This study was conducted to evaluate the acceptability of QFS to inform a randomized trial of QFS in Southern India (NCT03853304) (60). The parent trial is a randomized, double-blind, efficacy trial of QFS, among 1,000 women 18–49 y and their households in Chittoor, Andhra Pradesh (60). The trial design (60), periconceptional surveillance program (61–63), and pre-intervention biomarker survey (61–63) in this study population have been previously described.

The acceptability study of QFS was conducted in two villages adjacent to the trial catchment area and included a 3-week household-based randomized crossover trial and triangle tests at a central research facility to evaluate sensory discrimination between QFS, DFS, and iodized salt consumed in foods.

Participants were women aged 18–49 y and their households. Individuals were excluded if they did not meet inclusion criteria for the parent randomized trial (i.e., women 18–49 y, not pregnant, not lactating, healthy [not currently diagnosed with conditions such as malaria, active tuberculosis disease, or COVID-19; no self-reported fever, diarrhea, or vomiting in the last 2 weeks; no chronic conditions requiring daily medications], and planned to remain in the catchment area for the study period), and additional criteria relating to sensory discrimination: no mouth lesions/ulcers, dental work in last 2 weeks, history of middle ear surgery, neurological disorders, recent loss or change in smell or taste, or self-reported food allergies. Among households with >1 eligible woman, one woman was randomly selected using an a priori algorithm corresponding to the number of eligible women in the household (e.g., if there were n eligible women, order by ascending age and select the nth one). Programmatically, other eligible adult household members (18–65 y) were also invited to participate in data collection visits (i.e., triangle test, hedonic questionnaire) to include information on other potential household decision-makers in secondary analyses. Regardless of eligibility criteria for the acceptability study, individuals who self-reported as the primary preparer(s) of food in the home (≥18y) were also interviewed (if different from the selected woman) to specifically collect data on household salt use during food preparation and cooking.

The trial protocol was reviewed and approved by the Cornell University Institutional Review Board and the Arogyavaram Medical Centre (AMC) Institutional Ethics Committee and received clearance from the India Council of Medical Research and Health Ministry’s Screening Committee. The parent trial was registered at ClinicalTrials.gov (NCT03853304) and the Clinical Trials Registry of India (REF/2019/03/024479). Written consent was obtained from all participants before enrollment; participants who were unable to read were asked to provide a thumbprint in the presence of a literate witness who signed the consent form. The informed consent process was audio-visually recorded, as recommended by the Government of India for clinical trials.

The micronutrient-fortified premix was manufactured at JVS Foods Pvt. Ltd. in Jaipur, Rajasthan, India, and the salt interventions were blended and packaged locally by Sahayamatha Salt Refinery Pvt. Ltd. in Tamil Nadu, India, using the same protocol as the parent randomized trial (60). For the acceptability study, the three salt intervention arms were each labeled with a randomly assigned color, excluding colors used to label salt in the parent randomized trial; participants and study staff members involved in data collection and field activities were blinded to this assignment. Staff who coordinated salt production and storage were unblinded to the color code assignment but did not participate in data collection or other field activities.

Data collection was conducted in the local language (Telugu) at participants’ homes and at the central data collection facility at AMC using electronic tablets by trained nurse enumerators (64). Data collection instruments were developed, translated, and piloted as part of preparations for the parent trial, including review for accuracy of pilot data from mock participant profiles and practice interviews with nurse enumerators and for cultural context by local staff members. Revisions prior to data collection included updating available answer options (e.g., methods of salt storage) and standardized scripts on recommended salt use and storage. Revisions after data collection in preparation for the parent trial included refining questionnaires to reduce participant burden while maintaining data quality (e.g., self-reported salt use).

A flowchart of household and women recruitment is presented in Figure 1. After informed consent for the household, trained nurse enumerators conducted a household census, including member roster and sociodemographic information (i.e., household income, household dietary diversity score (HDDS; 0 to 12 food groups) (65), household food insecurity access scale (HFIAS) (66); age and sex of household members). Nurse enumerators returned to households with ≥1 woman 18–49 years of age for informed consent, screening, and eligibility determination for the QFS acceptability study. During screening, additional household (i.e., salt use) and individual (e.g., marital status, tobacco use, salt use at the table) data were collected. Eligible women (and their households) were provided an appointment for data collection at the central research facility at AMC for triangle tests, and transportation to AMC was provided. After completing the triangle tests, participants and their households were visited to begin the household-based randomized crossover trial.

COVID-19 safety precautions were followed throughout the study, in accordance with the regulations of the Government of India, the state of Andhra Pradesh, and the U.S. Centers for Disease Control and Prevention. Participants were screened for self-reported COVID-19 symptoms (via phone) before household visits and requested to complete a rapid SARS-CoV-2 antigen test immediately upon arrival at AMC. Study staff and participants were provided masks and requested to wear them during study visits and in-person interviews. Study staff monitored symptoms, and all study staff were instructed to stay home if they had any symptoms or had been potentially exposed to COVID-19 (e.g., travel, contact with someone with COVID-19). Data collection was conducted outdoors and with social distancing, where possible.

Power calculations were conducted a priori and replicated by a statistician otherwise unaffiliated with the study (LMJ). For the triangle tests, a sample size of n=55 participants would enable detection of a difference at 90% power at alpha=0.05, given a chance-adjusted percent correct of 30% (i.e., effect size parameter) (70). For the randomized crossover trial, a sample size of n=54 total households would enable detection of a 0.5-point difference in paired hedonic scale ratings with standard deviation of the difference of 1.1 (i.e., effect size of 0.45), based on mean ratings of DFS and QFS in a rice dish from a previous publication (57), with 90% power at alpha=0.05. Based on these power calculations, we planned to recruit at least 55 women and their households from two pre-selected villages. All eligible households within each village had the opportunity to participate.

Descriptive statistics were presented as n (%), mean (standard deviation [SD]), and predicted population margins (least squares mean [SE or 95% confidence interval]). Primary analyses were conducted among women 18–49 y (1 per household); we also conducted analyses in all participating members of the household (i.e., including women and additional household members). Findings were similar, and analyses among women are presented in tables and results.

A flowchart of households and women is summarized in Figure 1. A total of 138 households had at least one woman aged 18–49 y. After informed consent and screening, a total of 77 households had at least one woman 18–49 y who was eligible for the randomized crossover trial, and 75 women were eligible for the triangle test. A total of 66 women 18–49 y participated in the triangle tests at our central data collection facility at AMC.

Characteristics of women and their households are presented in Table 1 and Supplemental Table 1. The mean age of participants was 35 years; most were married (91%) and had at least some formal schooling (84%). A total of 96% of women were the primary food preparer in the household and 96% reported following a non-vegetarian diet. Although the prevalence of tobacco use was low (3%), 35% of women reported using betel nut (i.e., areca nut, a palm fruit commonly chewed in Southeast Asia and classified as a carcinogen (71)) in the past two weeks.

At baseline, approximately half of women reported using salt at the table at least once per day, while 13% reported never adding salt at the table. Most households reported using both table salt and rock salt (82%). Of households that reported purchasing table salt (91%), most reported using it at the table (87%), while 48% used it for cooking. In contrast, of households that purchased rock salt (91%), most reported using it for cooking (97%) and other non-food purposes (87%), and 38% reported adding it at the table.

Results from the triangle test are presented in Table 2. Food samples (i.e., plain rice, lemon rice) prepared with QFS, DFS, or iodized salt were not distinguishable in pair-wise comparisons (QFS vs. DFS, QFS vs. iodized, DFS vs. iodized; p>0.05).

Results from self-reported sensory differences between food samples are presented in Table 3. When asked to describe the difference between the sample they selected compared to the other two samples, most participants (60–79%) reported a slight difference (vs. moderate or high). Most (85–97%) participants reported a difference in taste; 5–11% reported a difference in color between QFS or DFS and iodized salt. Self-reported sensory differences were consistent between women who selected the correct sample compared to women who selected the incorrect sample for most comparisons. However, women who selected the correct sample were more likely to report that the samples were slightly different (vs. moderate or high) for QFS compared to DFS in lemon rice samples (p=0.01), were more likely to report a difference in odor for DFS compared to iodized salt (p=0.03) and QFS compared to DFS (p=0.01) in plain rice samples, and were less likely to report a difference in taste for QFS compared to iodized salt in lemon rice samples (p=0.01).

Weighed household salt use by intervention type and by week is presented in Table 4. Total household salt use (mean [SD]; week 1: 145.3 [145.2] g; week 3: 198.9 [175.8] g) and salt use per person per day (mean [SD]; week 1: 6.0 [6.4] g; week 3: 9.1 [7.6] g) increased during the 3-week intervention period. Approximately half (week 1: 50.7%; week 3: 67.2%) of households used ≥5 g of salt per person per day, and three-fourths (week 1: 66.2%; week 3: 81.3%) used ≥1 g. Among households using ≥1.0 g of salt per person per day, salt use (mean [SD]) ranged from 9.0 [5.8] grams per person per day during week 1 to 11.1 [6.9] g during week 3.

Approximately one-fourth of households consumed <1.0 g salt per person per day (week 1: 33.8%; week 3: 18.8%). The most frequently reported reasons for not using salt during the study included traveling or not being home (week 1: n=11; week 3: n=4) and forgetting to use the salt (week 1: n=6; week 3: n=3). Other reasons included household member was sick (week 1: n=0; week 3: n=2), primary preparer of food in the household that week did not know to use the salt (week 1: n=1; week 3: n=1), disliking the salt (week 1: n=1; week 3: n=0), concerns regarding using the salt (week 1: n=3; week 3: n=0), or another household member (e.g., husband, father-in-law) did not permit the household to use the salt (week 1: n=0; week 3: n=3) (data not shown).

Total household salt use, salt use per day, and salt use per person per day, did not differ by salt type (QFS vs. DFS) in mixed models that included fixed effects for salt type, randomization sequence, and period. However, during the study period, salt use (regardless of type of salt) increased from week 1 to week 3, including total household salt use (least square mean difference [MD]: 54.4 g; 95% confidence interval [CI]: 19.3, 89.6; p=0.003) and salt use per person per day (MD: 3.0 g; 95% CI: 1.0, 5.1; p=0.005) (Figure 3), indicating a period effect independent of salt type and randomization sequence. Findings were similar in sensitivity analyses among households that consumed ≥1.0 g per person per day.

Hedonic ratings by salt intervention type and week are presented in Table 5 and Figure 4. Acceptability of the salt interventions was high: mean hedonic rating for most of the five sensory domains (color, odor, taste, texture, overall) was ≥7.0 (i.e., like moderately) for each salt type. Hedonic ratings did not differ by salt type (QFS vs. DFS) for any of the sensory domains in mixed models that included fixed effects for salt type, randomization sequence, and period. However, during the study period, hedonic ratings significantly increased from week 1 to week 3 for all sensory domains except texture, indicating a period effect independent of salt type and randomization sequence.

In the current study, foods prepared with QFS, DFS, and iodized salt were not distinguishable in triangle tests, and few participants noted differences between foods prepared when salt was added prior to (plain rice) or after (lemon rice) cooking. To our knowledge, no sensory discrimination studies to date have been conducted to evaluate micronutrient-fortified salt formulated using the method in the current study—i.e., DFS or QFS with microencapsulated iron (ferrous fumarate) via extrusion—to which findings are directly comparable. However, findings are consistent with a study among women (n=18) in Bangalore, India, that used triangle tests to compare two formulations of DFS (micronized ferric pyrophosphate and microencapsulated ferrous fumarate via fluid bed) with iodized salt, prepared in plain rice, lemon rice, and rice with sambar (vegetable stew) (45). Foods prepared with DFS were indistinguishable from iodized salt except in plain rice, in which DFS at higher concentrations (microencapsulated ferrous fumarate via fluid bed; 2 mg iron/g salt) appeared as black specks (45). Findings are also consistent with previous triangle tests conducted among adults in Morocco, comparing traditional foods prepared with formulations of DFS [micronized ferric pyrophosphate (46) and microencapsulated ferrous fumarate via fluid bed method (47)] and salt fortified with vitamin A, iodine, and iron using micronized ferric pyrophosphate (72), to iodized salt. To our knowledge, no studies to date have used sensory discrimination tests to evaluate the acceptability of quadruple-fortified salt (i.e., iodine iron, folic acid, and vitamin B₁₂) compared to DFS or iodized salt.

In the current study, individual hedonic ratings were high, and did not differ by salt type in any of the sensory domains evaluated. Although other studies have reported high acceptability of multiple micronutrient-fortified salt, color differences were reported. In a 10-month randomized trial among children and their households in Bangalore, India, >98% of surveyed households (n=345) reported that the salt intervention (i.e., DFS with micronized ferric pyrophosphate, DFS with microencapsulated ferrous fumarate via fluid bed, vs. iodized salt) was acceptable, including color in both damp and dry seasons, taste, and overall acceptability (45). However, households receiving DFS were more likely to report a slight color change in foods (i.e., 26% [DFS with micronized ferric pyrophosphate] or 59% [DFS with microencapsulated ferrous fumarate via fluid bed]), compared to iodized salt (4%) (45). Similarly, other randomized trials (8–10 months) in Morocco reported high acceptability but color changes in foods with multiple micronutrient-fortified salt [DFS with micronized ferric pyrophosphate (46); DFS with microencapsulated ferrous fumarate via fluid bed method (47); vitamin A, iodine, and iron using micronized ferric pyrophosphate (72)] compared to iodized salt. In a study in New Delhi, India, women (≥18 y; n=26) gave high hedonic ratings (≥3.7 using a 5-point scale) to foods prepared with quintuple-fortified salt (Q5FS; zinc + iodine, iron, vitamin B₁₂, and folic acid) and DFS with microencapsulated ferrous fumarate via extrusion, and iodized salt (57). However, Q5FS and DFS had significantly lower ratings for color, compared to iodized salt. Another study in Tanzania provided households (n=899) with one of three salts (QFS [iodine, iron, vitamin B₁₂, and folic acid] or DFS with microencapsulated ferrous fumarate via extrusion, or iodized salt, similar to the current study) to use ad libitum for a week; 78% of all participants reported the intervention salt to be acceptable (vs. neutral or not acceptable) and hedonic ratings did not differ across salt types for all domains evaluated (58). However, the mean hedonic ratings were low (1.5–1.8 using a 5-point scale) across domains for all salt types. Although differences in methods used in previous studies, including study design, target population, salt interventions (i.e., micronutrient composition, formulation, and fortification methods), duration, and setting, constrain direct comparability of findings, the current study also demonstrated high acceptability of QFS.

In the current study, weighed salt use was ~6–9 g per person per day and ~9–11 g among the approximately 75% of households that used ≥1 g of salt per person per day. The salt use per person per day was similar to population-based estimates from adults in rural Andhra Pradesh using 24-hour urine sampling (mean [95% CI]: 9.15 [8.73–9.57] g) (68), which were used to inform power calculations in the parent trial. Salt use in the current study was consistent with previous estimates from randomized trials of DFS in India (mean [SD]; 11.3 [5.1] g via 4-day weighing at baseline (45); DFS: 12.4 [7.9], iodized salt: 15.0 [13.6] via 3-day weighing (42); DFS: 7.4 [4.8], iodized salt: 7.9 [4.9] via 24-hour dietary recall (42)). However, the current acceptability study was not designed to quantify individual- or household-level salt use, and its 3-week duration and pilot sample constrain direct comparability to population-based estimates (68) or longer term randomized trials (42, 45).

In this study, 75% of households used ≥1 g of salt per person per day and weighed household salt use and hedonic ratings increased over time, indicating a period effect. Perceptions and use of the salt intervention may be impacted by several factors, such as salt use practices, household decision-making, socioeconomic factors, and awareness of the benefits of the intervention salt.

Most households in the study population reported using rock salt for cooking (vs. table salt which has a comparable texture to the intervention salt), suggesting that households may require an acclimation period for intervention adoption. In the current study, baseline household salt practices (type of salt purchased, use of rock and table salt) were similar between households that used the salt (i.e., ≥1 g per person per day for at least 2 weeks) vs. non-users (i.e., <1 g per person per day for at least 2 weeks), and between households that used a higher amount of salt (i.e., ≥5 g per person per day for at least 2 weeks) vs. low-users (i.e., <5 g per person per day for at least 2 weeks). In the aforementioned randomized trial of DFS in Bangalore, India (n=80; 4-day weighing) (45), salt use at the end of the 10-month study did not differ from baseline; however, usual household salt practices prior to the start of the trial were not reported.

Although there is limited research on correlates of salt intervention use, an effectiveness study of DFS in Uttar Pradesh, India, reported that lower wealth quintile was associated with higher DFS adherence (73). In the current study, in secondary analyses households with higher salt use (i.e., ≥5 g per person per day for at least 2 weeks) were more likely to have a lower household income (≤10,000 INR), compared to households with lower salt use (i.e., <5 g per person per day for at least 2 weeks; 59% vs. 31%; data not shown), suggesting households with lower income may use more salt per week compared to households with income >10,000 INR.

In the current study, although most women were the primary food preparers in the household, only 13% self-reported as the head of the household. In secondary analyses, women identified barriers to salt use (e.g., such as another household member did not permit the salt to be used in the home, other primary food preparers did not know to use the salt), which may influence decisions regarding salt use and food preparation in the household and are important areas of future research.

In analyses of reasons for not using salt, barriers such as concerns about using the salt and disliking the salt were identified; this is consistent with previous research on barriers for DFS in India’s safety net programs, which include low consumer awareness of the benefits, quality, and safety of DFS (52, 73). In the current study, hedonic ratings for overall acceptability were higher among households that used the salt (i.e., ≥1 g per person per day for at least 2 weeks) vs. non-users (i.e., <1 g per person per day for at least 2 weeks), and in households that used a higher amount of salt (i.e., ≥5 g per person per day for at least 2 weeks) vs. low-users (i.e., <5 g per person per day for at least 2 weeks; data not shown).

Findings from this acceptability study informed the parent randomized trial, including counseling households on potential barriers identified, potential benefits of the intervention, how to acclimate from rock salt to table salt for food preparation, and reminders for using the salt to support retention and adherence.

This study contributed additional information on the acceptability of quadruple-fortified salt, using different methods from previous studies (i.e., a randomized crossover trial design and triangle tests), and included participants (women 18–49 y and their households) that are similar to the target population for the parent randomized trial. This study has several limitations. This study was conducted in two villages outside of the trial catchment area and was not a random sample; this may limit generalizability to the catchment area for the randomized trial. The number of triangle tests was limited to prevent sensory fatigue; therefore, we were unable to assess differences in raw salt and other common foods, such as sambar and egg or milk-based dishes in which color changes have been noted in other studies (47, 72). Findings may not reflect sensory discrimination between salt types prepared in other foods; however, hedonic ratings were similar across salt types when prepared in dishes at the household as part of the randomized crossover trial. Although we invited other household members to participate in data collection, we were not able to recruit enough additional household members to conduct sub-analyses within this group. The focus of the randomized crossover trial was to compare intervention arms in the parent randomized trial (QFS vs. DFS); however, including iodized salt in the random allocation would have allowed additional comparisons to evaluate acceptability. Lastly, we assessed weighed salt use and were not able to distinguish how the intervention salt was used (e.g., purposes besides food preparation and at the table) or changes in baseline salt use practices. Weighed salt use increased over a 3-week period; understanding the reasons for the increase and changes in salt use and acceptance beyond three weeks are critical for determining the efficacy of QFS. Although reasons reported for not using the salt (predominantly traveling or forgetting during the first week) may have contributed to the observed increase, other potential factors, such as using the salt for other purposes or increasing salt consumption above baseline salt intake, were not specifically assessed and could impact the efficacy of the intervention and health in the population. In addition to counseling on salt use for food preparation, participants in the parent trial will receive information on salt reduction per WHO recommendations for sodium intake (60, 74). Adjusting from rock salt to table salt may have had a strong influence on salt use and hedonic ratings; detectable differences between salt interventions (QFS vs. DFS) may require a longer study period beyond three weeks. As part of the parent trial, long-term acceptance and perceptions of using multiple-micronutrient fortified salt within this population will be assessed to further inform public health programs, particularly given the large number of households that reported using rock salt for food preparation.

In summary, in this study, acceptability of quadruple-fortified salt was high, based on individual hedonic ratings and weighed household salt use. Foods prepared with DFS, QFS, and iodized salt were not distinguishable, and intervention salts were highly rated across sensory domains and for overall acceptability. Findings informed the design of a randomized efficacy trial of QFS in this population. The parent randomized trial of QFS (i.e., iron, iodine + folic acid and/or vitamin B₁₂), in 1,000 women and their households, will evaluate the independent addition of folic acid or vitamin B₁₂ compared to DFS on micronutrient status prospectively over 12 months (60). Findings from this acceptability study and the parent randomized trial will inform interventions and public health programs for anemia and birth defects prevention.