Micronutrient deficiencies, also known as hidden hunger, have been said to affect 1 in 2 children and 2 in 3 women worldwide [1]. Micronutrient deficiencies contribute to poor growth, intellectual impairment, and increased risk of morbidity and mortality [2]. Having timely, valid data on micronutrient biomarkers is essential to justify, plan, and implement successful nutrition intervention programs [3]. One of the biggest challenges in interpreting micronutrient deficiencies is that inflammation can affect many micronutrient biomarkers and can therefore lead to incorrect diagnosis of individuals and to over- or underestimation of the prevalence of micronutrient deficiency in a population [4]. To improve the interpretation of micro-nutrient biomarkers in areas with high inflammation and to generate context-specific estimates of risk factors for anemia [4], a multiagency research group, the Biomarkers Reflecting Inflammation and Nutritional Determinants of Anemia (BRINDA) project, was formed in 2012 by CDC, Eunice Kennedy Shriver National Institute of Child Health and Human Development, and Global Alliance for Improved Nutrition, with the support from the Bill and Melinda Gates Foundation and under the secretariat of Emory University [5].

BRINDA’s primary goal is to improve global micronutrient assessment. Reliable data are essential for both policy making and practice to ensure that programs are appropriately designed and that nutritionally deficient populations are appropriately targeted. To date, BRINDA has compiled a global micronutrient database with 30 datasets on preschool-age children (PSC) (n = 20,000) from 26 countries and 24 datasets on nonpregnant women of reproductive age (WRA) (n = 56,000) from 23 countries. Using its rich data source, BRINDA published a series of papers to provide methods to adjust for inflammation on key micronutrient biomarkers, such as biomarkers for vitamin A, iron, folate, vitamin B-12, zinc, and vitamin D [4,6–13]. BRINDA’s effort to systematically address the role of inflammation in assessing micronutrient status has been recognized by the international community and is included in WHO guidelines [14] and CDC’s Micronutrient Manual and Toolkit [15].

The BRINDA inflammation adjustment method was developed to serve a wide range of users from the global nutrition community, including policy analysts who aid national nutrition strategy planning, researchers who study micronutrients, and field practitioners in charge of monitoring and evaluating micronutrient interventions. However, practically applying the BRINDA inflammation adjustment method to data can be complex. There is varying guidance on the adjustment approaches for different biomarkers. In addition, for the same iron and vitamin A bio-markers, according to the BRINDA’s initial publication [7], malaria status was added as a covariate and was also included in the WHO ferritin guideline [14]; however, later BRINDA papers on adjusting the role of inflammation on vitamin A, iron, and zinc [8, 9,11] did not recommend adding malaria as a binary covariate. Furthermore, users are required to develop statistical programming to conduct these analyses and must consider the time and technical skills in statistics and programming required to do so. Therefore, the time spent learning the BRINDA inflammation adjustment method and compiling correct programming codes may result in lost opportunities to have timely micronutrient data results during critical decision-making periods.

The need for global efforts to standardize guidance for micronutrient analysis and interpretation has been repeatedly emphasized by recent nutrition initiatives [3,16,17]. The objectives of this paper are to 1) provide an all-in-one summary of the BRINDA inflammation adjustment method, 2) evaluate whether malaria as a binary variable should be included in the BRINDA inflammation adjustment method, and 3) present user-friendly, standardized BRINDA inflammation adjustment R package and SAS macro. It is important to mention that both the BRINDA inflammation adjustment method and this paper focus on WRA and PSC. However, the general principle of the BRINDA method can apply to other population groups.

In the past few years, BRINDA has published a series of papers providing guidance on how to adjust micronutrient biomarkers, namely retinol binding protein (RBP) [8,9], serum retinol [8,9], serum ferritin [8], soluble transferrin receptor (sTfR) [10], serum zinc [11], serum and red blood cell (RBC) folate [12], serum B-12 [12], and vitamin D [13], using inflammation markers α-1-acid glycoprotein (AGP) and/or C-Reactive Protein (CRP) [6]. In this section, we will first present whether these micronutrient biomarkers should be adjusted for inflammation, and if so, by which inflammation markers (both AGP and CRP, only AGP, or only CRP). Second, we will summarize the procedure for applying the BRINDA inflammation adjustment method. Third, we will discuss whether including malaria as a binary variable in the BRINDA inflammation adjustment method is necessary. It has been widely recognized that serum and plasma values for micronutrient biomarkers such as RBP, retinol, ferritin, sTfR, zinc, folate, and B-12 have approximately the same values. Hereafter, we will use serum micronutrient biomarkers to refer to both plasma and serum micronutrient biomarkers.

The BRINDA inflammation adjustment method should only be used when there is both biological and statistical evidence of a relationship between micronutrient biomarkers and inflammation markers in the population being studied. It should not be applied indiscriminately to populations with high or low levels of inflammation without first establishing such a relation. Further, evidence suggests that inflammation may confound micronutrient biomarkers even in settings with low infection burden [18]. Based on current evidence in WRA and PSC groups, not all micronutrient biomarkers require adjustment for inflammation, nor do they need to be adjusted by both AGP and CRP. A schematic for when micronutrient biomarkers should be adjusted by which inflammation marker(s) is summarized in Figure 1. For example, serum and RBC folate and vitamin B-12 do not need to be adjusted for inflammation in either WRA or PSC [12], whereas sTfR should only be adjusted for AGP (but not CRP) [10] in both groups. Serum zinc does not routinely need adjustment for inflammation in WRA [11]. However, in PSC, serum zinc should be adjusted by both AGP and CRP only if Spearman correlation between serum zinc and either AGP or CRP is a negative (r < −0.1) and marginally significant (P < 0.1) [11]. A decision tree on when serum zinc should be adjusted is presented in Figure 2. If only one inflammation biomarker is available for micronutrient biomarkers that need to be adjusted by both AGP and CRP, it is still recommended to use that inflammation marker to adjust for inflammation.

In BRINDA’s initial publication [7], malaria status was added as a covariate in the BRINDA inflammation adjustment method and was also included in the WHO ferritin guideline [14]. Later BRINDA analysis showed that including malaria only altered the adjusted values of retinol and RBP slightly [8,9]. In this study, the role of malaria as a binary variable on a series of micronutrient biomarkers was evaluated using the latest BRINDA database. Five datasets in WRA (n = 4768) and 8 datasets in PSC (n = 6662) for ferritin, 5 datasets in WRA (n = 4768) and 7 datasets in PSC (n = 6251) for sTfR, 7 datasets in PSC (n = 6251) for RBP, 3 datasets in PSC (n = 1023) for serum retinol, and 2 datasets in PSC (n = 1841) for zinc were used in the analysis (data source is shown in Supplementary File 1). Overall, the median prevalence of iron deficiency in WRA was very similar in comparing the BRINDA adjustment method with and without malaria terms (ferritin: 18.8% compared with 19.2%; sTfR: 77.8% compared with 77.7%). Similarly, in PSC, when estimating iron deficiency using ferritin and sTfR, vitamin A deficiency using RBP and retinol, and zinc deficiency using serum zinc, the prevalence estimates of micronutrient deficiencies were similar using the BRINDA inflammation adjustment method with or without malaria as a binary variable (Figure 4). Therefore, adding malaria as a binary variable in the BRINDA inflammation adjustment for biomarkers of iron, vitamin A, and zinc is not necessary.

The BRINDA inflammation adjustment R package and SAS macro are user-friendly, all-in-one statistical analysis software that facilitate the application of the BRINDA adjustment method for 5 micronutrient biomarkers, namely, serum RBP, retinol, ferritin, sTfR, and zinc (when appropriate), using AGP and CRP for various population groups. Because it is not recommended to adjust for inflammation for serum and RBC folate and vitamin B-12, the package and macro do not include those micronutrient biomarkers. The BRINDA inflammation adjustment R package and SAS macro have 3 main functions: 1) evaluating whether applying the BRINDA adjustment method is needed for certain micronutrient biomarkers based on previous publications (Figure 1); 2) determining whether zinc needs adjustment based on the correlation between zinc and inflammation markers from users’ data (Figure 2); and 3) applying the BRINDA inflammation adjustment method (Figure 3 steps 4 to 6).

The BRINDA inflammation adjustment R package and SAS macro do not include functionality for data checks nor determining the statistical relation between micronutrient biomarkers and inflammation for user’s data. These steps (Figure 3 steps 1–3) require visualization and research-specific knowledge, such as information on LoDs and laboratory methods, which are beyond the scope of the package or macro. Users are responsible for completing these steps prior to applying the BRINDA inflammation adjustment R package and SAS macro to their data.

An overview of inputs, core content, and outputs for the BRINDA inflammation adjustment R package and SAS macro are illustrated in Supplementary File 2. The training materials and video tutorials for using the package and macro are available on the BRINDA website (https://www.brinda-nutrition.org/the-brinda-approach) [21,22]. In short, users need to specify the 1) name of their dataset, 2) variable names of the micronutrient and inflammation markers, 3) population group of the dataset, 4) user specified AGP and CRP reference values (which can be input only if the population group is “manual”), and 5) output format. After a successful run, the BRINDA inflammation adjustment package will generate a new dataset containing additional variables of adjusted micronutrient biomarkers (by default). If users specify the output format as “full,” the output dataset will also include additional variables such as coefficients of regressions of micronutrient biomarkers on AGP and CRP and natural logs of AGP/CRP reference values. It is of note that the BRINDA software allows users to input one inflammation marker (AGP or CRP) instead of both inflammation markers and provides the adjusted micronutrient biomarker values accordingly.

This paper provides current and practical guidance on applying the BRINDA inflammation adjustment method, and the BRINDA inflammation adjustment R package and SAS macro provide a streamlined structure that can shorten the time spent applying the BRINDA method.

Addressing the role of inflammation in assessing micro-nutrient status is essential to interpret micronutrient deficiencies correctly [14]. The BRINDA method has been used in various studies and national surveys [23–25] and included in WHO ferritin guidelines [14] and CDC’s Micronutrient Manual and Toolkit [15]. The evidence gained from the BRINDA project has provided important insights to research, policy, and programming [26]. The role of malaria as a binary variable does not affect the results of BRINDA inflammation adjustment on a series of micronutrient biomarkers, namely, ferritin, sTfR, retinol, RBP, and zinc, using the latest BRINDA database. However, the stage of malaria infection may influence the results for the BRINDA inflammation adjustment method [24] and warrants more consideration in the future. Although the BRINDA inflammation adjustment method has not been applied within clinical settings, research on its utility in this setting is warranted.

In this paper, the open-access BRINDA inflammation adjustment R package and SAS macro were introduced. The package and macro, available sample code, video, and training materials [21,22] can further enhance the use of the BRINDA inflammation adjustment methods by a diverse audience and can shorten the time between data collection and interpretation. Although similar packages for STATA and SPSS software have not been developed, an interactive web portal on the BRINDA inflammation adjustment method is being developed. This can allow users to apply the BRINDA inflammation adjustment without the use of any software. Users will 1) upload their data to an encrypted website portal and 2) select the variable names of micronutrient and inflammation markers and desired cut-off values, and then BRINDA inflammation-adjusted micronutrient values and a summary of micronutrient deficiencies will be downloaded.

In conclusion, this paper provides clear guidance and introduces the user-friendly R package and SAS macro to apply the BRINDA inflammation adjustment method, which can facilitate micronutrient biomarker analysis for research and to inform nutrition policies and programs.