Prevalence of morbidity symptoms among pregnant and postpartum women receiving different nutrient supplements in Ghana and Malawi: A secondary outcome analysis of two randomised controlled trials

Abstract Little is known about the impact of small‐quantity lipid‐based nutrient supplements (SQ‐LNSs) on maternal morbidity. This secondary outcome analysis aimed to compare morbidity symptoms among women in two trials evaluating the efficacy of SQ‐LNSs. From enrolment (≤20‐week gestation) to 6 months postpartum, Ghanaian (n = 1320) and Malawian (n = 1391) women were assigned to consume daily: 60 mg iron and 400 µg folic acid until childbirth and placebo thereafter (iron and folic acid [IFA] group); or multiple micronutrients (MMN); or 20 g/day SQ‐LNSs. Within country, we used repeated measures logistic regression and analysis of variance models to compare group differences in the period prevalence and percentage of days of monitoring when women had fever, gastrointestinal, reproductive, and respiratory symptoms during the second and third trimesters of pregnancy (n ~ 1243 in Ghana, 1200 in Malawi) and 0–3 and 3–6 months postpartum (n ~ 1212 in Ghana, 730 in Malawi). Most outcomes did not differ significantly among groups, with the following exceptions: in Ghana, overall, the prevalence of vomiting was lower in the LNS (21.5%) than MMN (25.6%) group, with the IFA group (23.2%) in‐between (p = 0.046); mean ± SD percentage of days with nausea was greater in the LNS (3.5 ± 10.3) and MMN (3.3 ± 10.4) groups than the IFA (2.7 ± 8.3) group (p = 0.002). In Malawi, during 3–6 month postpartum, the prevalence of severe diarrhoea was greater in the LNS (8.1%) than the MMN (2.9%) group, with IFA (4.6%) in‐between, p = 0.041). We conclude that the type of nutrient supplement received during pregnancy and lactation generally does not influence morbidity symptoms in these settings. Clinicaltrials.gov identifiers: NCT00970866; NCT01239693.


| INTRODUCTION
Common morbidity symptoms such as fever, poor appetite, nausea and vomiting are experienced by many pregnant and lactating women (Regodón Wallin et al., 2020) and can have debilitating consequences such as reduced quality of life, depressive symptoms and malnutrition (Regodón Wallin et al., 2020;Temming et al., 2014). However, these symptoms have generally received little research attention (Parikh et al., 2021;Roos-Hesselink et al., 2019;Vogel et al., 2021).
Low nutrient status during pregnancy and lactation could contribute to the occurrence and severity of morbidity symptoms (Papathakis et al., 2016;Wu, Imhoff-Kunsch, & Girard, 2012).
Current estimates suggest that 80% of nonpregnant women 15-49 years of age in Sub-Saharan Africa (161 million) have deficiencies of at least one of three core micronutrients, including iron, zinc and folate (Stevens et al., 2022) attributed primarily to poor access to nutritious foods (Darnton-Hill & Mkparu, 2015;Lee et al., 2013). Similar estimates are not available for pregnant and lactating women (Stevens et al., 2022), but the prevalence of deficiency could be even higher due to increased physiological requirements during those periods (Adu-Afarwuah, Lartey, & . Small-quantity (20 g/day) lipid-based nutrient supplements (SQ-LNSs) were developed for pregnant and lactating women by collaborators of the International Lipid-based Nutrient Supplements (iLiNS) Project  to add to the menu of interventions for addressing poor access to nutritious foods in lowincome settings (Das et al., 2018). Besides micronutrients, SQ-LNSs provide high-quality protein and essential fatty acids (EFAs) linoleic acid and α-linolenic acid, which are not usually included in multiple micronutrient supplements.
We previously reported pregnancy outcomes from the iLiNS-DYAD trials in Ghana (Adu-Afarwuah et al., 2015) and Malawi  that evaluated the efficacy of maternal SQ-LNSs. In Ghana, maternal SQ-LNS had a positive impact on several outcomes including greater foetal growth (especially among primiparous women) (Adu-Afarwuah et al., 2015), decreased prevalence of low gestational weight gain (Adu-Afarwuah, Lartey & Okronipa, , and increased likelihood of adequate iodine status (Adu-Afarwuah et al., 2018). In Malawi , results for foetal growth were less conclusive but generally consistent with those from Ghana (Adu-Afarwuah et al., 2015). Evaluating the impact of SQ-LNS consumption on women's morbidity is important for identifying additional potential benefits of this intervention, as well as potential risks.
The aim of this secondary outcome analysis was to compare morbidity symptoms among women enrolled in the iLiNS-DYAD trials in Ghana and Malawi. We hypothesised that within country, the three treatment groups would differ in the prevalence and mean percentage of days of monitoring when women had morbidity symptoms during pregnancy and lactation.
2 | METHODS 2.1 | Study design, setting and participants As described previously, the trials in Ghana (Adu-Afarwuah et al., 2015 and Malawi (Ashorn, Alho, Ashorn, Cheung, Dewey, Gondwe, et al., 2015; were partially double-blind, parallel, individually randomised, controlled trials with three equal-size groups. In brief, we conducted the Ghana trial (ClinicalTrials.org Identifier: NCT00970866) in the Somanya-Odumase-Kpong area, a semi-urban setting about 70 km north of Accra. At the time of the study, the inhabitants were mainly subsistence farmers and petty traders; by Ghanaian standards, most had low socioeconomic status.
Adult literacy was~44%. The staple diets consisted of maize, cassava, plantain, rice, fish and leafy vegetables; mango and watermelon consumption was seasonal. Health services were provided mainly by two hospitals, one polyclinic (providing more services than a clinic but not to a hospital status) and one clinic run by the government.
Between December 2009 and December 2011, women on routine attendance to antenatal clinics in the four health facilities were eligible if they were ≥18 years old, ≤20 weeks pregnant, and their antenatal cards contained the information necessary for determining eligibility. We included women ≥18 years of age because the legal age of adulthood was 18 years, and the prevalence of pregnancy among girls <18 years at the study site was relatively low. Exclusion criteria were: intention to move out of the area during the period of the intervention; milk or peanut allergy; unwillingness to receive field workers or take study supplement; gestational age (GA) > 20 weeks before completion of enrolment; antenatal card indicated HIV

Key messages
• Ghanaian and Malawian women assigned to receive iron and folic acid (pregnancy only) or multiple micronutrients (pregnancy and lactation) or small-quantity lipid-based nutrient supplements (SQ-LNSs) (pregnancy and lactation) generally did not differ in morbidity outcomes during the second and third trimesters of pregnancy and 0-3 and 3-6 months lactation.
• The lack of group differences in most of the morbidity outcomes corroborates previous results of no significant group differences in the percentage of women who experienced severe adverse events or mean concentrations of two biomarkers of inflammation (C-reactive protein and alpha-1 glycoprotein) or percentage of women with elevated concentrations of the biomarkers.
• Overall, there was no consistent evidence that SQ-LNS or multiple micronutrients increased or decreased maternal morbidity during pregnancy or lactation, compared to iron and folic acid. In Malawi, we conducted the trial (ClinicalTrials.org Identifier NCT01239693) in the Mangochi District, a predominantly rural area south of the country. The main sources of livelihood in the area were subsistence farming and fishing, and the diets consisted predominantly of cereals, mainly corn, with relatively small amounts of roots, tubers, fish, fruit and legumes (Ndekha et al., 2000). In both countries, haemoglobin concentration between 50 and 109 g/L (WHO, 2021) at baseline was not an exclusion criterion because the intervention provided at least daily iron and folic acid to all participants, and in addition, we monitored supplement intakes and morbidity bi-weekly.
After baseline assessments, pregnant women were randomised to one of three supplementation groups. In Ghana, women were assigned to receive daily: iron and folic acid capsule until childbirth and thereafter placebo (200 mg/day Ca) until the end of follow-up at 6 months postpartum (IFA supplement or group), or multiple micronutrient capsule throughout follow-up (MMN supplement or group); or 20 g/day SQ-LNS throughout follow-up (SQ-LNS supplement or LNS group). In Malawi, women were assigned to receive one of the same three supplements as in Ghana until childbirth. After childbirth, only those women enrolled on the 'complete follow-up' scheme (n = 864) received the assigned placebo or supplement and underwent a full follow-up morbidity monitoring until 6 months postpartum (lactation) as in Ghana; those in the 'simplified follow-up' scheme (n = 527) received no placebo or supplement postpartum and did not undergo any morbidity monitoring after childbirth.
The randomisation techniques in Ghana (Adu-Afarwuah et al., 2015) and Malawi  were similar: the Study Statistician at UC Davis used a computer-generated (SAS version 9.4) scheme to develop the group assignments in blocks of 9, and each supplement or group was coded with three different colours (Adu-Afarwuah et al., 2015) or alphabetical letters . At each site, group assignments were placed in opaque envelopes and labelled in a manner inconspicuous to the women. Subsequently, the envelopes were stacked in increasing order of block numbers. At each enrolment, the individual performing the randomisation shuffled the nine (Adu-Afarwuah et al., 2015) or six  topmost envelopes in the stack, asked the potential participant to choose one (to determine the group assignment), and then returned the unused envelopes to the top of the stack. This process was repeated until all the envelopes for the enrolment site were used. When there were less than nine (Adu-Afarwuah et al., 2015) or six  envelopes left for participants to pick from, the individual performing the randomisation presented whatever number of envelopes that remained. It was not possible for anyone to guess the group assignments since none had knowledge of the randomisation scheme. Group allocation information was kept only by the field supervisor at the project site, and the Study Statistician at UC Davis.

| Nutrient supplements
The nutrient contents of the supplements women consumed are shown in Table 1. The IFA capsule contained 60 mg/day of iron and 400 µg/day of folic acid reflecting the standard micronutrient supplementation for pregnant women in Ghana (WHO, 2012) and Malawi  at the time of the trials. The MMN supplement contained 18 vitamins and minerals providing either 1x or 2x the recommended dietary allowance (RDA) of nutrients for pregnancy (Adu-Afarwuah et al., 2015) except for iron, which we kept at 20 mg/ day assuming that together with the amount from the usual diet, this dose would give a total daily intake close to the UNICEF/WHO/UNU International Multiple Micronutrient Preparation UNIMMAP iron content for pregnancy (UNICEF/WHO/UNU, 1999) without greatly exceeding the RDA (9 mg/day) for lactation (Adu-Afarwuah et al., 2015;IOM, 2001). The SQ-LNS contained the same micronutrients as the MMN group, and in addition, four minerals (calcium, phosphorus, potassium and magnesium), protein (2.6 g/day), energy (118 kcal/day), and EFAs (4.59 g/day linoleic acid and 0.59 g/ day α-linolenic acid). The oil content of the SQ-LNS consisted of soybean oil and the oil supplied by the peanut paste ingredient.
The IFA and MMN (each in 10-capsule blister packs) were produced by DSM South Africa, and the SQ-LNSs (individual 20-g sachets for women and 10-g sachets for infants) by Nutriset S.A.S.

| Intervention
Intervention procedures in Ghana (Adu-Afarwuah et al., 2015) and Malawi  were similar. At enrolment, women received a 2-week supply of the assigned supplement with the advice to those in the IFA and MMN groups to consume one capsule each day with water after a meal, and those in the LNS group to consume one sachet each day after mixing the entire content of the sachet with a small amount of food. In addition, women received the standard message: 'Do not forget to eat meat, fish, eggs, fruits and vegetables whenever you can; you still need these foods even as you take the supplement we have given you'. We advised women not to consume more than one capsule (IFA and MMN groups) or sachet (LNS group) per day, even if they forgot to take the supplement the previous day or days and to take their assigned supplement along if they had to travel out of the study area. Women who travelled and did not return before the next biweekly visit were given an extra supply for the period they intended to be away.
From enrolment to childbirth (hereafter, pregnancy), field workers visited women in the homes biweekly during which they delivered fresh supplies of supplements or placebo and collected selfreports of supplement intakes and morbidity. From childbirth till exit at 6 months postpartum (hereafter, lactation), field workers visited the women weekly (partly to collect infant morbidity data) but delivered the women's supplement or placebo and collected the women's morbidity data biweekly as before. To corroborate selfreports of supplement intake, field workers recovered and counted any unused supplements.

| Morbidity symptoms data collection
At each biweekly visit in Ghana (Adu-Afarwuah et al., 2015) and Malawi , field workers asked and probed women about the number of days preceding the visit when women experienced T A B L E 1 Composition of supplements used in the study.  (Kaestel et al., 2005), except iron. c Same micronutrient content as MMN.
d Nutrient concentrations include contributions from the ingredients as well as from the multiple micronutrient premix. Oil content consisted of soybean oil and oil from the peanut paste ingredient, which were combined to meet the target amounts of α-linolenic acid and linoleic acid in the final product. morbidity symptoms. If a woman had a symptom during the days in question, she was asked whether the symptom was present 'yesterday' (yes/no). We considered 13 symptoms categorised into generalised (fever); gastrointestinal (poor appetite, nausea, vomiting, abdominal pain, diarrhoea and severe diarrhoea); reproductive (pregnancy-related bleeding and mastitis); and respiratory (cough, sore throat, any nasal discharge and severe nasal discharge) morbidity. These symptoms were selected based on a similar study in Bangladesh (Kim et al., 2012) suggesting that they may be common in our study settings. Women were asked about abdominal pain and pregnancy-related bleeding during pregnancy only, and mastitis during lactation only. We defined diarrhoea as the passage of ≥3 watery stools (Ghosh et al., 2010) or by women's report of increased frequency and reduced consistency of stools (Bhutta et al., 1991) in a 24h period, severe diarrhoea as having either bloody stools or mucus in stools or both and severe nasal discharge as having a thick or yellowish nasal discharge.
In Ghana, field workers unable to complete the morbidity data collection on the usual 14-day schedule (typically because of a participant's travel) had up to 7 days to complete the visit or else consider it missed. At the first home visit 14-21 days after enrolment, field workers collected data 'since enrolment'; in subsequent visits, data were collected 'since last visit' if the current visit occurred within 7-21 days after the preceding visit, or 'during the past 2 weeks' if the current visit occurred >21 days after the preceding visit. With this data collection scheme, it was possible to obtain self-reports of women's morbidity over all or nearly all of the entire follow-up period.
In Malawi, the first home visit was completed within 7-13 days after enrolment, the second occurred 7 or 14 days after the first visit, and subsequent visits were completed every 14 days thereafter. At Weeks 32 and 36 of pregnancy, supplement delivery and data collection were accomplished during the women's visits to the laboratory. Although women were visited biweekly, the recall of selfreported morbidity was conducted over the 'past 7 days' because of fears that recall over a longer period might be problematic due to a low average level of education.
It was not possible to blind field workers or study participants to those who received the SQ-LNSs, because the capsules (IFA and MMN) and SQ-LNS sachets looked different.

| Measurement time intervals and outcome measures
Because the appearance of certain morbidity symptoms is related to the duration of the pregnancy (e.g., nausea and vomiting often appear and peak during the first trimester, and disappear by the end of the second trimester (Lacasse & Bérard, 2008)

| Sample size and statistical analysis
The sample sizes in the Ghana and Malawi trials were based on detecting a small-to-moderate Cohen's effect size d of 0.3 between any two groups for either of the two primary continuous variables (i.e., length-for-age z-score at birth and at 18 months of age), with a two-sided 5% test and 80% power. As described previously, the number of pregnant women enrolled in Ghana was 1320 Our statistical analysis plan was posted online (https://ilins. ucdavis.edu/) before analysis. We used SAS version 9.4 (SAS Institute) to perform the analyses according to the complete-case intention-to-treat principle (Groenwold et al., 2014) and the group to which women were assigned, regardless of any protocol violations.
Data on women who were lost to follow-up for any reason were included in the analysis if available. Background and household variables were summarised using percentages for categorical variables and mean ± SD for continuous variables. We examined morbidity variables by time interval, that is, the second and third trimesters during pregnancy and 0-3 and 3-6 months during lactation, after excluding the first trimester due to a small number of women who attended antenatal clinics and were enrolled during the first trimester. We evaluated the effects of time interval and intervention group (IFA, MMN or LNS) on the occurrence of morbidity symptoms (yes/no) by using repeated measures logistic regression, and on the percentage of days with morbidity symptoms by using repeated measures analysis of variance (ANOVA). The fixed effects in these models were one within-subjects factor (i.e., time intervals during pregnancy and lactation) and one between-subjects factor (intervention group), while treating participants as a random effect. We used a two-tailed test for the analysis of all outcome variables because we did not predict the direction of any group differences. We set the level of significance (α) at 0.05 and considered 0.05 < p ≤ 0.08 to indicate a tendency toward significance In adjusted analysis, we controlled for prespecified baseline variables, including age, height, BMI, GA, years of education, marital status, household assets index (proxy indicator for socioeconomic status), HFIAS score, the season in which woman was enrolled, primiparity, blood haemoglobin concentration, and positive malaria RDT if they were significantly associated with the outcome at 10% level of significance. Finally, we performed two additional exploratory analyses: first, we used a 'per protocol' sensitivity analysis to assess the treatment effect among women considered to have high adherence to supplementation (i.e., added supplements to their foods >70% of days during follow-up); second, we evaluated potential effect modification (p-for-interaction) by seven baseline characteristics, including age, GA, years of education, household assets index, household toilet facility category, HFIAS score, and primiparity. We did not perform stratified analyses because the effect modification analyses were exploratory.

| RESULTS
The background characteristics of the women enrolled in the iLiNS DYAD trials are shown in  third trimester, 0-3 mo lactation and 3-6 months lactation were 56 ± 19, 89 ± 18, 81 ± 13 and 80 ± 12, respectively for Ghana, and 29 ± 9, 40 ± 10, 36 ± 9 and 38 ± 9, respectively for Malawi. In both countries, the mean number of days of morbidity monitoring was smallest for the second trimester compared with the other periods, since many of the women were enrolled during the second trimester. In Malawi, cough, nausea and fever were the most common morbidity symptoms during the second trimester of pregnancy (35.3%-49.1% period prevalence; 6.3-12.3 mean percentage of days), while cough, fever and sore throat were the most common symptoms during the third trimester (35.1%-55.7% period prevalence; 5.3-11.2 mean percentage of days). During lactation, cough, nasal discharge and fever were the most common symptoms during the 0-3 months period (17.5%-31.4% period prevalence; 2.0-5.5 mean percentage of days), while cough, nasal discharge and sore throat were the most common during the 3-6 months period (17.8%-29.0% period prevalence; 2.5-4.8 mean percentage of days).

| Summary of morbidity results by time interval
In both countries, the main effect of time interval was statistically significant for all morbidity outcomes, except for the prevalence of abdominal pain, which was assessed during pregnancy (only) in Ghana but not during any time interval in Malawi. For most gastrointestinal symptoms (notably poor appetite, nausea and vomiting), the prevalence or mean percentage of days declined significantly from the second trimester of pregnancy through 0-3 months or 3-6 months of lactation. A similar pattern was shown for the mean percentage of days of monitoring when women had fever and respiratory symptoms (cough, sore throat, any nasal discharge and severe nasal discharge), but the prevalence of fever and respiratory symptoms increased or appeared to increase from the second to the third trimesters of pregnancy, before declining through 0-3 months or 3-6 months of lactation. The mean ± SD percentage of days of monitoring when women had morbidity symptoms during pregnancy and lactation, by intervention group, in Ghana and Malawi are shown in Table 5. In Ghana, we found significant group differences (p = 0.002) and

| Main effect of the intervention group
intervention × time interval interaction (p = 0.03) for mean ± SD percentage of days women had nausea during the follow-up period.
In the analysis based on the separate time intervals, the differences in the mean percentage of days women had nausea occurred during the third trimester of pregnancy (p = 0.040) and 0-3 months of lactation (p = 0.030). In the third trimester of pregnancy, the mean percentage of days women had nausea was greater in the LNS (4.6 ± 12.1) group than in the IFA group (2.8 ± 6.8) and the difference in mean (95% CI) was 1.8 (0.4, 3.2); the MMN (4.8 ± 11.3) did not differ significantly from the other groups. At 0-3 months of lactation, the mean percentage of days women had nausea was greater in the MMN (0.8 ± 4.0) and LNS (0.7 ± 4.2) groups than in the IFA group (0.3 ± 1.7); the difference in mean (95% CI) when comparing the MMN and LNS groups with the IFA group was 0.5 (0.0, 1.0) for the MMN group and 0.4 (0.0, 0.9) for the LNS group. There were significant intervention × time interval interactions for the percentage of days of monitoring when women had mastitis (p = 0.04) and severe nasal discharge (p = 0.05), but in none of the separate time intervals did the groups differ in either of these outcomes.
In Malawi, the groups did not differ significantly in the mean percentage of days with morbidity symptoms during the follow-up period. We found a significant intervention group × time interval interaction (p = 0.06) for the percentage of days women had severe nasal discharge, but in none of the separate time intervals did the groups differ in this outcome. These findings did not change in either country after adjusting for baseline characteristics (results not shown). 1.0 ± 5.6 (720) a Ghana: All enrolled women received intervention or placebo and underwent a full follow-up morbidity monitoring until 6 months postpartum. Malawi: All enrolled women received intervention and underwent a full follow-up morbidity monitoring until childbirth; thereafter, only women (n = 864) in the 'complete follow-up' scheme received the assigned placebo or intervention and underwent a full follow-up morbidity monitoring until 6 months postpartum. The mean ± SD number of days of morbidity monitoring in the second trimester, third trimester, 0-3 months lactation and 3-6 months lactation were 56 ± 19, 89 ± 18, 81 ± 13 and 80 ± 12, respectively for Ghana and 29 ± 9, 40 ± 10, 36 ± 9 and 38 ± 9, respectively for Malawi. b p-values for these analyses were not reported because the study did not have prespecified hypotheses for the effect of time interval on the prevalence or percentage of days of monitoring when women had morbidity symptoms.
T A B L E 4 Prevalence of morbidity symptoms during pregnancy and lactation, by intervention group, in Ghana and Malawi a .

| Exploratory analyses
Results of the per-protocol analysis limiting the analysis to women with high adherence to supplementation (i.e., added supplements to their foods >70% of days during follow-up) did not differ from those of the main analysis (results not shown). Regarding effect modification, there were some significant interactions between intervention group and certain baseline characteristics, such as maternal age in Ghana and household toilet facility in Malawi, but there were no clear or consistent patterns across time intervals. Given the number of interactions examined, these few significant P-for-interaction values could have been due to chance.

| DISCUSSION
In Ghana and Malawi, we found that after excluding the first trimester due to the small number of women enrolled in that period, maternal morbidity symptoms were most common in the second trimester of pregnancy and prevalence declined significantly through 3-6 months of lactation. In Ghana, the three intervention groups did not differ significantly in the morbidity outcomes measured, except for the period prevalence of vomiting in the third trimester of pregnancy (greater in the MMN group than the LNS group) and 0-3 months lactation (greater in the MMN group than the IFA group) and mean percentage of days with nausea in the third trimester of pregnancy and during 0-3 months of lactation (greater in the MMN and LNS groups than in the IFA group). Likewise, in Malawi, the three intervention groups differed significantly only in the period prevalence of severe diarrhoea during 3-6 months of lactation (greater in the LNS group than the MMN group). In both countries, the differences between groups in these three outcomes were small and possibly due to chance. Overall, we did not find consistent evidence that SQ-LNS or MMN increased or decreased maternal morbidity during pregnancy or lactation, compared to IFA.
The strengths of our study included the random allocation of participants to intervention groups and the frequent home visits by study staff to collect the morbidity data. There are potential limitations. First, the self-reported morbidity data could not be clinically verified and may also be compromised by recall bias.
Second, the period prevalence of morbidity outcomes during the second trimester of pregnancy may be underestimates because many women in both countries were enrolled later in that trimester rather than earlier, and thus some morbidity symptoms during that period may have been missed. In Malawi, the biweekly data collection focused on the 'past 7 days', which reduced the risk of recall bias (relative to a 14-day recall period) but meant that we missed collecting morbidity data for approximately one-half of the days during the follow-up period. Third, we tested multiple hypotheses involving 11 or 10 morbidity symptoms in either country and therefore, at least some of the observed differences may be due to chance (Li et al., 2017). We did not correct for multiplicity in our hypotheses testing because we considered any such correction as T A B L E 4 (Continued) The mean (SD) number of days of morbidity monitoring in the second trimester, third trimester, 0-3 months lactation and 3-6 months lactation were 56 ± 19, 89 ± 18, 81 ± 13 and 80 ± 12, respectively for Ghana and 29 ± 9, 40 ± 10, 36 ± 9 and 38 ± 9, respectively for Malawi.
Severe nasal discharge Overall The mean ± SD number of days of morbidity monitoring in the second trimester, third trimester, 0-3 months lactation and 3-6 months lactation were 56 ± 19, 89 ± 18, 81 ± 13 and 80 ± 12, respectively for Ghana and 29 ± 9, 40 ± 10, 36 ± 9 and 38 ± 9, respectively for Malawi. For nausea, mastitis and severe nasal discharge, the impact of intervention group for each time interval was determined separately because the p-value for the time interval × intervention group interaction was significant at α = 0.1.
Despite these limitations, self-reported data may reflect community experiences not often captured in clinic or hospital settings (Kim et al., 2012). These trials included approximately 12 months of maternal follow-up, which allowed the women to become accustomed to monitoring their own morbidity symptoms, thereby possibly minimising the risk of recall bias. It is also unlikely that any recall bias or possible underestimates of morbidity prevalence in the second trimester of pregnancy would be unique to a specific intervention group or groups.
We reported previously that the intervention groups in Ghana in which pregnant and lactating women received 40 g/day LNS but the comparison groups received no supplements. In Nepal (Christian et al., 2009), women receiving folic acid, iron-folic acid, iron-folic acid-zinc, or multiple micronutrients from early pregnancy through 3 months postpartum did not differ in several morbidity outcomes measured during the first 9 days after giving birth, including poor appetite for >2 days, ≥4 watery stools and/or mucus/blood in stools, or persistent cough for >2 days.
In Ghana, the observed group differences in the mean percentage of days of monitoring when women had nausea during the third trimester of pregnancy and 0-3 months postpartum may potentially be related to the perceived metallic 'taste' or 'aftertaste' of LNS when consumed without being mixed with food (Klevor et al., 2016).
However, the finding that the MMN group also reported a greater percentage of days with nausea than the IFA group argues against that potential explanation, given that women swallowed the MMN with water. In Malawi, it is difficult to explain why the period prevalence of severe diarrhoea during 3-6 months postpartum was greater in the LNS group than the MMN group, as there were no differences during any other time interval nor for the mean percentage of days of monitoring when women had severe diarrhoea.
These observed differences were small and perhaps not clinically meaningful.
We conclude that Ghanaian and Malawian women receiving IFA, MMN or SQ-LNS supplements during pregnancy and lactation generally do not appear to differ in morbidity symptoms. Additional studies to examine the impact of SQ-LNS consumption on maternal morbidity would be useful.