A systematic review of thyroid dysfunction in preterm neonates exposed to topical iodine
- Population Health Sciences, Medical Research Institute, Mackenzie Building, University of Dundee, Dundee, Scotland
- Correspondence to Dr Fiona Williams, Population Health Sciences, Medical Research Institute, Mackenzie Building, University of Dundee, Kirsty Semple Way, Dundee DD2 4BF Scotland;
- Received 1 February 2013
- Revised 13 September 2013
- Accepted 16 September 2013
- Published Online First 8 October 2013
Objective To determine whether maternal exposure to iodine or neonatal exposure to topical iodine-containing solutions increases the risk of transient thyroid dysfunction in neonates born <32 weeks’ gestation or <1.5 kg.
Design Systematic review.
Search strategy Electronic searches were conducted using Medline and the Cochrane Library.
Eligibility criteria A study was eligible for review if it reported neonatal exposure to topical iodine or maternal iodine exposure. The key outcome measure was neonatal thyroid function. The search had no restrictions on date of publication, type of study or language.
Results 794 papers were identified during the initial search; 15 studies were fully reviewed. The incidence of (transient) hypothyroidism/hyperthyrotropinaemia following exposure to topical iodine ranged from 12 to 33 per 100 infants; the incidence in non-exposed infants was 0.
Conclusions There is evidence that neonatal exposure to iodine-containing disinfectants causes thyroid dysfunction in infants born <32 weeks. None of the studies evaluated neurodevelopment. Larger scale studies are needed to determine definitively the nature of the relationship and the impact of exposure on neurodevelopment. In the meantime, it would seem prudent to restrict exposure of iodine-containing skin disinfectants in preterm infants; chlorhexidine might be a credible alternative.
What is already known on this topic
Extreme preterm infants are at risk of iodine overload as the Wolff–Chaikoff effect does not mature until 36–40 weeks’ gestation.
The British National Formulary recommends that the use of povidone-iodine in neonates <32 weeks or <1500 g is contraindicated.
Preterm infants are exposed to iodine through the use of neonatal- and/or maternal-iodinated skin disinfectants.
What this study adds
Urinary iodine levels in infants not exposed to topical iodine ranged from 55–400 µg/L; levels in exposed infants range from 400–18 900 µg/L.
The incidence of hypothyroidism/hyperthyrotropinaemia following exposure to topical iodine ranged from 12 to 33 per 100 infants; the incidence in non-exposed infants was 0.
Current evidence is suggestive that maternal and/or neonatal exposure to topical iodine causes neonatal thyroid dysfunction, but definitive evidence is lacking.
The human body has a defence mechanism against iodine overload. Once serum iodine levels reach a critical level, the uptake of iodine by the thyroid and the process of iodination of tyrosine are stopped, thereby preventing the overproduction of thyroid hormones; a mechanism known as the Wolff–Chaikoff effect.1 In adults and children, this mechanism maintains near normal levels of thyroid stimulating hormone (TSH) and thyroid hormones when there is challenge with iodine overload. But the Wolff–Chaikoff effect does not mature until 36–40 weeks’ gestation,2 which renders preterm infants vulnerable to the effects of iodine overload, the immediate consequences of which can be development of goitre and/or transient hypothyroidism3 and neurodevelopment compromise.
Direct iodine overload of the newborn can be caused through use of iodinated disinfectants4 ,5 and by use of iodinated contrast media.6 The foetus and newborn can be further exposed indirectly via high maternal iodine concentrations either prenatally via transplacental transfer or postnatally through breast milk.7 ,8 The major source of maternal exposure is through skin disinfection with povidone-iodine prior to caesarean section. As maternal iodine levels are correlated with breast milk iodine levels, breastfed infants are doubly exposed.9
In addition to an inability to mount a Wolff–Chaikoff response,1 there is evidence that preterm infants are more vulnerable to iodine exposure than term infants4 ,10–12 because of immaturity of the thyroid gland, increased permeability of their skin and decreased renal clearance of iodine.13–15 Presumably in reflection of this the British National Formulary (BNF) for Children16 lists neonates <32 weeks and infants <1.5 kg as contraindications to the use of povidone-iodine for skin disinfection. Furthermore, it advises that its use should be avoided in women who are breastfeeding and it should be used with caution in pregnancy.
Nevertheless, iodine-containing disinfectants are still a feature of neonatal and obstetric units. As part of another study, in 2006, we surveyed 42 of the largest neonatal and obstetric units in the UK and asked about their use of iodine-containing solutions; 57% of obstetric units and 21% of the neonatal units reported its use.17 If maternal exposure is an important source of infants’ iodine overload, the use of iodine disinfectants in obstetric units is problematic as the majority of preterm births are delivered by caesarean section.18 Because of this potential for iatrogenic harm, we undertook a systematic review to determine whether obstetric or neonatal exposure to topical iodine increases the risk of thyroid dysfunction in neonates born preterm.
A study was eligible for review if it reported exposure to topical iodine-containing solutions in either neonates, born <32 weeks’ gestation or with a birth weight of <1500 g, or women during pregnancy. Maternal exposure to iodine could occur, for example, through vaginal douching, epidural/spinal anaesthesia, caesarean section or vaginal delivery, or to treat postoperative wound infection. In neonates, exposure could occur, for example, during surgery, for antisepsis prior to procedures such as insertion of a central line or for disinfection of the umbilical stump. Neonates could also be exposed to excess iodine in breast milk if their mothers had been exposed to topical iodine.
The key outcome measure was thyroid function, and to be eligible, a study should also report thyroid function in the exposed neonates, for example, TSH, thyroxine (T4) or free thyroxine (FT4). The search had no restrictions on date of publication, type of study or language.
Information sources and search strategy
Systematic electronic searches were conducted using MEDLINE (EBSCOhost Interface) and Cochrane Library (Wiley Online Library Interface). Additional searches were done using ISI Web of Knowledge and Google scholar. The reference lists of the included articles were scanned to look for literature that had not been obtained by searches. The search strategy was developed using the structured approach of the Patient/Population, Intervention, Comparison, Outcome Studies (PICOS).19 MeSH terms were generated for each of the PICOS subject headings and formatted according to which generated the most results, either by exploding or searching the MeSH term as a major concept. If a search term was not available as a MeSH heading, it was searched as a keyword. The MeSH terms for each PICOS subject heading were combined using Boolean terms enabling multiple combinations of terms to be searched at once. The same search strategy was used for both Medline and Cochrane Library search, suitably adapted to the vagaries of each system.
Study selection, data abstraction and risk of bias assessment
The titles and abstracts of articles obtained from the electronic searches were screened and assessed for their relevance according to the eligibility criteria. Selected articles identified after the initial screen were retrieved in full text; data were abstracted and recorded about route and source of exposure to iodine, gestational age, birth weight, number in the study, study design, results of thyroid function tests, urinary iodine levels, diagnosis of (transient) hypothyroidism/hyperthyrotropinaemia and whether or not neurodevelopment was assessed.
The methodological robustness of the cohort studies included in the review was assessed by the Downs Black Scale.20 This scale rates papers on the basis of four criteria (reporting, external validity, internal validity and power), which allocates maximum scores of respectively 11, 3, 13 and 5 for the criteria, with a maximum possible score of 32 (the higher the score, the more robust the methodology).
At each stage of study selection, data abstraction and assessment of bias, reviews were undertaken separately by JA and FLRW. Discrepancies between the reviewers were resolved by discussion.
A meta-analysis was impracticable due to heterogeneity of study design: TSH and iodine levels were often estimated from graphs, TSH levels were measured in varying blood media, iodine and TSH estimations were made at varying times postnatally and also at varying times postexposure.
During the initial search, 794 papers were identified; following exclusions, 14 papers4 ,21–33 that reported 15 studies remained for full review (figure 1). One paper24 reported two distinct studies, an inter-NICU and intra-NICU study. Nine articles matched the inclusion criteria exactly4 ,21 ,22 ,25–29 ,32; a further six articles that matched the inclusion criteria very closely were also included.23 ,24 ,30 ,31 ,33 The reasons for slightly less-than-perfect match of the additional six articles varied. Three studies23 ,24 included preterm infants >32 weeks, although the means for the groups were between 32.1 and 32.7 weeks, which indicates that the infants were skewed towards the lower gestational ages. In one case report, the route of exposure was implied but not definitive and the infant was 32 weeks’ gestation30; in one case report, the birth weight was 1500 g31, and one case report included one infant of 34 weeks’ gestation33 (table 1).
No randomised controlled trials were found; five cohort studies were identified4 ,22–24 and ten case reports/series.21 ,25–33 The cohort studies included a mean of 32 exposed infants (range 9–73) and 27 unexposed infants (range 14–55). Ten studies reported neonatal exposure to povidone-iodine,4 ,22 ,23 ,25–27 30–33 three reported maternal exposure to povidone-iodine21 ,28 ,29 and two reported both neonatal and maternal exposures (table 1).24
Risk of bias within studies
Overall none of the cohort studies4 ,22–24 scored highly against the Downs Black Scale20, with scores ranging from 13/32 to 19/32. Generally the papers scored well on the reporting criteria, reasonably on the external validity, but less well on the internal validity (essentially determining bias) and power criteria (table 2). The low scores are not necessarily reflective of the robustness of the paper. Some items scored an automatic zero as the methodology was not (and could not be) randomised; sometimes an item was scored zero because it was impossible to determine whether or not they had ascribed the item. The single main contributor to the low scores of the papers was the small sample sizes.
Results of individual studies
All studies reported large differences in levels of urinary iodine excretion between the exposed and non-exposed infants. Over days 1–3 postnatal, non-exposed infants typically had iodine excretion levels around 100 µg/L (mean range 55–148 µg/L), while levels in exposed infants were much higher (mean range 1100–18 900 µg/L) (table 3). Not unexpectedly urinary iodine levels were raised on the day of exposure23 ,24 and still raised 3 days postexposure.4 ,22 ,23
TSH levels were measured on filter paper in three studies,23 ,24 in serum in one study22, and the fifth study reports taking filter paper blood spots but undertaking assays in serum.4 TSH levels on filter paper are approximately half of those in serum. Four of the cohort studies4 ,22 ,24 reported mean/median TSH levels; these papers reported statistically significant higher TSH levels in iodine-exposed compared with non-exposed infants (table 3). One study23 reported no difference in levels of TSH between the exposed and non-exposed infants. The incidence of (transient) hypothyroidism/hyperthyrotropinaemia following exposure to topical iodine ranged from 12 to 33 per 100 infants; the incidence in non-exposed infants was 0 (table 3). T4 or FT4 levels were reported by all cohort studies, and the mean or median levels did not differ appreciably between exposed and non-exposed infants.
The duration of follow-up varied between the cohort studies. Two studies reported that neonatal TSH levels had normalised by day 28 postnatal,4 ,22 two studies made no comment23 ,24 and the Intra-NICU study24 found statistically significant high TSH levels in the exposed infants at day 90 postnatal.
Neurodevelopment was not reported by any of the studies.
Ten papers reported single case histories.21 ,25–33 Twenty-nine infants who met the inclusion criteria were reported by the ten papers. The source of iodine exposure was neonatal in seven reports,25–27 30–33 maternal via breast milk for two21 ,29 and maternal in utero exposure for one report.28
Levels of urinary iodine were measured at various times postexposure and ranged from 160 µmol/L to 3932 µg/L. The highest level (3932 µg/L) was recorded consequent to exposure from maternal use of iodine-soaked tampons for treatment of an abdominal wall abscess on postnatal days 7–32 (table 3).
TSH levels were measured, with one exception31 from serum/plasma samples (table 3). Eight papers reported significant increases in levels of TSH following exposure to iodine. Two studies25 ,33 found no increase in TSH levels following exposure to iodine, but observed appreciable drops in the levels of T4/FT4. Of the 29 infants included in the case series, 11 were reported to have transient hypothyroidism/hyperthyrotropinaemia. When reported, T4 levels were normal or low (table 3).
All studies reviewed reported some type of thyroid dysfunction in exposed neonates. Thirteen studies reported cases of transient hypothyroidism/hyperthyrotropinaemia. All studies that measured urinary iodine showed hugely increased levels in infants exposed to iodine compared with non-exposed infants. The raised urinary iodine levels confirmed exposure but were not always associated with greatly elevated TSH levels. The incidence of transient hypothyroidism/hyperthyrotropinaemia in the exposed infants in the cohort studies in this review ranged from 12 per 100 to 33 per 100, which is somewhat similar to the incidence of 18.2% found in a systematic review that investigated exposure of preterm infants to iodinated contrast media.14
A definitive answer to the question posed by this review is impossible due to limitations in the current body of research.
First, the number of infants recruited to the cohort studies was typically few, and methodological differences between the studies did not facilitate a meta-analysis. Biochemical data were estimated from graphs in four studies,4 ,22 ,24 ,26 so the levels for these studies reported in table 3 are not precise.
Second, the gestational age range of four studies23 ,24 ,33 included infants >32 weeks; although, as the extreme preterm are most vulnerable to the effects of iodine exposure4 ,11, this would lead to an underestimation of the effect size. We elected to include these studies as the mean age of the reported infants was typically only slightly above 32 weeks (ranging from 29.5 to 32.7 weeks). The BNF's rationale for <32 weeks is not explicit, and it is very unlikely that a few extra days will make any physiological/clinical difference.
Third, hormone levels were measured using a variety of assays, in serum or in whole blood, on different postnatal days, and for varying lengths of follow-up. Concordance between TSH assays is variable34, and several factors that are common in preterm infants such as acute severe illness35 and certain prescription drugs can appreciably alter TSH and thyroid hormone concentrations.36 There is also a marked gestational age effect on TSH and iodothyronine concentrations that confounds interpretation of levels.37 Knowledge and account of these factors is critical when interpreting the impact of iodine exposure on preterm neonatal thyroid function.
Fourth, three cohort studies4 ,22 ,23 reported transient hypothyroidism and two studies24 reported transient hyperthyrotropinaemia in iodine-exposed infants. We report a joint incidence, as the number classified depends on how hypothyroidism is defined. For example, Linder et al's24 cases of hyperthyrotropinaemia with TSH levels >30 mIU/L would be classified as hypothyroidism by others. Hyperthyrotropinaemia and hypothyroidism are essentially on the same continuum with current convention defining hyperthyrotropinaemia as serum levels of TSH≥10 and ≤20 mU/L, and hypothyroidism as >20 mU/L.
Fifth, the time between exposure to iodine and thyroid function tests was not always clear. Many studies undertook serial measurements of thyroid function, and as neonatal exposure is often due to skin antisepsis, exposure can occur on multiple occasions. Urinary iodine levels were measured in most studies, which allowed confirmation of exposure, but the time between exposure and measurement was variable. Similarly it was not possible to quantify the dose of iodine nor to estimate dose–response. Some infants might have had a single exposure; some were exposed to relatively small amounts, for example, for skin antisepsis prior to catheter placement, but others had much larger doses, for instance, prior to surgery. In all studies that measured iodine, its excretion was clearly elevated, and while the TSH levels were often elevated they were not hugely so. The two instances when neonatal TSH was greatly elevated and that may be indicative of dose–response were associated with 25 days of maternal exposure via iodinated tampons and 20 days of neonatal×3 times daily use of povidone-iodine for scalp antisepsis.
Sixth, it is likely that the follow-up of infants reported by most studies is not long enough to detect (all cases of) altered thyroid function. Immaturity of the hypothalamus–pituitary–thyroid axis, critical illness and drugs such as dopamine that inhibit TSH may mask an immediate rise associated with iodine exposure.38
Finally, the findings we report may be subject, for example, to outcome reporting bias and/or publication bias, where study results are suppressed or omitted in study publications. Such biases are very difficult to detect.
There are alternatives to povidone-iodine as chlorhexidine is an effective skin disinfectant. In children, use of chlorhexidine when compared with povidone-iodine clearly reduced bacterial colonisation following skin disinfection before catheter placement.39 ,40 However, some units are reluctant to use chlorhexidine because the typical preparation is alcohol–based, and if allowed to pool on the infants’ skin it may cause burns or it could pose a fire hazard during diathermy. Careful practice can minimise the risk of burns or fire, or units could use aqueous chlorhexidine solutions.41
In summary, there is evidence that exposure of extreme preterm infants to topical iodine disinfectants causes thyroid dysfunction, although the extent of the dysfunction is unknown. None of the studies included in this review undertook evaluated neurodevelopmental status of the exposed infants. Larger studies would be needed to determine definitively the nature of the relationship and the impact of exposure on neurodevelopmental outcome. In the meantime, it would seem prudent to follow the advice of the BNF to restrict exposure of iodine-containing skin disinfectants in preterm infants; chlorhexidine or octenidine might be credible alternatives.
Contributors JA contributed to the conceptualisation and design of the systematic review, undertook the independent review of the journal titles and abstracts, and full-text articles, reviewed and revised the manuscript, and approved the final manuscript as submitted. JA undertook this systematic review as a supervised project in the fourth year of medical undergraduate studies. FLRW contributed to the conceptualisation and design of the systematic review, undertook the independent review of the journal titles and abstracts, and full-text articles, drafted, reviewed and revised the manuscript, and approved the final manuscript as submitted.
Funding Tenovus Scotland, Anonymous Trust Dundee.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.