Abstract
Background: Accurate and precise estimates of discretionary salt intake are critical for the design of salt fortification programs and salt reduction interventions.
Objective: To compare four methods of estimating discretionary salt intake among non-pregnant women of reproductive age in Punjab, India.
Methods: One-day, observer-recorded, weighed food records (WFR), household salt disappearance (HHSD) data, duplicate diet composites, and samples of household salt and milk samples were collected from 100 women and repeated in a subset of 40 to adjust for intra-person variation and estimate usual discretionary salt intake. Diet composites were also replicated from 40 randomly selected WFR but prepared without the addition of discretionary salt. The duplicate diet composites' sodium and iodine contents were analyzed using Inductively Coupled Plasma (ICP)-Optical Emission Spectrometry and ICP-Mass Spectrometry, respectively. The iodine content of household salt samples was analyzed using the Ion Selective Electrode method. The association and agreement between the WFR method, the selected reference method, and the HHSD, replicate diet (RD) and iodine (IM) methods was explored using correlation and Bland-Altman analyses.
Results: Mean ± SD discretionary salt intakes according to the WFR, HHSD, RD, and IM methods were 4.7 ± 1.8 g/d, 5.7 ± 2.6 g/d, 4.1 ± 2.1 g/d, and 7.8 ± 5.3 g/d, respectively. The RD method showed the strongest correlation (ρ = 0.76, p < 0.001) and the smallest mean difference ± SD (−0.68 ± 1.25 g/d), with limits of agreement from −3.18 to 1.82 g/d, compared to the WFR method. However, the HHSD method was also moderately correlated (ρ=0.49: p < 0.001) and showed good agreement (0.98 ± 2.12 [-3.27,5.23] g/d) with the WFR despite lower precision.
Conclusions: Though intensive to implement, the WFR and RD methods produce precise estimates of discretionary salt intake. Repeated measurements may improve the precision of the HHSD method for large population-based surveys.