Methods/design
Registration and methodology
The study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) at the National Institute for Health Research (CRD42016053141). The guidelines of PRISMA-P (Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols) were followed while reporting the study protocol.
Inclusion criteria for studies
Type of study
All human or animal studies, including randomized and non-randomized placebo-controlled clinical trials, performed during 1999–2016 will be included. Research with cross-sectional and clustered designs, with or without blinding, will be evaluated. Case reports, as well as quasi-experimental and observational studies, will be excluded. No language limitations will be imposed and papers in languages other than English and Persian will be translated before use.
Participants
The human studies will be eligible if their participants:
-
1.
Aged 18 years and older at the time of type II DM diagnosis;
-
2.
Underwent fasting blood sugar (FBS) or hemoglobin A1c (HbA1c) level measurements, glucose challenge test (GCT), glucose tolerance test (GTT), oral GTT (OGTT), or Glucose, 2 h Post Prandial (2HPP) to confirm hyperglycemia;
-
3.
Used A. nilotica to induce hypoglycemia;
-
4.
Used A. nilotica or placebo for at least 4 weeks; and
-
5.
Completed the treatment course at a rate of more than 70%.
The animal studies will be eligible if their participants:
-
1.
The induced diabetic—animal’s which recived A. nilotica (regardless to the type of diabetes induce).No limitation in sex or race of animals.
-
2.
FBS, OGTT, GCT, and GTT were measured to assess blood glucose levels in animals.
Types of intervention
During the preliminary analysis, studies will be included if they involved:
-
1.
The use of A. nilotica as the intervention;
-
2.
The use of placebo in the control group; and
-
3.
The use of placebo to alleviate symptoms [18]
Primary outcome
The rate of response to treatment (e.g. reductions in blood glucose levels) in the intervention and placebo groups will be regarded as the primary outcome. The outcomes will be measured through FBS, GCT, GTT, and OGTT in all of studies and in addition to these tests, will be measured 2HPP and HbA1c level in human study.
Secondary outcome
The side effects of A. nilotica, determining the most effective plant part, and the mean FBS or HbA1c, GCT, GTT, OGTT, or 2HPP levels will be considered as secondary outcomes.
Search strategies for selecting relevant studies
A search strategy will be adopted to find both published and unpublished articles. The Cochrane Central Register of Controlled Trials, MEDLINE, Google Scholar, EMBASE, ProQuest, Scopus, PsycINFO, and CINAHL databases will be searched using a number of key terms including A. nilotica, hypoglycemia, A. arabica, diabetes, and type II diabetes. Boolean operators “and” and “or” will be applied to make combinations of key terms. The search will involve three stages. During the first stage, a limited preliminary search using some of the mentioned key terms will be conducted in MEDLINE and CINAHL. Relevant articles will be selected based on their title, abstract, and keywords. In the second stage, all databases will be searched using all key terms. Finally, the references of all reports and articles will be searched as additional resources which were not listed in bibliographic databases. In order to find unpublished studies, government reports, protocols, gray literature, and student theses indexed in ProQuest will be searched.
Database of ongoing clinical trials
The following databases will be searched to find ongoing clinical trials:
Searching other resources
Key journals in the field will be manually searched. As mentioned earlier, government reports, student theses, studies published by different research committees, and abstracts presented at various conferences and seminars will also be assessed.
Data collection and analysis
Selection of relevant studies
The author (L.D) will initially evaluate the eligibility of the selected studies by reviewing their titles and abstracts. The two co-authors (F.A and N.R) will then assess the eligibility of the papers by independently evaluating their full texts. The authors will discuss their viewpoints to resolve any cases of disagreement. An external evaluator will be consulted if the discussions fail. Authors of papers whose abstracts are presented on posters will be contacted and asked to send the full text of their papers if possible.
Data extraction and management
Two co-authors (F.A and N.R) will individually assess full texts to extract the required data and enter them into a relevant form. The following pieces of information will be collected [19].
-
1.
Research characteristics including the first author’s name, location and dates of publication and conduct of the study, research design, sample size, and duration of follow-up;
-
2.
Patient characteristics including their age and gender, number of participants, the inclusion and exclusion criteria, keywords definitions, and measurement tools;
-
3.
Intervention details including number of groups, blinding procedure, dose, duration, and type of intervention, determinants of treatment length, causes of treatment discontinuation, and sample loss); and
-
4.
Outcome measures including details about the tools and methods used for the measurement of outcomes, side effects, and serious side effects.
The collected data will be reviewed by the third author and cases of disagreement will be resolved by consulting an external evaluator.
Quality assessment of studies
For quality assessment for human study, the Cochrane Risk of Bias Tool and for animal study the SYRCLE (Systematic Review Centre for Laboratory animal Experimentation) checklist will be use. Two external evaluators will independently. Again, an external evaluator will be consulted if any disagreements occur [18].
Data synthesis
Whenever possible, quantitative data will be pooled in statistical meta-analysis using STATA software. Two models of meta-analysis, i.e. the fixed-effect model and the random-effect model, will be used for outcomes. The fixed-effect model (which is based on the Mantel–Haenszel method) assumes that research samples are taken from populations with the same effect size. It thus weights studies according to the in-study variance. In contrast, the random-effect model assumes research samples are selected from populations with different effect sizes and weights studies based on both in-study and between-study variances (based on their level of heterogeneity). The latter model is more suitable in cases of higher heterogeneity. We will apply Chi square-based Q statistic to determine the between-study heterogeneity for each model. All results will be subject to double data entry. Effect sizes will be presented as odds ratio for categorical data and weighted mean differences for continuous data. Analysis will be performed on 95% confidence intervals of the calculated effect sizes. Chi square tests will be used to evaluate heterogeneity. Whenever statistical pooling is impossible, the findings will be merely provided in tables and figures. Meta-analyses will be performed on animal and human studies separately and the results of analyze will be compared in two groups. Subgroup meta-analyses based on outcome will also be conducted.