The decision this page serves. Should a woman take supplemental iodine in the preconception / pregnancy / postpartum window, and does the answer turn on her baseline iodine status? The honest top-line, from the only RCT-restricted synthesis (Cochrane), is that the randomized evidence is too thin to settle routine supplementation — and that the benefit that does show up is concentrated where deficiency is severe, which is the same deficiency-repletion shape the Deficiency Repletion vs Enhancement concept predicts. — this lead is the wiki’s framing of the page; the underlying Cochrane verdict and effect sizes are extracted and tagged in the sections below.
The Cochrane verdict: insufficient, not null
The Cochrane SR restricted to RCTs/quasi-RCTs (14 studies, 11 with data, >2700 women) reached no firm conclusion: «There were insufficient data to reach any meaningful conclusions on the benefits and harms of routine iodine supplementation in women before, during or afer pregnancy» (Harding et al., 2017). This is the insufficient- evidence state, not the no-effect state — most outcomes are single-trial, low or very-low GRADE, with CIs wide enough to contain both meaningful benefit and none. Treat no clear effect here as unmeasured, not no effect.
A structural reason the RCT base is thin — and stays thin where it matters most. A placebo RCT in a severely deficient population «may be unethical», and where iodine-containing prenatal supplements are already routine an RCT is unfeasible (Harding et al., 2017). So the strongest design is unavailable exactly in the stratum with the largest expected effect — the streetlight effect on the outcome side: the RCT lamp does not shine where the key was dropped.
What the trials do show — an adverse-effect ledger with two signals
Two effects were large enough to clear the noise (both in mild-to-moderate deficiency settings, both graded low / very-low):
- Postpartum hyperthyroidism DOWN ~68% — average RR 0.32 (95% CI 0.11 to 0.91, 3 trials, 543 women, I2=0) (Harding et al., 2017). (The maternal Summary-of-findings table mis-prints the upper bound as 1.91; the abstract and the results text both give 0.91 — a documented typo.)
- Digestive intolerance UP ~15x — average RR 15.33 (95% CI 2.07 to 113.70, 1 trial, 76 women, very low) (Harding et al., 2017). A tolerability/ adherence cost, not a hard-outcome harm, but real: the adherence penalty of nausea/vomiting can sink a supplement in practice.
Everything else — maternal hypo/hyperthyroidism, preterm birth, TPO-antibodies, low birthweight, neonatal thyroid function — showed no clear difference, all low/very-low quality, all consistent with the insufficient-evidence reading rather than a demonstrated null (Harding et al., 2017). No trial reported any child outcome beyond the neonatal period — a hard limit on what the RCT base can say about the neurodevelopmental endpoint that motivates supplementation.
The child-cognition signal is a deficiency-severity gradient
The one place a large, precise benefit appears is severe deficiency: pooling the two old injected-iodine trials in severely deficient settings, «the MD was 11.21 points (95% CI 7.96 to 14.46)» on child IQ, with no heterogeneity (I2=0) (Harding et al., 2017). Severe deficiency also drove prevention of cretinism / congenital anomalies (RR 0.27, 0.12-0.60) and neonatal goitre (RR 0.11, 0.02-0.56). In mild-moderate deficiency the single oral RCT (Zhou 2015, Bayley-III at 18 months) was null (Harding et al., 2017). The authors could not run the planned baseline-status subgroup analysis — «we did not identify enough trials» (Harding et al., 2017) — so the gradient is read across trials (severe-setting trials positive, mild-setting trial null), not within a subgroup contrast. Route-(a)/(b) caveat: this is suggestive effect-modification by baseline status, but it rests on between-trial comparison, not a powered interaction test.
How this refines the held iodine evidence (Bougma 2013) — a type-F advance, NOT independent corroboration
The wiki already holds Bougma 2013 (iodine -> child mental development; the held instance is on Deficiency Repletion vs Enhancement). Harding is a refinement (type F) of the same evidence lineage, not an independent corroboration (type E): Kimberly B Harding is a co-author of Bougma 2013, the two syntheses share their severe-deficiency IQ trials (Kevany 1969, Thilly 1978, Pharoah 1971), and WHO-affiliated authors recur across both. A shared author defeats independence outright, so this adds no E-count and no independent-backing confidence to the child-IQ claim — it sharpens the same claim.
The parameter table (built before any prose comparison, per the cross-source rule):
| Parameter | Harding 2017 (Cochrane) | Bougma 2013 | Same quantity? |
|---|---|---|---|
| Evidence base | RCT / quasi-RCT only | 2 RCT + 8 non-RCT trials + 13 prospective cohorts | NO — RCT-restricted subset vs observational-inclusive |
| Child IQ, severe deficiency, RCT-grade | MD +11.21 points (7.96-14.46), 2 trials, 174 children | RCT-only pooled d=0.68 (~+10.2 IQ) | ~YES — both severe-deficiency RCT estimates; converge ~+10-11 IQ, overlapping trials |
| Child IQ, all designs pooled | not computed (RCT-restricted) | d=0.49 = +7.4 IQ (range 6.9-10.2) (Bougma et al., 2013) | NO — Bougma includes observational; Harding excludes it |
| Child IQ, mild-moderate deficiency | null (Zhou 2015, oral) | not isolated (has timing window: early d=0.51 vs late 0.17) | NO — Harding adds a mild-deficiency null Bougma lacks |
| Overall verdict on child development | «mixed» / insufficient for routine | «regardless of study design, iodine deficiency had a substantial impact on mental development» (Bougma et al., 2013) | These ARE the divergence — see below |
The divergence is a design artifact, adjudicated by the source itself — not a tension to file.
Harding names the gap and its cause: «Findings showed a positive impact of iodine, regardless as to
the study design, ranging from 6.9 to 10.2 IQ points, whereas our findings on child development were
mixed» (Harding et al., 2017); and of the similar
Zimmermann 2012a review, «This review reached different conclusions to ours, however this may be
explained by its design» — Zimmermann did not specify study-inclusion criteria and included
non-randomized trials (Harding et al., 2017). Run the not-joined
check: the two verdicts hold on different evidence bases (RCT-restricted vs observational-
inclusive) and are consistent once matched — where designs overlap (severe-deficiency RCTs) both
give ~+10-11 IQ; Bougma’s larger, broader estimate rides on observational data an RCT-only synthesis
discards. This is scope-mismatch (not-joined check (ii)), a distinction, not a [[tension]].
Net F-composite: Bougma alone reads as iodine adds ~7 IQ points broadly; Harding bounds that — the RCT-grade, severe-deficiency benefit is real and ~+11 IQ, but the broad estimate is inflated by observational designs, and in mild-moderate deficiency the randomized evidence is null-to-thin. The composite is the deficiency-repletion shape, not a universal enhancement.
Supplementation is also a route to iodine EXCESS in pregnancy
Iodine is not monotone-benign: the same trials show supplementation raised excessive iodine intake — «7.0% of women who received iodine supplements and 1.6% of women who did not had excessive iodine intake during pregnancy (average RR 4.33; 95% CI 1.24 to 15.07)» (Harding et al., 2017), and paradoxically raised insufficient neonatal intake (RR 2.14, 1.04-4.37) by an unexplained mechanism. The upper arm is a real harm channel -> Excess Iodine Intake and Thyroid Disease; a deficiency benefit plus an excess harm on the same nutrient is the genuine two-armed U, not the artifactual kind -> The U-Shaped Association Artifact. Note the tolerable-upper-intake figures (IOM 1100 µg/day, EC 600 µg/day) are «not intended for iodine-deficient populations» (Harding et al., 2017) — a deficient thyroid can be harmed at lower intakes than a replete one, so the safe window is narrower, not wider, in the very stratum that benefits most.
Decision map by baseline iodine status (the stratification this supports)
- Severe deficiency (endemic). Strongest case: RCT-grade IQ benefit (~+11), cretinism and congenital-anomaly prevention, a non-significant 34% lower perinatal mortality concentrated in the one severe-deficiency trial (Harding et al., 2017). This is deficiency-repletion, and it is why placebo RCTs here are now considered unethical. Rare in the wiki’s reasonably-healthy default stratum.
- Mild-moderate deficiency (most of the trial evidence, and most of the developed world). Benefit on hard child outcomes is unproven by randomized evidence (the one oral RCT was null); the demonstrable effects are a lower postpartum-hyperthyroidism risk and a digestive-tolerability cost. Guidance bodies still recommend ~150-250 µg/day (WHO/UNICEF/ICCIDD 250; EFSA 200; IOM 220/290) on the precautionary logic that fetal brain development is time-critical and status before conception matters (Harding et al., 2017) — a population-standpoint call the RCT evidence does not itself compel.
- Replete. No evidence of benefit and a live excess-intake channel (RR 4.33) — the enhancement leg, where more buys nothing and the upper bound is the operative risk.
Gaps (type G)
- No powered baseline-status subgroup analysis exists — the single most decision-relevant cut (who benefits) is exactly the one too few trials could support.
- No child outcome beyond the neonatal period in any RCT — the neurodevelopmental endpoint is RCT-unmeasured past infancy; the +11.21 rests on two old injected-iodine trials.
- Mild-moderate-deficiency child cognition rests on one small aborted oral RCT (Zhou 2015) — a modern RCT pooling of mild-deficiency oral supplementation on validated child
cognition.
confidence: low. - G (needs aggregation): a pooled RCT-grade absolute effect on child IQ by baseline status is a magnitude the current trial set cannot compute — an honest structural limit, not a queue item.