The primary objective is to assess the prevalence of malnutrition among aged people in Iran.

The secondary objectives are (1) to evaluate the prevalence of malnutrition among elderly people in Iran according to the assessment tools, study setting, severity of the problem, and regional setting and (2) to evaluate the heterogeneity and determine its potential causes.

In this systematic review, we will include all studies that reported the prevalence of malnutrition among elderly people in Iran or provided main data to calculate this measure. These may be included as descriptive and observational studies. We will include population-based prevention clinical trials, cohort studies, and case-control studies if the baseline data were available and consistent with the objectives of this systematic review.

We will include studies conducted on people of both genders, aged 60 years and above, who live in elderly nursing homes or are community indwelling elders. Studies on hospitalized elderly people will be excluded. There will be no restriction in terms of race and ethnicity.

We will include studies that measured and reported malnutrition by using standard and valid measuring tools (questionnaire, anthropometric, or biological indices).

We will include studies with a minimum sample size of 30.

A comprehensive search will be conducted in local and international databases to obtain published and unpublished articles until September 30, 2018.

PubMed, Scopus, and Web of Science are the international databases that will be searched. In addition, the local databases that will be searched include Iranian Databases of Medical literature, Scientific Information Database, Magiran, Iran Medical Sciences Thesis Database and the Iranian Research Institute for Information Science and Technology.

Furthermore, relevant gray literature (eg, thesis or dissertations and conference papers) will be included by searching in the international and local databases and Google Scholar search engine. We will not consider any language limitation.

We will test the primary search strategy (Multimedia Appendix 1) in PubMed, the main database in the field of medical sciences. At first, we will find equivalent keywords from Medical Subject Headings database and will create appropriate syntaxes for PubMed. Then, we will adopt it for other databases.

To ensure access to all relevant articles and to find those not previously found, the list of included articles, references, and the relevant national and international organizations’ reports, such as WHO, will be searched manually.

The study selection will be done in 2 stages. At first, the search results will be imported into the reference manager software (EndNote X7, Thomson Reuters) for storage and organization. After detecting and deleting duplicates, 2 researchers (SE and HKH) will screen the titles and abstracts of all retrieved documents based on the inclusion and exclusion criteria. Then, the full text of potentially eligible studies will be screened independently by the same researchers to finalize the review and start data extraction. Any disagreement between reviewers will be resolved by discussing or consulting with a third person.

Irrelevant articles, which do not meet the inclusion criteria, will be deleted by providing reasons. All the mentioned steps will be presented on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram [54].

If there is a lack of access to the full text of some studies or if there is any unclear information, the researchers will contact the corresponding author thrice during a period of 7 to 14 days [55,56]. We will report all the missing data if we cannot find them. At the end, all other eligible articles retrieved from other sources will be added to the final list.

At this stage, the 2 researchers will independently extract data from the included articles using data extraction form. The data extraction form includes review information; bibliographic data such as authors, title, year, issue; and study data such as design, setting/context, year, sample size, sampling method, data gathering method, data collection period, participants, and main results. Data extraction will be done by 2 independent reviewers (MM and HKH), and any disagreement will be resolved by discussing or consulting with a third expert person.

The quality assessment and risk of bias of the included studies will be assessed by 2 independent assessors (MN and SE), and the strategy of discussion or consulting with a third reviewer will be applied in case of disagreement. For assessing the observational studies, assessors will use a 10-item rating tool developed by Hoy et al [57] in 2012. This tool has been developed based on literature review, expert opinion, and the researchers’ experiences and has a high interrater agreement in assessing the quality of prevalence studies. On the basis of this, each item scores 0 for the response No and 1 for Yes. The total score will be between 0 and 10 [57]. We will use the Cochrane Collaboration tool for assessing the clinical trials [58]. The result of risk of bias and quality assessment will be presented in a table.

We will deal with missing data using 2 strategies. In case of missing the full text of studies from the review, or if there is any unclear information, the researchers will contact the corresponding author(s) thrice during a period of 7 to 14 days [55,56]. When outcome data are not reported, we will calculate the outcome data if there is additional information in the publications [59,60]. However, if these 2 strategies are not successful, we will analyze only the available data and report all missing data and discuss their impact on the findings [59].

We will assess the heterogeneity among studies by using forest plot, I-square (I²) test, and Cochran Q test [61-63]. The significance level of less than 0.05 and I² equal to or more than 50% will be assumed as a severe heterogeneity. In cases of severe heterogeneity, subgroup analysis considering age, gender, elderly residency, and malnutrition severity will be performed to identify the potential source of heterogeneity.

We will check the possibility of publication bias graphically (funnel plot) and statistically (Egger test) [61,64], using Stata software, version 12 (StataCorp LP). A P value of less than .05 will be considered as statistically significant or nonignorable publication bias.

The main outcome of this review is malnutrition. If there are enough studies reporting the prevalence of malnutrition in elderly people, we will do a meta-analysis to calculate the pooled estimate of prevalence and subgroups analysis (gender, age, and residence place) 95% CIs using Stata software.

On the basis of the level of methodological heterogeneity, we will use fixed or random effects model to conduct meta-analyses [61].

On the basis of the quality of heterogeneity, subgroup analysis will be conducted on the following variables: age, sex, geographical region, residency (nursing home/home), risk (at risk or affected), and severity of malnutrition. By this way, the potential sources of heterogeneity will be identified.

Although the need for doing sensitivity analysis is not often specified before completing the systematic review and meta-analysis [61], we will use it to determine the impact of the quality of the study, study design, sample size, and methods of analysis, if needed. In fact, this is a repeated meta-analysis to assess the robustness of the process of selection and inclusion of the studies [65].