Dietary Fiber for the Treatment of Type 2 Diabetes Mellitus

A Meta-analysis

Robert E. Post, MD, MS; Arch G. Mainous III, PhD; Dana E. King, MD, MS; Kit N. Simpson, DrPH

Disclosures

J Am Board Fam Med. 2012;25(1):16-23. 

In This Article

Methods

Data Sources and Searches

A search of PubMed materials dated January 1, 1980, to December 31, 2010, was performed on February 9, 2011, using the keywords "dietary fiber" and "diabetes mellitus." A search of OVID, the Cochrane Clinical Register of Controlled Trials, and Cumulative Index to Nursing and Allied Health Literature also was performed using the same keywords. The references within studies that met inclusion criteria were searched for any possible relevant articles that may have been missed by these queries.

Study Selection

Inclusion criteria for this meta-analysis included randomized trials that involved an increase in dietary fiber intake as an intervention, evaluated HbA1c and/or fasting blood glucose as an outcome, used human participants with known type 2 diabetes mellitus, and was written in English. Exclusion criteria included are detailed in Figure 1.

Figure 1.

Flow diagram for selection of studies for meta-analysis.

If a study included multiple arms, where one arm included diet and a medication versus another arm of just dietary intervention, then only the dietary intervention arm was included in the analysis. Patients in crossover trials, in which patients are their own controls, were treated as separate control and intervention arms for analysis.

Data Extraction

Data were extracted from each study and entered into an Excel spreadsheet (Microsoft, Corp., Redmond, WA). Demographic factors included sample population, mean age, mean body mass index, and sex distribution of the control and treatment groups. Outcome measures included final means for HbA1c and fasting blood glucose of the control and intervention groups. For studies that reported standard deviation of the mean, those data were used. For studies that reported standard error of the mean, standard error was multiplied by the square root of the sample population to obtain the appropriate standard deviations. If studies reported fasting blood glucose in units of mg/dL, this was converted to the standardized international unit of mmol/L by multiplying the fasting blood glucose value in mg/dL by 0.0555.

Study quality was assessed using the GRADE assessment as detailed in the Cochrane Handbook.[18] This assessment takes into account the bias of studies as well as their methodology (eg, randomized vs observational) and their limitations to determine a quality grade of high, moderate, low, or very low.

Data Synthesis and Analysis

Meta-analysis for the mean differences was performed with Review Manager (version 5.0.23, The Nordic Cochrane Center, Copenhagen, Denmark). Separate analyses were performed for fasting blood glucose and HbA1c. Because measures of fasting blood glucose are on the same scale (mmol/L), and measures of HbA1c are on the same scale (percent of glycohemoglobin), a standardized mean difference was not necessary. The final means in the control and intervention groups were compared by computing a mean difference using the inverse variance method, where studies with less variance in their effect estimate are given more weight. Both fixed- and random-effects models were used for analysis. Comparison of final means is considered an appropriate analysis method in randomized trials because, in theory, the baseline data of the control and intervention groups in a randomized trial are not statistically different, and therefore final values for each trial arm would be representative of a change from a common baseline.[18] A test of heterogeneity was performed to determine if results from a fixed-effect model could be considered valid. If the test of heterogeneity exhibited P < .05, then only the results of a random effect model would be considered valid. Forest plots each for fasting blood glucose and HbA1c were created. Funnel plots for each analysis to assess publication bias visually were created as well.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....