Table 2.

Validity Characteristics of included studies.

Study	Selection Bias or Detection Bias	Performance or Detection Bias	Attrition Bias	Controlled Variables	Reviewer Notes/Possible Limitations
Bruner et al., 1996 [16]	Randomization clear, double-blinded design	Researchers and participants blinded to allocation	Not all randomized included in outcomes; dropouts and reasons for dropouts reported	Additional vitamins and minerals	All participants had ID. There was no measurement of baseline differences between iron-sufficient and ID adolescents. Serum ferritin was a marker of iron status. Statistical analysis was by group assignment, rather than ferritin response.
Beard et al., 2005 [21]	Randomization clear, double-blinded design	Participants unaware of group allocation	Intention-to-treat protocol utilized to deal with missing data; dropouts reported; however, reasons for dropouts not reported	Age, socioeconomic status, income, education, nutritional intakes	Staff members administered supplementations at health clinics with instructions for the mothers on when to take and when to return, etc. Standardized outcome measures were used.
Murray-Kolb and Beard, 2007 [2]	Randomization clear, double-blind placebo-controlled design	One researcher knew the group allocation; outcome assessors, as well as participants were blinded to group allocation	Not all of those randomized included in outcomes; dropout rates reported; however, reasons for dropouts not reported	Socioeconomic status, grade point average, level of physical activity and other demographics;other vitamins/minerals; inflammation, contraception and menstrual cycle; intellectual ability (IQ) was used as a covariate	Standardized and validated measures of were cognition used. High reliability. Analysis was completed on women who were classified as ferritin responders or non-responders. Determination of the relationship between changes in iron status and changes in cognition. 16 weeks used to allow more time for brain iron concentrations to replenish.
McClung et al., 2009 [16]	Randomization reported, double-blind, placebo-controlled design	Researchers, outcome assessors and participants all blinded to allocation	Unclear if all of those randomized included in outcomes; dropout rates and reasons for dropouts reported		Compliance monitored and assessed as a confounding variable. Research assistants administered capsules to participants each day. Treatment occurred on the background of battle combat training. Randomization procedures were not described.
Verdon et al., 2003 [24]	Randomization clear, double-blind design	Researchers, outcome assessors and participants blind to allocation	Intention-to-treat protocol completed to deal with missing data; dropout rates recorded and reasons for dropout recorded	Other disorders that could explain fatigue	Intervention period was relatively shorter than other included studies. A validated self-administered questionnaire was used, although fatigue was assessed on a visual analogue scale. Adherence to treatment was monitored by an electronic device that recorded the date and time that the pill container was opened
Vaucher et al., 2012 [22]	Multi-centre study, randomization clear, double-blind, placebo-controlled design	Researchers, outcome assessors and participants blinded to allocation	Intention-to-treat protocol completed to deal with missing data		A validated self-administered questionnaire used. Intention-to-treat analysis was used, and attrition and compliance were less of an issue. Dropout rates were not reported
Lambert et al., 2002 [23]	Randomization occurred, but allocation procedures not described	Researcher and participants blinded to allocation	Unclear if randomized participants included in outcomes/dropout rates reported/reasons for dropouts reported		Standardized and validated outcome measures were used.
Devaki et al., 2009 [20]	No randomization described	No blinding methods described	Unclear if randomized participants included in outcomes; dropout rates and reasons for dropout reported		Compliance was monitored as supplements were administered by a supervisor at breakfast and lunch. Potential for performance, selection and detection bias.
Siwek et al., 2009 [17]	Randomization clear	Researchers and participants blinded to allocation			Outpatient sample.
Nowak et al., 2003 [18]	Randomization described	Researchers and participants blinded to allocation	Unclear if randomized participants included in outcome; dropout rates reported however reasons for dropout not reported		Clinical sample, so it is difficult to generalize to wider population. Standardized and validated measures of depressive symptoms were used. Allocation procedures were not described.
Sawada and Yokoi, 2010 [19]	Randomization clear, A double-blind, double-blinded, placebo-controlled design	Researchers and participants blinded to allocation	Information around dropout rates and missing data was not reported	Age, height, body mass, body weight	Pilot study, first of its kind. There are missing data and information around how this was dealt with. Treatment was possibly confounded by the use of multivitamins, alongside zinc