Effects of nutrition education and home gardening interventions on feto-maternal outcomes among pregnant women in Jimma Zone, Southwest Ethiopia: A cluster randomized controlled trial

**Intervention:** A combined package of (1) nutrition education sessions and (2) home gardening support, delivered to pregnant women and their households.

**Nutrition education:** Six structured group sessions (one per month) covering: importance of diverse diets during pregnancy, food sources of key nutrients (iron, zinc, vitamin A, protein), food preparation and preservation, hygiene practices, and how to use garden produce for family meals. Sessions were delivered by trained community health workers (Health Extension Workers) at local health posts.

**Home gardening support:** Provision of vegetable seeds (kale, Swiss chard, carrots, beetroot, tomatoes, onions, peppers), gardening tools (hoe, watering can), and technical training on raised-bed construction, composting, pest management, and year-round planting. Households were encouraged to maintain a garden of at least 10 m².

**Comparator:** Standard antenatal care (ANC) only, which in this setting includes routine check-ups, iron-folate supplementation, tetanus vaccination, and general health advice — but no structured nutrition education or gardening support.

**Outcome measures (primary and secondary):**

**Primary:** Low birth weight (LBW, defined as <2500 g at delivery)

**Secondary:** Maternal dietary diversity score (MDDS, 0–10 scale, higher = more food groups consumed in past 24 hours), maternal anemia (hemoglobin <11 g/dL in third trimester), gestational weight gain, preterm birth (<37 weeks), stillbirth, neonatal mortality, and maternal mid-upper arm circumference (MUAC)

**Sample size:** 468 pregnant women (234 in intervention group, 234 in control group) recruited from 12 rural kebeles (the smallest administrative unit in Ethiopia) in Jimma Zone, Southwest Ethiopia.

**Population:** Pregnant women aged 18–49 years, at 12–20 weeks gestation at enrollment, who had lived in the study area for at least 6 months and planned to deliver there. Exclusion criteria: multiple pregnancy (twins/triplets), known chronic illness (diabetes, hypertension, HIV), or severe anemia at baseline (Hb <7 g/dL).

**Setting:** Rural agrarian communities in Jimma Zone, Oromia Region, Ethiopia. Most households were subsistence farmers with limited market access. Baseline dietary diversity was low (mean MDDS ~3.5 out of 10). About 40% of women had no formal education. The region has a high prevalence of undernutrition and food insecurity.

**Duration:** Women were enrolled at 12–20 weeks gestation and followed through delivery (approximately 20–28 weeks of intervention exposure). The intervention began at enrollment and continued until delivery.

**Birth weight:** Measured within 24 hours of delivery using a calibrated digital infant scale (Seca 354, accuracy ±10 g). Measured by trained data collectors at health facilities or at home births.

**Maternal dietary diversity:** Assessed using the Women's Dietary Diversity Score (WDDS), a 10-food-group recall tool (grains, white roots/tubers, legumes, nuts/seeds, dairy, meat/fish/eggs, dark leafy greens, vitamin A-rich fruits/veg, other vegetables, other fruits). Women reported all foods consumed in the previous 24 hours. Score = number of groups consumed (0–10).

**Maternal anemia:** Venous blood sample drawn at 34–36 weeks gestation, analyzed using a hematology analyzer (Mindray BC-3000) for hemoglobin concentration. Anemia defined as Hb <11 g/dL.

**Gestational weight gain:** Weight measured at enrollment and at 34–36 weeks using a calibrated digital scale (Seca 874). Total gain calculated as difference.

**MUAC:** Measured at enrollment and at 34–36 weeks using a non-stretchable tape, midpoint between acromion and olecranon.

**Obstetric outcomes:** Preterm birth (<37 weeks), stillbirth, neonatal death — recorded via maternal interview and health facility records.

**Process measures:** Attendance at education sessions, garden establishment and maintenance (observed by field workers), seed use, and vegetable consumption frequency.

**Study design:** Cluster randomized controlled trial (cRCT). Twelve kebeles were randomly assigned to either intervention (6 kebeles) or control (6 kebeles). All eligible pregnant women within each kebele were enrolled. This is a parallel-group design with the cluster (kebele) as the unit of randomization, not the individual woman.

**Why cluster randomization?** The intervention included community-level components (gardening training, group education) that could "spill over" between women in the same village. Randomizing by kebele prevents contamination — if women in the same village were randomized individually, those in the control group might learn gardening techniques from intervention neighbors. Clustering also reflects how the intervention would be delivered in real-world programs (by community health workers serving entire villages).

**Randomization:** Twelve kebeles were randomly allocated 1:1 to intervention or control using a computer-generated random sequence. The randomization was done by a statistician not involved in data collection. Allocation was concealed until after baseline data collection.

**Blinding:** This was an open-label trial — participants, health workers, and data collectors knew which kebeles received the intervention. Blinding was impossible because the intervention involved visible activities (building gardens, attending group sessions). However, outcome assessors (those measuring birth weight and analyzing blood samples) were blinded to group assignment. The statistician analyzing data was also blinded.

**Duration:** The intervention lasted from enrollment (12–20 weeks gestation) until delivery — approximately 20–28 weeks. This is a moderate duration for a nutrition intervention during pregnancy. The critical window for affecting birth weight is the third trimester (rapid fetal weight gain), so the intervention covered this period for most women.

**Statistical approach:** Intention-to-treat analysis (all women analyzed in their assigned group regardless of compliance). For binary outcomes (LBW, anemia, preterm birth), they used generalized estimating equations (GEE) with a log-binomial model to account for clustering. For continuous outcomes (dietary diversity, weight gain), they used linear mixed models with random effects for kebele. Results are reported as risk ratios (RR) or mean differences with 95% confidence intervals. They adjusted for baseline maternal age, education, parity, and household food security.

**What this design can prove:** A well-conducted cRCT can establish causality — if the intervention group has better outcomes, we can be confident the intervention caused it (assuming no major confounds). The cluster design strengthens external validity because it mimics real-world program delivery.

**What this design cannot prove:** (1) Cannot separate the effects of nutrition education from gardening — the combined package was tested, not individual components. (2) Cannot determine which specific nutrients or foods drove the effect. (3) Open-label design means placebo effects or differential care (Hawthorne effect) could influence outcomes — women in intervention kebeles may have received more attention or changed behavior because they knew they were being studied. (4) Cluster randomization with only 12 clusters is relatively low — there is risk of baseline imbalance between groups despite randomization. (5) Cannot generalize to urban populations, women with higher education, or settings with different food environments.

**Major methodological weaknesses:**

Small number of clusters (12) increases risk of type I error and limits ability to adjust for cluster-level confounds

No blinding of participants or field workers

Self-reported dietary data (24-hour recall) is subject to recall bias and social desirability bias

Loss to follow-up: 42 women (9%) were lost (18 intervention, 24 control) — differential loss could bias results

No data on actual nutrient intake or biomarkers (e.g., serum ferritin, vitamin A) — only dietary diversity and anemia

Short follow-up: no data on long-term child growth, development, or maternal postpartum nutrition

**Primary outcome — Low birth weight:**

Intervention group: 7.8% (17/218) of infants had LBW

Control group: 24.3% (51/210) of infants had LBW

Risk ratio (RR): 0.32 (95% CI: 0.19 to 0.54, p < 0.001)

This means a 68% reduction in LBW risk

Number needed to treat (NNT): 6 women would need to receive the intervention to prevent one LBW case

**Secondary outcomes — Maternal nutrition:**

**Dietary diversity score (MDDS, 0–10):**

- Intervention: mean 5.2 (SD 1.4)

- Control: mean 3.7 (SD 1.1)

- Mean difference: +1.5 points (95% CI: 1.2 to 1.8, p < 0.001)

- The intervention group consumed about 1.5 more food groups per day

**Maternal anemia (Hb <11 g/dL at 34–36 weeks):**

- Intervention: 18.3% (40/218)

- Control: 32.4% (68/210)

- RR: 0.57 (95% CI: 0.40 to 0.81, p = 0.002)

- 43% reduction in anemia risk

**Gestational weight gain (kg):**

- Intervention: mean 8.2 kg (SD 2.8)

- Control: mean 6.9 kg (SD 2.5)

- Mean difference: +1.3 kg (95% CI: 0.7 to 1.9, p < 0.001)

**MUAC (cm, change from baseline to 34–36 weeks):**

- Intervention: +0.8 cm (SD 0.6)

- Control: +0.3 cm (SD 0.5)

- Mean difference: +0.5 cm (95% CI: 0.3 to 0.7, p < 0.001)

**Other secondary outcomes:**

**Preterm birth (<37 weeks):**

- Intervention: 6.4% (14/218)

- Control: 12.4% (26/210)

- RR: 0.52 (95% CI: 0.28 to 0.97, p = 0.04)

- 48% reduction in preterm birth

**Stillbirth:**

- Intervention: 1.4% (3/218)

- Control: 2.9% (6/210)

- RR: 0.48 (95% CI: 0.12 to 1.91, p = 0.30)

- Not statistically significant (wide confidence interval)

**Neonatal death (within 28 days):**

- Intervention: 0.9% (2/218)

- Control: 1.9% (4/210)

- RR: 0.48 (95% CI: 0.09 to 2.59, p = 0.39)

- Not statistically significant

**Compliance and process measures:**

89% of intervention women attended at least 4 of 6 education sessions

82% of intervention households had an active garden at delivery (observed by field workers)

74% reported consuming vegetables from their garden at least 3 times per week

**Low birth weight:** The risk dropped from about 1 in 4 babies (24.3%) to about 1 in 13 babies (7.8%). This is a large effect — comparable to the difference between a high-income country (where LBW rates are typically 5–10%) and a low-income setting with high food insecurity. For context, the global average LBW rate is about 15%. The intervention brought this rural Ethiopian population below the global average.

**Dietary diversity:** Women went from eating about 3.7 food groups per day to about 5.2 — roughly equivalent to adding one serving of vegetables and one serving of protein-rich food (e.g., eggs, legumes) to their daily diet. This is a meaningful shift: the minimum recommended dietary diversity for pregnant women is 5 food groups, so the intervention moved the average woman from below to above this threshold.

**Anemia:** The prevalence dropped from 32.4% to 18.3% — a 43% reduction. This is clinically significant because maternal anemia is associated with fatigue, poor pregnancy outcomes, and increased risk of postpartum hemorrhage. The effect is roughly equivalent to what you might see from daily iron supplementation alone in similar populations.

**Gestational weight gain:** The intervention group gained about 1.3 kg more total weight. For a 40-week pregnancy, this is about 0.05 kg per week extra — modest but meaningful given that inadequate weight gain is a risk factor for LBW.

**What the authors acknowledge:**

Open-label design (no blinding of participants or field workers) — possible Hawthorne effect

Self-reported dietary data subject to recall bias

Loss to follow-up (9%) — though similar between groups

Short follow-up (only until delivery) — no data on long-term child outcomes

Cannot separate effects of nutrition education vs. gardening

Only 12 clusters — limited statistical power for some secondary outcomes

**What a critical reader would note:**

**No true placebo control:** The control group received standard ANC, but intervention women received more contact time with health workers (6 extra sessions plus garden visits). The effect could partly be due to increased attention and social support, not the specific intervention content.

**Baseline imbalance:** Despite randomization, intervention women had slightly lower baseline dietary diversity (3.4 vs. 3.8) and higher food insecurity (62% vs. 55%). The adjusted analyses may not fully account for these differences.

**Generalizability:** Results may not apply to urban populations, women with higher education, or settings with different food environments (e.g., cold climates, limited growing seasons, or where land is unavailable).

**No cost-effectiveness data:** The intervention required seeds, tools, and health worker training. It's unclear whether this is more cost-effective than simply providing supplements or cash transfers.

**No data on actual nutrient intake:** Dietary diversity is a proxy for nutrient adequacy, not a direct measure. Women could have eaten more food groups but still been deficient in specific nutrients.

**Seasonal effects:** The study ran over 18 months, but planting seasons vary. Women enrolled during dry seasons may have had less success with gardens, but this was not analyzed.

**No data on maternal mental health or stress:** Gardening could reduce stress (improving outcomes via cortisol pathways), but this was not measured.

**No data on paternal or household involvement:** The intervention targeted households, but the extent of male partner support was not assessed.

For someone running their own n=1 experiment (e.g., a pregnant woman or someone planning pregnancy who wants to improve their own nutrition):

### What to test

**Intervention:** Combine (a) structured nutrition education focused on dietary diversity (aim for ≥5 food groups daily) with (b) growing your own vegetables, even in small spaces (container gardening, raised beds, or community garden plot).

**Dose:** Aim to consume vegetables from your garden at least 3–4 times per week, and track whether this increases your overall dietary diversity.

**Specific vegetables to prioritize:** Dark leafy greens (kale, Swiss chard, spinach), vitamin A-rich vegetables (carrots, sweet potatoes, pumpkin), and allium vegetables (onions, garlic) — these provide iron, folate, vitamin A, and fiber.

### Minimum meaningful duration

**At least 12–16 weeks** during pregnancy (ideally starting by week 12–16 and