 kg), BMI (+0.3 vs. + 0.7 kg/m2 ), waist circum- strategy. Discrepancies were resolved by consensus agree-
ference (–1.3 vs. + 2.4 cm) and overall energy intake (−548 ment. When, consensus was not achieved, a senior author
vs. −74 kcal/day) significantly differed between groups, with (S.J.M.R.) involved in the study helped to resolve disagree-
nutritional therapy demonstrating greater improvement com- ments. We extracted data on the following items from each
pared to the control group [20]. Another study by Bhopal, study: name of the first author, year of publication, type of
reported significant weight loss in intervention group, vs study population, number of participants in the interven-
weight gain in control group following nutritional therapy [21]. tion and control groups, gender, participants mean age, study
Furthermore, Mohammadi and colleagues reported improve- location, study design, intervention components, intervention
ment in anthropometric measures, fasting blood sugar, 2-h duration, and type of diabetes. For the results, we extracted
postprandial blood sugar, serum total cholesterol, serum ala- information on the anthropometric, nutritional and biochem-
nine transaminase and increased circulating following diet ical variables. Means and standard deviations of metabolic
therapy [22]. variables at baseline, end of study and/or changes between
However, despite a substantial number of independent baseline and period of intervention delivery). When this data
studies reporting positive, and sustained, outcomes following was unavailable, we emailed the corresponding author to
nutritional therapy, there is no consensus on its’ overarch- obtain information missing in the published study report.
ing effect. Given the clinical importance of such nutritional
management in clinical practice and poor uptake of current
guideline recommendations. We sought to conduct a system-
atic review and meta-analysis of the health benefits of medical 2.4. Quality assessment of studies
nutrition therapy among patients with diabetes.
We evaluated the quality of included trials using the Cochrane
2. Methods quality assessment tool which comprises of the following
domains: random sequence generation, allocation conceal-
This systematic review and meta-analysis was conducted ment, blinding of participants and personnel, blinding of
using recommendations outlined by the Preferred Reporting outcome assessment, incomplete outcome data, selective
Items of Systematic Reviews and Meta-Analysis statement reporting and other probable sources of biases. To assess the
guidelines [23]. quality of studies, each study was assigned a label (yes, no or
unclear). This information was used to classify the study as
having a low risk, high risk or unknown risk of bias, respec-
2.1. Search strategy
tively [24]. Studies were judged to be of high risk of bias when
the randomization, allocation concealment and blinding was
A systematic search was conducted by combining med-
not reported or not performed. Studies were judged to be of
ical subject headings (MeSH) and non-MeSH terms in
low risk of bias when all critical items or more items on the
PubMed/MEDLINE, Cochrane and SCOPUS with no language
assessment domains were reported [24].
or date restrictions. Databases were searched from inception
to February 2019 (Supplemental Table1). To avoid missing any
relevant studies, reference lists of eligible studies and related
reviews were searched manually.
2.5. Quantitative data synthesis
2.2. Eligibility criteria
Mean change and standard deviation (SD) of the outcome
We included studies that met the following inclusion criteria: measures were used to estimate the overall effect size.
All studies that evaluated the effect of Medical Nutrition If the SD of the mean difference was not reported in
Therapy on diabetes patients were included in this meta- the studies, we derived this value using the following for-
analysis. We defined medical nutrition therapy as nutritional mula: SD change = square root [(SD baseline 2 + SD final 2 ) –
consultation provided by a registered dietitian. We included (2 × r × SD baseline × SD final )] [25]. The random-effects model
studies that met the following inclusion criteria: (1) studies, (using DerSimonian–Laird method) was used to estimate the
irrespective of design that had control groups receiving usual weighted mean difference (WMD) and corresponding 95%
care; (2) studies that evaluated the effectiveness of prescribing confidence intervals (95% CI). We carried out predefined sub-
Medical Nutrition Therapy by a Registered Dietitians; and (3) group analysis for region (location of study) to detect potential
reported sufficient information on metabolic variables both in sources of heterogeneity among the studies. Meta-regression
control and intervention groups. Prospective studies without was used to determine the effect of duration of intervention
a suitable control group were excluded. We excluded studies and participant age on intervention outcomes. Publication
that did not report outcome measures (or changes in outcome bias was evaluated by means of visual assessment of fun-
measures) at baseline to the end of intervention or follow-up. nel plots and Egger’s tests [26]. When publication bias was
detected, it was re-evaluated using the ‘trim and fill’ approach
2.3. Data extraction and quality assessment [27]. Sensitivity analysis was performed to investigate the
effect of each study on overall analysis. All statistical anal-
Two independent researchers (H.K.V and J.R) screened and yses were executed using Stata software (Stata Corp. College
extracted relevant data for all studies identified by the search Station, Texas, USA).
 402
Table 1 – Baseline characteristics of included studies in the meta-analysis.
Studies Author Country Study year Age (years) Patients, n Follow up (in Type of Duration of Intervention Control
weeks) diabetes intervention
(week)
1 Marincic US 2018 – 165 12 Type 2 diabetes 12 Medical nutrition Diabetes
therapy self-management

p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408
education
2 Mohammadi IRAN 2017 49 30 10 T2DM 10 Diet therapy Usual care
3 Liu China 2015 62 117 52 T2DM 52 Dietitian-led Usual care
intervention
4 Parker US 2014 50 76 12 Prediabetes 12 Medical nutrition Usual care
therapy
5 Niswender US 2014 57 478 26 T2DM 26 Dietary interven- Usual care + insulin
tion + insulin detemir
detemir
6 Miller US 2014 50 48 18 T2DM 18 Medical nutrition Self-Care
therapy
7 Battista Canada 2012 59 88 104 T1,2DM 104 Dietitian-coached Usual care
8 Huang Taiwan 2010 56 154 52 Type 2 diabetes 52 Registered Usual care
dietitian–led
intervention group
9 Timmerberg US 2009 45 26 16 Type 2 diabetes 16 Group nutrition Group nutrition
class + follow up class
with dietitians
10 Trento Italy 2008 66 45 104 Type 2 diabetes 104 Nurse-, dietitianand Usual care
pedagogist-led
Group Care
11 Eriksson Finland 1999 54 212 52 Type 1I diabetes 52 Medical nutrition Received general
therapy information about
the lifestyle changes
 p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408 403

3. Results

other sources
Other bias
We retrieved 266 studies using our search strategy after remov-

of bias
ing duplicates papers (Supplemental Figure 1). At title and

Low
Low
Low
Low

Low
Low
Low
Low
Low
abstract screening, 241 papers were removed using the study
selection criteria and 25 articles were retained for full text
eligibility assessment. Eleven articles [28–38] met the review

bias selective
inclusion criteria. Fourteen studies were excluded for the fol-
lowing reasons: (1) outcome data was presented in unsuitable

Reporting

reporting
format and attempts to obtain complete data was unsuccess-
ful (n = 7), (2) did not include control group comparisons (n = 4),

Low
Low
Low
Low

Low
Low
Low
Low
Low
and (3) intervention was designed without including a nutri-
tionist (n = 3).

outcome data
Attrition bias
3.1. Study characteristics

incomplete
Table 1 presents the characteristics of included studies. Stud-

Low
Low
Low
Low

Low
Low
Low
Low
Low
ies were conducted in the US [28,31–33,36], China [30], Iran [29],
Canada [34], Taiwan [35], Italy [37], and Finland [38]. Studies
were published between 1999–2018 and duration of interven-
tion was between 10–104 weeks. Mean age of participants was

bias blinding
54 (45–66) years. Ten studies were evaluated among patients

Detection
with diabetes [28–30,32–38] and one conducted on pre-patients

One-arm interrupted time series design, a quasi-experimental research design

with diabetes [31]. Routine care was the common interven-

Low
Low
Low
Low

Low
Low
Low
Low
Low
Retrospective chart review from the electronic medical records of patients
tions evaluated among the control groups. Table 2 shows the
risk of bias assessment of studies included in this review. One
study in selection bias (Random sequence generation) [35] and
two study in selection bias (allocation concealment) [16,38]
Performance

had low risk of bias.

blinding

3.2. Interventions
bias

Low
Low
Low
Low

Low
Low
Low
Low
Low
Most intervention groups in studies included this review
received recommendations for physical activity and individ-
ualized nutrition counselling to promote health and behavior
Selection bias

concealment

change by a registered dietitian.

allocation

Unclear
Unclear
Unclear
Unclear

Unclear
Unclear
Unclear

3.3. Results of meta-analysis

Low
Low

Eight studies providing a total of 612 participants in inter-

vention group and 588 participants in control group with
Selection bias

data reported on levels of Fasting blood sugar as an out-

Table 2 – Risk of bias assessment of the studies.

generation

come variable [29–32,34,35,37,38]. Results combined using

sequence
random

random effects model showed a significant reduction in FBS

Unclear
Unclear
Unclear

Unclear

Unclear
Unclear
Unclear

levels following medical nutrition therapy (MNT) interven-

High

Low

tion (Weighted mean difference(WMD): −8.85 mg/dl, 95% CI:

−14.41, −3.28) (Fig. 1). Significant heterogeneity was identified
among studies (p = 0.001, I2 = 76%).
Year

2018
2017
2015
2014
2014
2014
2012
2010
2009
2008
1999

We combined results from eight studies [30–37] that

reported HBA1C as an outcome measure using a random-
effects model. This provided a total of 1032 participants
(intervention = 522 and control = 510). The results demon-
Mohammadi et al. [22]

Timmerberg et al. [36]

Niswender et al. [32]

strated that MNT intervention significantly reduced HBA1C

Marincic et al. [28]

Eriksson et al. [38]

levels (WMD: −0.43%, 95% CI: −0.69, −0.17) in the intervention

Battista et al. [20]
Huang et al. [35]

Trento et al. [37]

Parker et al. [31]

Miller et al. [33]

Study name

group compared to the control group (routine care). How-

Liu et al. [30]

ever, we identified significant heterogeneity among studies

(p = 0.004, I2 = 94%).
In six studies (seven arms) that reported weight as an
outcome variable, there was a total of 556 participants in
404 p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408

Fig. 1 – (a–k) Forest plots of randomized controlled trials investigating the effects of Medical nutrition therapy.
 p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408 405

Fig. 1 – (Continued)

intervention group and 534 participants in control group For TG, the combined effect size using a random effects
[28–30,32,34,38]. Pooled results using random effects model model was -8.82 mg/dl (95% CI −20.10, 2.45) [30,31,34,35,37,38].
showed a significant reduction in weight in the MNT group We identified significant heterogeneity between studies
compared with the control group (WMD: −1.54 kg, 95% CI: (p = 0.03, I2 = 57%).
−2.44, −0.64). There was significant heterogeneity among Four studies reported LDL as outcome variables
included studies (p = 0.001, I2 = 77%). [29,31,34,35]. MNT group showed significant reduction in
Six studies (seven arms) containing 240 participants in LDL levels (WMD: −4.43 mg/dl, 95% CI: −13.66, 4.80) compared
MNT group and 242 participants in control group reported BMI with the control group. There was no significant heterogeneity
as an outcome variable [29,30,33–35,37]. Pooled results showed between studies (p = 0.12, I2 = 47%).
that BMI was reduced in the intervention group compared Overall, results from six studies [29,31,34,35,37,38] con-
with the control group (WMD: −0.34 Kg/m2, 95% CI: −0.52, ducted among 605 participants were combined using a
−0.17). There was no significant heterogeneity among studies random-effects model. The results showed that MNT did not
(p = 0.39, I2 = 4.5%). have any significant effect on HDL levels (WMD: −0.40 mg/dl,
Five studies reported waist circumstance as an outcome 95% CI: −3.20, 2.40). Significant heterogeneity was found
measure [29,32,34,37,38]. In comparison with the control among studies (p = 0.001, I2 = 89).
group, the MNT group showed a significant reduction in Seven studies providing 694 participants reported SBP as
waist circumstance (WMD: −2.16 cm, 95% CI: −4.09, −0.23). an outcome variable [29,30,33–35,37,38]. Results pooled using
There was significant heterogeneity among included studies random effects model demonstrated a significant reduction
(p = 0.001, I2 = 93). in the MNT group compared with the control group (WMD:
Eight studies containing a total of 748 participants (382 −7.90 mmHg, 95% CI: −13.03, −2.77). There was significant
participants in intervention group and 366 participants in con- heterogeneity among studies (p = 0.001, I2 = 93). Furthermore,
trol group) reported cholesterol levels as an outcome measure using a random-effects model we combined results of stud-
[29–31,34–38]. We combined results using a random effects ies evaluating the impact of nutritional therapy on DBP
model and demonstrated a significant reduction in cholesterol [29,30,33–35,37,38]. The results indicated a significant reduc-
levels following MNT intervention (WMD: −4.06 mg/dl, 95% CI: tion in DBP levels (WMD: −2.60 mmHg, 95% CI: −4.27, 0.94) with
−7.31, −0.81) (Fig. 1). There was no significant heterogeneity significant heterogeneity among included studies (p = 0.001,
between studies (p = 0.29, I2 = 16). I2 = 72).
406 p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408

95% CI between calculated SESs for MNT intervention studies

Table 3 – Results of subgroup analysis of included
randomized controlled trials in meta-analysis. (Supplemental Fig. 5).

Variables Region by continent

North America Eurasian
continents
4. Discussion
FBS
Number of studies 3 5 This systematic review evaluated the effectiveness of medical
Weighted mean difference −0.54 −13.59 nutrition in developing the treatment plan for patients with
(WMD) diabetes. We found evidence to support the effectiveness of
95% CI −5.25, 4.17 −20.65, −6.53 medical nutrition therapy (MNT) on almost all anthropometric
p-heterogeneity 0.88 0.01
and biochemical outcomes, expect high and low density lipo-
HbA1C proteins. However, the clinical relevance of this differences
Number of studies 5 3 remains unknown as most interventions involved multiple
Weighted mean difference −0.31 −0.59
components. Furthermore, the duration of the intervention
(WMD)
across included studies suggest that multiple encounters may
95% CI −0.59, 0.04 −0.85, -0.33
p-heterogeneity 0.01 0.01 be required to observe desired changes. However, lifestyle
interventions such as MNT, which demonstrate small effect
Weight
may exert large impact at a population level.
Number of studies 3 4
Weighted mean difference −0.92 −1.98
Results of our sub-group analysis showed higher improve-
(WMD) ment in some biochemical parameters - FBS, HBA1C and
95% CI −2.01, 0.17 −3.20, −0.75 weight gain among participants in the Eurasian region
p-heterogeneity 0.04 0.01 compared with their North American counterparts. This dif-
ferences in outcomes between participants from different
continents might be due to genetic variations and differences
in lifestyle. Suggesting that health practitioners may need to
3.4. Subgroup analysis and meta-regression tailor their approach when managing different patient groups.
Although, we did not observe differences in outcome with a
Table 3 shows results of the subgroup analyses. Studies were sub-group analysis after meta-regression for age or duration of
stratified based on continents (North American and Eurasian intervention, this may have been masked by the small number
continents). Subgroup analyses showed that Medical Nutri- of studies evaluating long-term outcomes. Sub-group analysis
tional therapy had significantly higher effects on FBS, HBA1C, by intervention duration was limited by the nature of the data
and weight reduction among participant within Eurasian con- retrieved from studies included in this review.
tinent compared with their North American counterparts. We
did not consider subgroup analysis for BMI, Cholesterol, and
4.1. Comparison with previous findings
LDL because these outcomes showed negligible heterogeneity.
Also, we did not consider subgroup analysis for waist circum-
A previous systematic review by Moller et al. [39] showed sim-
stance, TG, HDL, SBP, and DBP because there were insufficient
ilar results to our study. At short-term (6–12 months). The
studies within these subgroups.
authors [39] reported improvement in HbA1c 0.55 (95% CI:
We performed meta regression analyses on FBS, HBA1C,
0.02–1.1) BMI, 2.1-kg (95% CI: 1.2–2.9-kg) and LDL cholesterol
weight, BMI, Cholesterol, LDL, waist circumstance, TG, HDL,
0.17-mmol/L (95% CI: 0.11–0.23-mmol/L) due to nutrition ther-
SBP, and DBP based on age of participants and duration of
apy compared to dietary advice. However, the size and number
intervention (Supplemental Figs. 2 and 3) but there is no
of studies included in that review, as well as the methodolog-
significant relationship between changes in outcome and par-
ical qualities of studies made it difficult to provide definite
ticipants age or intervention duration.
conclusions about the effectiveness of nutrition therapy. Cur-
rent guidelines for managing diabetes recommend MNT for
3.5. Publication bias and sensitivity analysis patients with diabetes. In this study we found evidence to
support the multi-faceted impact of MNT shown by previ-
The Funnel plot (Supplemental Fig. 4), Egger’s and Begg’s ous reports [40,41]. Also, there is evidence to recommend
tests did not show any publication bias for FBS (p = 0.39, the provision of MNT by qualified health care profession-
p = 0.32), HBA1C (p = 0.09, p = 0.45), Weight (p = 0.13, p = 0.17), als (i.e. Dietitians) and incorporate tailored approaches to
BMI (p = 0.14, p = 0.29), waist circumstance (p = 0.05, p = 0.32), yield sustainable outcomes. Medical nutrition therapy has
TG (p = 0.17, p = 0.85), and HDL (p = 0.93, p = 0.85) (Supplemen- also been recommended by American guidelines, document-
tal Fig. 4). There was a significant publication bias (Egger’s ing evidence of clinical benefits for physiological parameters
and Begg’s tests) for cholesterol (p = 0.02, p = 0.13), LDL (p = 0.01, and disease management [42]. However, the evidence to sup-
p = 0.04), SBP (p = 0.04, p = 0.65), and DBP (p = 0.02, p = 0.02). We port the long-term provision of this therapy remains scarce
used ‘trim and fill’ method for adjusting publication bias but and at best uncertain [41]. Considering the resource intensive
did not detect potentially missing studies that could have nature of MNT, studies evaluating long-term clinical benefits
biased the results of this meta-analyses. Sensitivity analysis and cost-effectiveness of MNT are needed to inform guide-
did not show any significant differences beyond the limits of lines.
 p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408 407

4.2. Limitations
Conflict of interest
Although, the results of this review were generated using
The authors declare no conflict of interest.
evidence from a larger total sample size, there were some lim-
itations with the review process. The results of this review
are limited by inherent bias in the original studies used to Appendix A. Supplementary data
synthesis evidence on MNT. A number of studies included
in this review were heterogeneous used short-term follow-up Supplementary material related to this article can be
durations and varying assessment techniques. There was sig- found, in the online version, at doi:https://doi.org/10.1016/
nificant heterogeneity between most studies used to evaluate j.pcd.2019.05.001.
these outcomes that remained even after sensitivity anal-
ysis. Most studies showed high or unclear risk of bias for
references
critical domains of randomization or allocation concealment.
This may have inflated the effect size observed, limiting the
strength of the evidence and consequently, the recommenda-
[1] E.A. Spencer, K.L. Pirie, R.J. Stevens, et al., Diabetes and
tions generated by this review. We performed a comprehensive modifiable risk factors for cardiovascular disease: the
search process and assessments for publication bias, poorly or prospective Million Women Study, Eur. J. Epidemiol. 23 (2008)
non-indexed papers could still have been missed that could 793.
have influenced the results of our study. Also, we were limited [2] T. Schramm, G. Glslason, L. Kober, S. Rasmussen, J.N.
by the reporting characteristics nature of the studies included Rasmussen, S.Z. Abildstrom, et al., Diabetes patients
requiring glucose-lowering therapy and nondiabetics with a
in this review. We were unable to retrieve complete outcome
prior myocardial infarction carry the cardiovascular risk. A
data despite attempts to contact corresponding authors.
population study of 33 million people, Circulation 117 (2008)
Lifestyle interventions such as medical nutrition ther- 1945–1954.
apy are mostly delivered as part of a multi-component [3] Organization WH, Global Report on Diabetes, 2016.
management strategies. The synergistic effects from other [4] W. Yang, W. Zhao, J. Xiao, et al., Medical care and payment
complimentary interventions, cultural and contextual fac- for diabetes in China: enormous threat and great
tors may have contributed to the weight of the effect sizes opportunity, PLoS One 7 (2012), e39513.
[5] Association AD, Economic costs of diabetes in the US in
observed. In this study, we did not adjust for known or
2012, Diabetes Care 36 (2013) 1033–1046.
unknown mediating factors that could explain a significant
[6] N. Hex, C. Bartlett, D. Wright, et al., Estimating the current
part of the intervention effect we observed. Therefore, we and future costs of Type 1 and Type 2 diabetes in the UK,
advise caution when interpreting the results of this study as including direct health costs and indirect societal and
mediating factors such as physical activity levels were not productivity costs, Diabet. Med. 29 (2012) 855–862.
adjusted for and most studies were evaluated at short-term [7] M.G. Albano, C. Crozet, J. d’Ivernois, Analysis of the
follow-up. Future studies should explore the effectiveness of 2004–2007 literature on therapeutic patient education in
diabetes: results and trends, Acta Diabetol. 45 (2008) 211–219.
medical nutrition using robust designs in a clearly defined pre-
[8] M.J. Davies, S. Heller, T. Skinner, et al., Effectiveness of the
diabetic or diabetic patient group and use longer follow up diabetes education and self management for ongoing and
duration. newly diagnosed (DESMOND) programme for people with
newly diagnosed type 2 diabetes: cluster randomised
controlled trial, BMJ 336 (2008) 491–495.
[9] L. Blonde, J. Dempster, J.M. Gallivan, et al., Reducing
cardiovascular disease risk in patients with diabetes: a
5. Conclusions message from the National Diabetes Education Program, J.
Am. Acad. Nurse Pract. 18 (2006) 524–533.
This systematic review and meta-analysis found evidence to [10] Committee CDACPGE, Clinical practice guidelines for the
support incorporating medical nutrition therapy as a cen- prevention and management of diabetes in Canada.
tral component of the management of diabetes mellitus. Canadian Diabetes Association, Can. J. Diabetes 27 (2003)
S10–S13.
Improvements were observed in outcome measures of FBS,
[11] S. Farsaei, A.M. Sabzghabaee, A.H. Zargarzadeh, et al., Effect
HBA1C, weight, BMI, waist circumference, cholesterol, and of pharmacist-led patient education on glycemic control of
SBP. Sub-group analysis showed difference by continents, but type 2 diabetics: a randomized controlled trial, J. Res. Med.
no significant associations were found for the effect of age or Sci. 16 (2011) 43.
intervention duration. Future studies should explore the effect [12] M.J. Franz, The role of nutrition therapy and dietitians in the
of MNT as an early preventative intervention in pre-diabetic management of the metabolic syndrome, Curr. Diab. Rep. 7
patient groups, evaluate long-term maintenance effects and (2007) 60–65.
[13] L.M. Gaetke, M.A. Stuart, H. Truszczynska, A single nutrition
self-management approaches among diabetic patient groups.
counseling session with a registered dietitian improves
short-term clinical outcomes for rural Kentucky patients
with chronic diseases, J. Am. Diet. Assoc. 106 (2006) 109–112.
[14] J.D. Goldhaber-Fiebert, S.N. Goldhaber-Fiebert, M.L. Tristán,
et al., Randomized controlled community-based nutrition
Funding and exercise intervention improves glycemia and
cardiovascular risk factors in type 2 diabetic patients in rural
No fund. Costa rica, Diabetes Care 26 (2003) 24–29.
408 p r i m a r y c a r e d i a b e t e s 1 3 ( 2 0 1 9 ) 399–408

[15] W.L. Haskell, K. Berra, E. Arias, et al., Multifactor [30] H. Liu, M. Zhang, X. Wu, et al., Effectiveness of a public
cardiovascular disease risk reduction in medically dietitian-led diabetes nutrition intervention on glycemic
underserved, high-risk patients, Am. J. Cardiol. 98 (2006) control in a community setting in China, Asia Pac. J. Clin.
1472–1479. Nutr. 24 (2015) 525–532.
[16] M. Trento, M. Basile, E. Borgo, et al., A randomised controlled [31] A.R. Parker, L. Byham-Gray, R. Denmark, et al., The effect of
clinical trial of nurse-, dietitian-and pedagogistled Group medical nutrition therapy by a registered dietitian
Care for the management of Type 2 diabetes, J. Endocrinol. nutritionist in patients with prediabetes participating in a
Invest. 31 (2008) 1038–1042. randomized controlled clinical research trial, J. Acad. Nutr.
[17] J. Ma, K. Berra, W.L. Haskell, et al., Case management to Diet. 114 (2014) 1739–1748.
reduce risk of cardiovascular disease in a county health care [32] K. Niswender, M. Piletic, H. Andersen, et al., Weight change
system, Arch. Intern. Med. 169 (2009) 1988–1995. upon once-daily initiation of insulin detemir with or
[18] J. Wylie-Rosett, L. Delahanty, Group DPPR, An integral role of without dietary intervention in overweight or obese
the dietitian: implications of the Diabetes Prevention insulin-naïve individuals with type 2 diabetes: results from
Program, J. Am. Diet. Assoc. 102 (2002) 1065. the DIET trial, Diabetes Obes. Metab. 16 (2014) 186–192.
[19] A. Al-Shookri, G. Khor, Y. Chan, et al., Effectiveness of [33] S.T. Miller, V.J. Oates, M.A. Brooks, et al., Preliminary efficacy
medical nutrition treatment delivered by dietitians on of group medical nutrition therapy and motivational
glycaemic outcomes and lipid profiles of Arab, Omani interviewing among obese African American women with
patients with type 2 diabetes, Diabet. Med. 29 (2012) 236–244. type 2 diabetes: a pilot study, J. Obes. 2014 (2014) 345941.
[20] M.-C. Battista, M. Labonté, J. Ménard, et al., [34] M.C. Battista, M. Labonté, J. Ménard, et al., Dietitian-coached
Dietitian-coached management in combination with annual management in combination with annual endocrinologist
endocrinologist follow up improves global metabolic and follow up improves global metabolic and cardiovascular
cardiovascular health in diabetic participants after 24 health in diabetic participants after 24 months, Appl.
months, Appl. Physiol. Nutr. Metab. 37 (2012) 610–620. Physiol. Nutr. Metab. 37 (2012) 610–620.
[21] R.S. Bhopal, A. Douglas, S. Wallia, et al., Effect of a lifestyle [35] M.C. Huang, C.C. Hsu, H.S. Wang, et al., Prospective
intervention on weight change in south Asian individuals in randomized controlled trial to evaluate effectiveness of
the UK at high risk of type 2 diabetes: a family-cluster registered dietitian-led diabetes management on glycemic
randomised controlled trial, Lancet Diabetes Endocrinol. 2 and diet control in a primary care setting in Taiwan,
(2014) 218–227. Diabetes Care 33 (2010) 233–239.
[22] S. Mohammadi, S.R. Arefhosseini, M.A. Jafarabadi, et al., [36] B.D. Timmerberg, J. Wurst, J. Patterson, et al., Feasibility of
Regulation of serum lipid profile, glucose, insulin, and using videoconferencing to provide diabetes education: a
adiponectin in obese diabetic women under diet therapy: a pilot study, J. Telemed. Telecare 15 (2009) 95–97.
randomized clinical controlled study, Iran. Red Crescent [37] M. Trento, M. Basile, E. Borgo, et al., A randomised controlled
Med. J. 19 (2017). clinical trial of nurse-, dietitian- and pedagogist-led
[23] D. Moher, A. Liberati, J. Tetzlaff, et al., Preferred reporting Group Care for the management of Type 2 diabetes, J.
items for systematic reviews and meta-analyses: the Endocrinol. Invest. 31 (2008) 1038–1042.
PRISMA statement, PLoS Med. 6 (2009), e1000097. [38] J. Eriksson, J. Lindström, T. Valle, et al., Prevention of Type II
[24] J.P. Higgins, D.G. Altman, P.C. Gøtzsche, et al., The Cochrane diabetes in subjects with impaired glucose tolerance: the
Collaboration’s tool for assessing risk of bias in randomised Diabetes Prevention Study (DPS) in Finland. Study design
trials, BMJ 343 (2011), d5928. and 1-year interim report on the feasibility of the lifestyle
[25] M. Borenstein, L.V. Hedges, J. Higgins, et al., References, intervention programme, Diabetologia 42 (1999) 793–801.
Wiley Online Library, 2009. [39] G. Moller, H.K. Andersen, O. Snorgaard, A systematic review
[26] M. Egger, G.D. Smith, M. Schneider, et al., Bias in and meta-analysis of nutrition therapy compared with
meta-analysis detected by a simple, graphical test, BMJ 315 dietary advice in patients with type 2 diabetes, Am. J. Clin.
(1997) 629–634. Nutr. 106 (2017) 1394–1400.
[27] T.M. Palmer, J.L. Peters, A.J. Sutton, et al., Contour-enhanced [40] E.B. Lynch, R. Liebman, J. Ventrelle, et al., Peer reviewed: a
funnel plots for meta-analysis, Stata J. 8 (2008) 242. self-management intervention for African Americans with
[28] P.Z. Marincic, M.V. Salazar, A. Hardin, S. Scott, S.X. Fan, P.R. comorbid diabetes and hypertension: a pilot randomized
Gaillard, C. Wyatt, L. Watson, P. Green, P. Glover, M. Hand, controlled trial, Prev. Chronic Dis. 11 (2014).
Diabetes self-management education and medical nutrition [41] E.B. Lynch, R. Liebman, J. Ventrelle, et al., Design of the
therapy: a multisite study documenting the efficacy of Lifestyle Improvement through Food and Exercise (LIFE)
registered dietitian nutritionist interventions in the study: a randomized controlled trial of self-management of
management of glycemic control and diabetic dyslipidemia type 2 diabetes among African American patients from
through retrospective chart review, J. Acad. Nutr. Diet. 119 (3) safety net health centers, Contemp. Clin. Trials 39 (2014)
(2019) 449–463. 246–255.
[29] S. Mohammadi, S.R. Arefhosseini, M.A. Jafarabadi, et al., [42] M.J. Franz, M.A. Powers, C. Leontos, et al., The evidence for
Regulation of serum lipid profile, glucose, insulin, and medical nutrition therapy for type 1 and type 2 diabetes in
adiponectin in obese diabetic women under diet therapy: a adults, J. Am. Diet. Assoc. 110 (2010) 1852–1889.
randomized clinical controlled study, Iran. Red Crescent
Med. J. 19 (2017).